extract vmsingle-specific code into vmstorageSingleNode type

Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>

.github/pull_request_template.md

@@ -1,10 +1,3 @@
-### Describe Your Changes
+**PLEASE REMOVE LINE BELOW BEFORE SUBMITTING**
 
-Please provide a brief description of the changes you made. Be as specific as possible to help others understand the purpose and impact of your modifications.
-
-### Checklist
-
-The following checks are **mandatory**:
-
-- [ ] My change adheres to [VictoriaMetrics contributing guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
-- [ ] My change adheres to [VictoriaMetrics development goals](https://docs.victoriametrics.com/victoriametrics/goals/).
+Before creating the PR, make sure you have read and followed the [VictoriaMetrics contributing guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).

.github/workflows/build.yml

@@ -57,15 +57,17 @@ jobs:
             arch: amd64
           - os: openbsd
             arch: amd64
+          - os: netbsd
+            arch: amd64
           - os: windows
             arch: amd64
     steps:
       - name: Code checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Setup Go
         id: go
-        uses: actions/setup-go@v6
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c  # v6.4.0
         with:
           cache-dependency-path: |
             go.sum

.github/workflows/changelog-linter.yml

@@ -9,7 +9,7 @@ jobs:
   tip-lint:
     runs-on: 'ubuntu-latest'
     steps:
-      - uses: 'actions/checkout@v6'
+      - uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
         with:
           # needed for proper diff
           fetch-depth: 0

.github/workflows/check-commit-signed.yml

@@ -8,7 +8,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout code
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
         with:
           fetch-depth: 0  # we need full history for commit verification
 

.github/workflows/check-licenses.yml

@@ -15,11 +15,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Code checkout
-        uses: actions/checkout@master
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Setup Go
         id: go
-        uses: actions/setup-go@v6
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c  # v6.4.0
         with:
           go-version-file: 'go.mod'
           cache: false
@@ -27,7 +27,7 @@ jobs:
       - run: go version
 
       - name: Cache Go artifacts
-        uses: actions/cache@v4
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae  # v5.0.5
         with:
           path: |
             ~/.cache/go-build

.github/workflows/codeql-analysis-go.yml

@@ -29,18 +29,18 @@ jobs:
 
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Set up Go
         id: go
-        uses: actions/setup-go@v6
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c  # v6.4.0
         with:
           cache: false
           go-version-file: 'go.mod'
       - run: go version
 
       - name: Cache Go artifacts
-        uses: actions/cache@v4
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae  # v5.0.5
         with:
           path: |
             ~/.cache/go-build
@@ -50,14 +50,14 @@ jobs:
           restore-keys: go-artifacts-${{ runner.os }}-codeql-analyze-${{ steps.go.outputs.go-version }}-
 
       - name: Initialize CodeQL
-        uses: github/codeql-action/init@v4
+        uses: github/codeql-action/init@e46ed2cbd01164d986452f91f178727624ae40d7  # v4.35.3
         with:
           languages: go
 
       - name: Autobuild
-        uses: github/codeql-action/autobuild@v4
+        uses: github/codeql-action/autobuild@e46ed2cbd01164d986452f91f178727624ae40d7  # v4.35.3
 
       - name: Perform CodeQL Analysis
-        uses: github/codeql-action/analyze@v4
+        uses: github/codeql-action/analyze@e46ed2cbd01164d986452f91f178727624ae40d7  # v4.35.3
         with:
           category: 'language:go'

.github/workflows/docs.yaml

@@ -16,19 +16,19 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Code checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
         with:
           path: __vm
 
       - name: Checkout private code
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
         with:
           repository: VictoriaMetrics/vmdocs
           token: ${{ secrets.VM_BOT_GH_TOKEN }}
           path: __vm-docs
 
       - name: Import GPG key
-        uses: crazy-max/ghaction-import-gpg@v7
+        uses: crazy-max/ghaction-import-gpg@2dc316deee8e90f13e1a351ab510b4d5bc0c82cd  # v7.0.0
         id: import-gpg
         with:
           gpg_private_key: ${{ secrets.VM_BOT_GPG_PRIVATE_KEY }}

.github/workflows/test.yml

@@ -32,11 +32,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Code checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Setup Go
         id: go
-        uses: actions/setup-go@v6
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c  # v6.4.0
         with:
           cache-dependency-path: |
             go.sum
@@ -47,7 +47,7 @@ jobs:
       - run: go version
 
       - name: Cache golangci-lint
-        uses: actions/cache@v4
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae  # v5.0.5
         with:
           path: |
             ~/.cache/golangci-lint
@@ -72,11 +72,11 @@ jobs:
 
     steps:
       - name: Code checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Setup Go
         id: go
-        uses: actions/setup-go@v6
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c  # v6.4.0
         with:
           cache-dependency-path: |
             go.sum
@@ -94,11 +94,11 @@ jobs:
 
     steps:
       - name: Code checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Setup Go
         id: go
-        uses: actions/setup-go@v6
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c  # v6.4.0
         with:
           cache-dependency-path: |
             go.sum

.github/workflows/vmui.yml

@@ -32,11 +32,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Code checkout
-        uses: actions/checkout@v6
+        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
 
       - name: Cache node_modules
         id: cache
-        uses: actions/cache@v5
+        uses: actions/cache@27d5ce7f107fe9357f9df03efb73ab90386fccae  # v5.0.5
         with:
           path: app/vmui/packages/vmui/node_modules
           key: vmui-deps-${{ runner.os }}-${{ hashFiles('app/vmui/packages/vmui/package-lock.json', 'app/vmui/Dockerfile-build') }}
@@ -69,7 +69,7 @@ jobs:
           VMUI_SKIP_INSTALL: true
 
       - name: Annotate Code Linting Results
-        uses: ataylorme/eslint-annotate-action@v3
+        uses: ataylorme/eslint-annotate-action@d57a1193d4c59cbfbf3f86c271f42612f9dbd9e9  # 3.0.0
         with:
           GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
           report-json: app/vmui/packages/vmui/vmui-lint-report.json

Makefile

@@ -485,8 +485,8 @@ apptest-legacy: victoria-metrics-race vmbackup-race vmrestore-race
 		curl --output-dir /tmp -LO $${URL}/$${VMSINGLE} && tar xzf /tmp/$${VMSINGLE} -C $${DIR} && \
 		curl --output-dir /tmp -LO $${URL}/$${VMCLUSTER} && tar xzf /tmp/$${VMCLUSTER} -C $${DIR} \
 	); \
-	VM_LEGACY_VMSINGLE_PATH=$${DIR}/victoria-metrics-prod \
-	VM_LEGACY_VMSTORAGE_PATH=$${DIR}/vmstorage-prod \
+	VMSINGLE_V1_132_0_PATH=$${DIR}/victoria-metrics-prod \
+	VMSTORAGE_V1_132_0_PATH=$${DIR}/vmstorage-prod \
 	go test ./apptest/tests -run="^TestLegacySingle.*"
 
 benchmark:
@@ -535,6 +535,15 @@ remove-golangci-lint:
 govulncheck: install-govulncheck
 	govulncheck ./...
 
+govulncheck-docker:
+	docker run -w $(PWD) -v $(PWD):$(PWD) \
+		-v govulncheck-gomod-cache:/root/go/pkg/mod \
+		-v govulncheck-gobuild-cache:/root/.cache/go-build \
+		-v govulncheck-go-bin:/root/go/bin \
+		--env="GOCACHE=/root/.cache/go-build" \
+		--env="GOMODCACHE=/root/go/pkg/mod" \
+		"$(GO_BUILDER_IMAGE)" /bin/sh -c "which govulncheck || go install golang.org/x/vuln/cmd/govulncheck@latest && govulncheck ./..."
+
 install-govulncheck:
 	which govulncheck || go install golang.org/x/vuln/cmd/govulncheck@latest
 

SECURITY.md

@@ -1,42 +1,4 @@
 # Security Policy
 
-## Supported Versions
+You can find out about our security policy and VictoriaMetrics version support on the [security page](https://docs.victoriametrics.com/victoriametrics/#security) in the documentation.
 
-The following versions of VictoriaMetrics receive regular security fixes:
-
-| Version                                                                        | Supported          |
-|--------------------------------------------------------------------------------|--------------------|
-| [Latest release](https://docs.victoriametrics.com/victoriametrics/changelog/)  | :white_check_mark: |
-| [LTS releases](https://docs.victoriametrics.com/victoriametrics/lts-releases/) | :white_check_mark: |
-| other releases                                                                 | :x:                |
-
-See [this page](https://victoriametrics.com/security/) for more details.
-
-## Software Bill of Materials (SBOM)
-
-Every VictoriaMetrics container{{% available_from "#" %}} image published to
-[Docker Hub](https://hub.docker.com/u/victoriametrics)
-and [Quay.io](https://quay.io/organization/victoriametrics)
-includes an [SPDX](https://spdx.dev/) SBOM attestation
-generated automatically by BuildKit during
-`docker buildx build`.
-
-To inspect the SBOM for an image:
-
-```sh
-docker buildx imagetools inspect \
-  docker.io/victoriametrics/victoria-metrics:latest \
-  --format "{{ json .SBOM }}"
-```
-
-To scan an image using its SBOM attestation with
-[Trivy](https://github.com/aquasecurity/trivy):
-
-```sh
-trivy image --sbom-sources oci \
-  docker.io/victoriametrics/victoria-metrics:latest
-```
-
-## Reporting a Vulnerability
-
-Please report any security issues to <security@victoriametrics.com>

app/victoria-metrics/main.go

@@ -22,7 +22,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/procutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/promscrape"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/pushmetrics"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
 )
 
 var (
@@ -30,23 +29,26 @@ var (
 	useProxyProtocol = flagutil.NewArrayBool("httpListenAddr.useProxyProtocol", "Whether to use proxy protocol for connections accepted at the corresponding -httpListenAddr . "+
 		"See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt . "+
 		"With enabled proxy protocol http server cannot serve regular /metrics endpoint. Use -pushmetrics.url for metrics pushing")
-	minScrapeInterval = flag.Duration("dedup.minScrapeInterval", 0, "Leave only the last sample in every time series per each discrete interval "+
-		"equal to -dedup.minScrapeInterval > 0. See also -streamAggr.dedupInterval and https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#deduplication")
 	dryRun = flag.Bool("dryRun", false, "Whether to check config files without running VictoriaMetrics. The following config files are checked: "+
 		"-promscrape.config, -relabelConfig and -streamAggr.config. Unknown config entries aren't allowed in -promscrape.config by default. "+
 		"This can be changed with -promscrape.config.strictParse=false command-line flag")
-	inmemoryDataFlushInterval = flag.Duration("inmemoryDataFlushInterval", 5*time.Second, "The interval for guaranteed saving of in-memory data to disk. "+
-		"The saved data survives unclean shutdowns such as OOM crash, hardware reset, SIGKILL, etc. "+
-		"Bigger intervals may help increase the lifetime of flash storage with limited write cycles (e.g. Raspberry PI). "+
-		"Smaller intervals increase disk IO load. Minimum supported value is 1s")
 	maxIngestionRate = flag.Int("maxIngestionRate", 0, "The maximum number of samples vmsingle can receive per second. Data ingestion is paused when the limit is exceeded. "+
 		"By default there are no limits on samples ingestion rate.")
-	finalDedupScheduleInterval = flag.Duration("storage.finalDedupScheduleCheckInterval", time.Hour, "The interval for checking when final deduplication process should be started."+
-		"Storage unconditionally adds 25% jitter to the interval value on each check evaluation."+
-		" Changing the interval to the bigger values may delay downsampling, deduplication for historical data."+
-		" See also https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#deduplication")
+	maxConcurrentRequests = flag.Int("search.maxConcurrentRequests", getDefaultMaxConcurrentRequests(), "The maximum number of concurrent search requests. "+
+		"It shouldn't be high, since a single request can saturate all the CPU cores, while many concurrently executed requests may require high amounts of memory. "+
+		"See also -search.maxQueueDuration and -search.maxMemoryPerQuery")
+	maxQueueDuration = flag.Duration("search.maxQueueDuration", 10*time.Second, "The maximum time the request waits for execution when -search.maxConcurrentRequests "+
+		"limit is reached; see also -search.maxQueryDuration")
 )
 
+func getDefaultMaxConcurrentRequests() int {
+	// A single request can saturate all the CPU cores, so there is no sense
+	// in allowing higher number of concurrent requests - they will just contend
+	// for unavailable CPU time.
+	n := min(cgroup.AvailableCPUs()*2, 16)
+	return n
+}
+
 func main() {
 	// VictoriaMetrics is optimized for reduced memory allocations,
 	// so it can run with the reduced GOGC in order to reduce the used memory,
@@ -87,14 +89,8 @@ func main() {
 	}
 	logger.Infof("starting VictoriaMetrics at %q...", listenAddrs)
 	startTime := time.Now()
-	storage.SetDedupInterval(*minScrapeInterval)
-	storage.SetDataFlushInterval(*inmemoryDataFlushInterval)
-	if *finalDedupScheduleInterval < time.Hour {
-		logger.Fatalf("-dedup.finalDedupScheduleCheckInterval cannot be smaller than 1 hour; got %s", *finalDedupScheduleInterval)
-	}
-	storage.SetFinalDedupScheduleInterval(*finalDedupScheduleInterval)
-	vmstorage.Init(promql.ResetRollupResultCacheIfNeeded)
-	vmselect.Init()
+	vmstorage.Init(*maxConcurrentRequests, promql.ResetRollupResultCacheIfNeeded)
+	vmselect.Init(*maxConcurrentRequests, *maxQueueDuration)
 	vminsertcommon.StartIngestionRateLimiter(*maxIngestionRate)
 	vminsert.Init()
 
@@ -118,8 +114,8 @@ func main() {
 		logger.Fatalf("cannot stop the webservice: %s", err)
 	}
 	logger.Infof("successfully shut down the webservice in %.3f seconds", time.Since(startTime).Seconds())
-	vminsert.Stop()
 	vminsertcommon.StopIngestionRateLimiter()
+	vminsert.Stop()
 
 	vmstorage.Stop()
 	vmselect.Stop()

app/victoria-metrics/self_scraper.go

@@ -93,7 +93,7 @@ func selfScraper(scrapeInterval time.Duration) {
 				mr.Value = r.Value
 			}
 		}
-		if err := vmstorage.AddRows(mrs); err != nil {
+		if err := vmstorage.VMInsertAPI.WriteRows(mrs); err != nil {
 			logger.Errorf("cannot store self-scraped metrics: %s", err)
 		}
 		if len(metadataRows.Rows) > 0 {
@@ -105,7 +105,7 @@ func selfScraper(scrapeInterval time.Duration) {
 					Type:             mm.Type,
 				})
 			}
-			if err := vmstorage.AddMetadataRows(mms); err != nil {
+			if err := vmstorage.VMInsertAPI.WriteMetadata(mms); err != nil {
 				logger.Errorf("cannot store self-scraped metrics metadata: %s", err)
 			}
 		}

app/vmagent/main.go

@@ -83,6 +83,9 @@ var (
 	maxLabelsPerTimeseries = flag.Int("maxLabelsPerTimeseries", 0, "The maximum number of labels per time series to be accepted. Series with superfluous labels are ignored. In this case the vm_rows_ignored_total{reason=\"too_many_labels\"} metric at /metrics page is incremented")
 	maxLabelNameLen        = flag.Int("maxLabelNameLen", 0, "The maximum length of label names in the accepted time series. Series with longer label name are ignored. In this case the vm_rows_ignored_total{reason=\"too_long_label_name\"} metric at /metrics page is incremented")
 	maxLabelValueLen       = flag.Int("maxLabelValueLen", 0, "The maximum length of label values in the accepted time series. Series with longer label value are ignored. In this case the vm_rows_ignored_total{reason=\"too_long_label_value\"} metric at /metrics page is incremented")
+
+	enableMultitenancyViaHeaders = flag.Bool("enableMultitenancyViaHeaders", false, "Enables multitenancy via HTTP headers. "+
+		"See https://docs.victoriametrics.com/victoriametrics/vmagent/#multitenancy")
 )
 
 var (
@@ -115,6 +118,7 @@ func main() {
 	remotewrite.InitSecretFlags()
 	buildinfo.Init()
 	logger.Init()
+	opentelemetry.Init()
 	timeserieslimits.Init(*maxLabelsPerTimeseries, *maxLabelNameLen, *maxLabelValueLen)
 
 	if promscrape.IsDryRun() {
@@ -216,7 +220,7 @@ func getOpenTSDBHTTPInsertHandler() func(req *http.Request) error {
 	}
 	return func(req *http.Request) error {
 		path := strings.ReplaceAll(req.URL.Path, "//", "/")
-		at, err := getAuthTokenFromPath(path)
+		at, err := getAuthTokenFromPath(path, req.Header)
 		if err != nil {
 			return fmt.Errorf("cannot obtain auth token from path %q: %w", path, err)
 		}
@@ -224,8 +228,15 @@ func getOpenTSDBHTTPInsertHandler() func(req *http.Request) error {
 	}
 }
 
-func getAuthTokenFromPath(path string) (*auth.Token, error) {
-	p, err := httpserver.ParsePath(path)
+func parsePath(path string, header http.Header) (*httpserver.Path, error) {
+	if *enableMultitenancyViaHeaders {
+		return httpserver.ParsePathAndHeaders(path, header)
+	}
+	return httpserver.ParsePath(path)
+}
+
+func getAuthTokenFromPath(path string, header http.Header) (*auth.Token, error) {
+	p, err := parsePath(path, header)
 	if err != nil {
 		return nil, fmt.Errorf("cannot parse multitenant path: %w", err)
 	}
@@ -559,14 +570,15 @@ func requestHandler(w http.ResponseWriter, r *http.Request) bool {
 }
 
 func processMultitenantRequest(w http.ResponseWriter, r *http.Request, path string) bool {
-	p, err := httpserver.ParsePath(path)
+	p, err := parsePath(path, r.Header)
 	if err != nil {
 		// Cannot parse multitenant path. Skip it - probably it will be parsed later.
 		return false
 	}
 	if p.Prefix != "insert" {
-		httpserver.Errorf(w, r, `unsupported multitenant prefix: %q; expected "insert"`, p.Prefix)
-		return true
+		// processMultitenantRequest is called for all unmatched path variants,
+		// but we should try parsing only /insert prefixed to avoid catching all possible paths.
+		return false
 	}
 	at, err := auth.NewTokenPossibleMultitenant(p.AuthToken)
 	if err != nil {

app/vmagent/opentelemetry/request_handler.go

@@ -25,6 +25,11 @@ var (
 	rowsPerInsert          = metrics.NewHistogram(`vmagent_rows_per_insert{type="opentelemetry"}`)
 )
 
+// Init must be called after flag.Parse and before using the opentelemetry package.
+func Init() {
+	stream.InitDecodeOptions()
+}
+
 // InsertHandlerForReader processes metrics from given reader.
 func InsertHandlerForReader(at *auth.Token, r io.Reader, encoding string) error {
 	return stream.ParseStream(r, encoding, nil, func(tss []prompb.TimeSeries, mms []prompb.MetricMetadata) error {
@@ -77,16 +82,6 @@ func insertRows(at *auth.Token, tss []prompb.TimeSeries, mms []prompb.MetricMeta
 
 	var metadataTotal int
 	if prommetadata.IsEnabled() {
-		var accountID, projectID uint32
-		if at != nil {
-			accountID = at.AccountID
-			projectID = at.ProjectID
-			for i := range mms {
-				mm := &mms[i]
-				mm.AccountID = accountID
-				mm.ProjectID = projectID
-			}
-		}
 		ctx.WriteRequest.Metadata = mms
 		metadataTotal = len(mms)
 	}

app/vmagent/prometheusimport/request_handler.go

@@ -75,11 +75,6 @@ func insertRows(at *auth.Token, rows []prometheus.Row, mms []prometheus.Metadata
 			Samples: samples[len(samples)-1:],
 		})
 	}
-	var accountID, projectID uint32
-	if at != nil {
-		accountID = at.AccountID
-		projectID = at.ProjectID
-	}
 	for i := range mms {
 		mm := &mms[i]
 		mmsDst = append(mmsDst, prompb.MetricMetadata{
@@ -88,8 +83,6 @@ func insertRows(at *auth.Token, rows []prometheus.Row, mms []prometheus.Metadata
 			Type:             mm.Type,
 			// there is no unit in Prometheus exposition formats
 
-			AccountID: accountID,
-			ProjectID: projectID,
 		})
 	}
 	ctx.WriteRequest.Timeseries = tssDst

app/vmagent/promremotewrite/request_handler.go

@@ -72,11 +72,6 @@ func insertRows(at *auth.Token, timeseries []prompb.TimeSeries, mms []prompb.Met
 
 	var metadataTotal int
 	if prommetadata.IsEnabled() {
-		var accountID, projectID uint32
-		if at != nil {
-			accountID = at.AccountID
-			projectID = at.ProjectID
-		}
 		for i := range mms {
 			mm := &mms[i]
 			mmsDst = append(mmsDst, prompb.MetricMetadata{
@@ -85,8 +80,8 @@ func insertRows(at *auth.Token, timeseries []prompb.TimeSeries, mms []prompb.Met
 				Type:             mm.Type,
 				Unit:             mm.Unit,
 
-				AccountID: accountID,
-				ProjectID: projectID,
+				AccountID: mm.AccountID,
+				ProjectID: mm.ProjectID,
 			})
 		}
 		ctx.WriteRequest.Metadata = mmsDst

app/vmagent/remotewrite/client.go

@@ -2,6 +2,7 @@ package remotewrite
 
 import (
 	"bytes"
+	"context"
 	"errors"
 	"fmt"
 	"io"
@@ -59,6 +60,8 @@ var (
 		"Multiple headers must be delimited by '^^': -remoteWrite.headers='header1:value1^^header2:value2'")
 
 	basicAuthUsername     = flagutil.NewArrayString("remoteWrite.basicAuth.username", "Optional basic auth username to use for the corresponding -remoteWrite.url")
+	basicAuthUsernameFile = flagutil.NewArrayString("remoteWrite.basicAuth.usernameFile", "Optional path to basic auth username to use for the corresponding -remoteWrite.url. "+
+		"The file is re-read every second")
 	basicAuthPassword     = flagutil.NewArrayString("remoteWrite.basicAuth.password", "Optional basic auth password to use for the corresponding -remoteWrite.url")
 	basicAuthPasswordFile = flagutil.NewArrayString("remoteWrite.basicAuth.passwordFile", "Optional path to basic auth password to use for the corresponding -remoteWrite.url. "+
 		"The file is re-read every second")
@@ -223,12 +226,14 @@ func getAuthConfig(argIdx int) (*promauth.Config, error) {
 		hdrs = strings.Split(headersValue, "^^")
 	}
 	username := basicAuthUsername.GetOptionalArg(argIdx)
+	usernameFile := basicAuthUsernameFile.GetOptionalArg(argIdx)
 	password := basicAuthPassword.GetOptionalArg(argIdx)
 	passwordFile := basicAuthPasswordFile.GetOptionalArg(argIdx)
 	var basicAuthCfg *promauth.BasicAuthConfig
-	if username != "" || password != "" || passwordFile != "" {
+	if username != "" || usernameFile != "" || password != "" || passwordFile != "" {
 		basicAuthCfg = &promauth.BasicAuthConfig{
 			Username:     username,
+			UsernameFile: usernameFile,
 			Password:     promauth.NewSecret(password),
 			PasswordFile: passwordFile,
 		}
@@ -306,11 +311,6 @@ func (c *client) runWorker() {
 		if !ok {
 			return
 		}
-		if len(block) == 0 {
-			// skip empty data blocks from sending
-			// see https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6241
-			continue
-		}
 		go func() {
 			startTime := time.Now()
 			ch <- c.sendBlock(block)
@@ -326,15 +326,20 @@ func (c *client) runWorker() {
 			c.fq.MustWriteBlockIgnoreDisabledPQ(block)
 			return
 		case <-c.stopCh:
-			// c must be stopped. Wait for a while in the hope the block will be sent.
-			graceDuration := 5 * time.Second
+			// c must be stopped. Wait up to 5 seconds for the in-flight request to complete.
+			// If it succeeds, drain the remaining in-memory queue before returning.
+			stopCtx, cancel := context.WithTimeout(context.Background(), time.Second*5)
+			defer cancel()
+
 			select {
 			case ok := <-ch:
 				if !ok {
 					// Return unsent block to the queue.
 					c.fq.MustWriteBlockIgnoreDisabledPQ(block)
+				} else {
+					c.drainInMemoryQueue(stopCtx, block[:0])
 				}
-			case <-time.After(graceDuration):
+			case <-stopCtx.Done():
 				// Return unsent block to the queue.
 				c.fq.MustWriteBlockIgnoreDisabledPQ(block)
 			}
@@ -466,7 +471,7 @@ again:
 				goto again
 			}
 
-			logger.Warnf("failed to repack zstd block (%s bytes) to snappy: %s; The block will be rejected. "+
+			logger.Warnf("failed to repack zstd block (%d bytes) to snappy: %s; The block will be rejected. "+
 				"Possible cause: ungraceful shutdown leading to persisted queue corruption.",
 				zstdBlockLen, err)
 		}
@@ -504,6 +509,32 @@ again:
 	goto again
 }
 
+func (c *client) drainInMemoryQueue(stopCtx context.Context, block []byte) {
+	var ok bool
+	for {
+		select {
+		case <-stopCtx.Done():
+			return
+		default:
+		}
+
+		block, ok = c.fq.MustReadInMemoryBlock(block[:0])
+		if !ok {
+			// The in memory queue has already been drained,
+			// or persisted queue is being used.
+			// In this case it is guaranteed that fq will be empty
+			return
+		}
+
+		// at this stage c.stopCh should be closed
+		// so sendBlock function should not perform retries
+		if ok := c.sendBlock(block); !ok {
+			c.fq.MustWriteBlockIgnoreDisabledPQ(block)
+			return
+		}
+	}
+}
+
 var remoteWriteRejectedLogger = logger.WithThrottler("remoteWriteRejected", 5*time.Second)
 var remoteWriteRetryLogger = logger.WithThrottler("remoteWriteRetry", 5*time.Second)
 

app/vmagent/remotewrite/client_test.go

@@ -9,6 +9,7 @@ import (
 	"github.com/golang/snappy"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 )
 
 func TestParseRetryAfterHeader(t *testing.T) {
@@ -36,6 +37,40 @@ func TestParseRetryAfterHeader(t *testing.T) {
 	f(time.Now().Add(10*time.Second).Format("Mon, 02 Jan 2006 15:04:05 FAKETZ"), 0)
 }
 
+func TestInitSecretFlags(t *testing.T) {
+	showRemoteWriteURLOrig := *showRemoteWriteURL
+	defer func() {
+		*showRemoteWriteURL = showRemoteWriteURLOrig
+		flagutil.UnregisterAllSecretFlags()
+	}()
+
+	flagutil.UnregisterAllSecretFlags()
+	*showRemoteWriteURL = false
+	InitSecretFlags()
+	if !flagutil.IsSecretFlag("remotewrite.url") {
+		t.Fatalf("expecting remoteWrite.url to be secret")
+	}
+	if !flagutil.IsSecretFlag("remotewrite.headers") {
+		t.Fatalf("expecting remoteWrite.headers to be secret")
+	}
+	if !flagutil.IsSecretFlag("remotewrite.proxyurl") {
+		t.Fatalf("expecting remoteWrite.proxyURL to be secret")
+	}
+
+	flagutil.UnregisterAllSecretFlags()
+	*showRemoteWriteURL = true
+	InitSecretFlags()
+	if flagutil.IsSecretFlag("remotewrite.url") {
+		t.Fatalf("remoteWrite.url must remain visible when -remoteWrite.showURL is set")
+	}
+	if !flagutil.IsSecretFlag("remotewrite.headers") {
+		t.Fatalf("expecting remoteWrite.headers to remain secret")
+	}
+	if !flagutil.IsSecretFlag("remotewrite.proxyurl") {
+		t.Fatalf("expecting remoteWrite.proxyURL to remain secret")
+	}
+}
+
 func TestRepackBlockFromZstdToSnappy(t *testing.T) {
 	expectedPlainBlock := []byte(`foobar`)
 

app/vmagent/remotewrite/pendingseries.go

@@ -211,6 +211,9 @@ func (wr *writeRequest) copyMetadata(dst, src *prompb.MetricMetadata) {
 	dst.Type = src.Type
 	dst.Unit = src.Unit
 
+	dst.AccountID = src.AccountID
+	dst.ProjectID = src.ProjectID
+
 	// Pre-allocate memory for all string fields.
 	neededBufLen := len(src.MetricFamilyName) + len(src.Help)
 	bufLen := len(wr.metadatabuf)

app/vmagent/remotewrite/remotewrite.go

@@ -151,6 +151,10 @@ func InitSecretFlags() {
 		// remoteWrite.url can contain authentication codes, so hide it at `/metrics` output.
 		flagutil.RegisterSecretFlag("remoteWrite.url")
 	}
+	// remoteWrite.proxyURL can contain authentication codes.
+	flagutil.RegisterSecretFlag("remoteWrite.proxyURL")
+	// remoteWrite.headers can contain auth headers such as Authorization and API keys.
+	flagutil.RegisterSecretFlag("remoteWrite.headers")
 }
 
 var (
@@ -167,6 +171,18 @@ func Init() {
 	if len(*remoteWriteURLs) == 0 {
 		logger.Fatalf("at least one `-remoteWrite.url` command-line flag must be set")
 	}
+	if *shardByURL && len(*disableOnDiskQueue) > 1 {
+		disableOnDiskQueues := *disableOnDiskQueue
+
+		firstValue := disableOnDiskQueues[0]
+		for _, v := range disableOnDiskQueues[1:] {
+			if firstValue != v {
+				logger.Fatalf("all -remoteWrite.url targets must have the same -remoteWrite.disableOnDiskQueue setting when -remoteWrite.shardByURL is enabled; " +
+					"either enable or disable -remoteWrite.disableOnDiskQueue for all targets")
+			}
+		}
+	}
+
 	if limit := getMaxHourlySeries(); limit > 0 {
 		hourlySeriesLimiter = bloomfilter.NewLimiter(limit, time.Hour)
 		_ = metrics.NewGauge(`vmagent_hourly_series_limit_max_series`, func() float64 {
@@ -285,6 +301,7 @@ func initRemoteWriteCtxs(urls []string) {
 		rwctxs[i] = newRemoteWriteCtx(i, remoteWriteURL, sanitizedURL)
 		rwctxIdx[i] = i
 	}
+	fs.RegisterPathFsMetrics(*tmpDataPath)
 
 	if *shardByURL {
 		consistentHashNodes := make([]string, 0, len(urls))
@@ -398,7 +415,7 @@ func tryPush(at *auth.Token, wr *prompb.WriteRequest, forceDropSamplesOnFailure
 
 	// Push metadata separately from time series, since it doesn't need sharding,
 	// relabeling, stream aggregation, deduplication, etc.
-	if !tryPushMetadataToRemoteStorages(rwctxs, mms, forceDropSamplesOnFailure) {
+	if !tryPushMetadataToRemoteStorages(at, rwctxs, mms, forceDropSamplesOnFailure) {
 		return false
 	}
 
@@ -498,7 +515,9 @@ func tryPush(at *auth.Token, wr *prompb.WriteRequest, forceDropSamplesOnFailure
 //
 // calculateHealthyRwctxIdx will rely on the order of rwctx to be in ascending order.
 func getEligibleRemoteWriteCtxs(tss []prompb.TimeSeries, forceDropSamplesOnFailure bool) ([]*remoteWriteCtx, bool) {
-	if !disableOnDiskQueueAny {
+	// When -remoteWrite.shardByURL=true always use all configured remote writes to preserve stable metrics distribution across shards.
+	// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/10507
+	if !disableOnDiskQueueAny || *shardByURL {
 		return rwctxsGlobal, true
 	}
 
@@ -513,12 +532,6 @@ func getEligibleRemoteWriteCtxs(tss []prompb.TimeSeries, forceDropSamplesOnFailu
 				return nil, false
 			}
 			rowsCount := getRowsCount(tss)
-			if *shardByURL {
-				// Todo: When shardByURL is enabled, the following metrics won't be 100% accurate. Because vmagent don't know
-				// which rwctx should data be pushed to yet. Let's consider the hashing algorithm fair and will distribute
-				// data to all rwctxs evenly.
-				rowsCount = rowsCount / len(rwctxsGlobal)
-			}
 			rwctx.rowsDroppedOnPushFailure.Add(rowsCount)
 		}
 	}
@@ -536,11 +549,18 @@ func pushTimeSeriesToRemoteStoragesTrackDropped(tss []prompb.TimeSeries) {
 	}
 }
 
-func tryPushMetadataToRemoteStorages(rwctxs []*remoteWriteCtx, mms []prompb.MetricMetadata, forceDropSamplesOnFailure bool) bool {
+func tryPushMetadataToRemoteStorages(at *auth.Token, rwctxs []*remoteWriteCtx, mms []prompb.MetricMetadata, forceDropSamplesOnFailure bool) bool {
 	if len(mms) == 0 {
 		// Nothing to push
 		return true
 	}
+	if at != nil {
+		for idx := range mms {
+			mm := &mms[idx]
+			mm.AccountID = at.AccountID
+			mm.ProjectID = at.ProjectID
+		}
+	}
 	// Do not shard metadata even if -remoteWrite.shardByURL is set, just replicate it among rwctxs.
 	// Since metadata is usually small and there is no guarantee that metadata can be sent to
 	// the same remote storage with the corresponding metrics.
@@ -691,7 +711,7 @@ func shardAmountRemoteWriteCtx(tssBlock []prompb.TimeSeries, shards [][]prompb.T
 			}
 			tmpLabels.Labels = hashLabels
 		}
-		h := getLabelsHash(hashLabels)
+		h := getLabelsHashForShard(hashLabels)
 
 		// Get the rwctxIdx through consistent hashing and then map it to the index in shards.
 		// The rwctxIdx is not always equal to the shardIdx, for example, when some rwctx are not available.
@@ -782,11 +802,28 @@ var (
 	dailySeriesLimitRowsDropped  = metrics.NewCounter(`vmagent_daily_series_limit_rows_dropped_total`)
 )
 
+// getLabelsHashForShard is a separate function from getLabelsHash because
+// it omits the '=' separator between label name and value for backward compatibility.
+// Changing it would re-shard all series across remoteWrite targets.
+func getLabelsHashForShard(labels []prompb.Label) uint64 {
+	bb := labelsHashBufPool.Get()
+	b := bb.B[:0]
+	for _, label := range labels {
+		b = append(b, label.Name...)
+		b = append(b, label.Value...)
+	}
+	h := xxhash.Sum64(b)
+	bb.B = b
+	labelsHashBufPool.Put(bb)
+	return h
+}
+
 func getLabelsHash(labels []prompb.Label) uint64 {
 	bb := labelsHashBufPool.Get()
 	b := bb.B[:0]
 	for _, label := range labels {
 		b = append(b, label.Name...)
+		b = append(b, '=')
 		b = append(b, label.Value...)
 	}
 	h := xxhash.Sum64(b)

app/vmagent/remotewrite/remotewrite_test.go

@@ -25,7 +25,7 @@ func TestGetLabelsHash_Distribution(t *testing.T) {
 		t.Helper()
 
 		// Distribute itemsCount hashes returned by getLabelsHash() across bucketsCount buckets.
-		itemsCount := 1_000 * bucketsCount
+		itemsCount := 10_000 * bucketsCount
 		m := make([]int, bucketsCount)
 		var labels []prompb.Label
 		for i := range itemsCount {
@@ -44,10 +44,12 @@ func TestGetLabelsHash_Distribution(t *testing.T) {
 		}
 
 		// Verify that the distribution is even
-		expectedItemsPerBucket := itemsCount / bucketsCount
+		expectedItemsPerBucket := float64(itemsCount / bucketsCount)
+		allowedDeviation := math.Round(float64(expectedItemsPerBucket) * 0.04)
 		for _, n := range m {
-			if math.Abs(1-float64(n)/float64(expectedItemsPerBucket)) > 0.04 {
-				t.Fatalf("unexpected items in the bucket for %d buckets; got %d; want around %d", bucketsCount, n, expectedItemsPerBucket)
+			if math.Abs(expectedItemsPerBucket-float64(n)) > allowedDeviation {
+				t.Fatalf("unexpected items in the bucket for %d buckets; got %d; want in range [%.0f, %.0f]",
+					bucketsCount, n, expectedItemsPerBucket-allowedDeviation, expectedItemsPerBucket+allowedDeviation)
 			}
 		}
 	}

app/vmalert-tool/unittest/input.go

@@ -52,7 +52,7 @@ func writeInputSeries(input []series, interval *promutil.Duration, startStamp ti
 	data := testutil.Compress(r)
 	// write input series to vm
 	httpWrite(dst, bytes.NewBuffer(data))
-	vmstorage.Storage.DebugFlush()
+	vmstorage.DebugFlush()
 	return nil
 }
 

app/vmalert-tool/unittest/unittest.go

@@ -61,7 +61,7 @@ func UnitTest(files []string, disableGroupLabel bool, externalLabels []string, e
 	}
 	eu, err := url.Parse(externalURL)
 	if err != nil {
-		logger.Fatalf("failed to parse external URL: %w", err)
+		logger.Fatalf("failed to parse external URL: %s", err)
 	}
 	if err := templates.Load([]string{}, *eu); err != nil {
 		logger.Fatalf("failed to load template: %v", err)
@@ -108,7 +108,9 @@ func UnitTest(files []string, disableGroupLabel bool, externalLabels []string, e
 	storagePath = tmpFolder
 	processFlags()
 	vminsert.Init()
-	vmselect.Init()
+	const maxConcurrentRequests = 4
+	maxQueueDuration := 5 * time.Second
+	vmselect.Init(maxConcurrentRequests, maxQueueDuration)
 	// storagePath will be created again when closing vmselect, so remove it again.
 	defer fs.MustRemoveDir(storagePath)
 	defer vminsert.Stop()
@@ -279,7 +281,8 @@ func processFlags() {
 }
 
 func setUp() {
-	vmstorage.Init(promql.ResetRollupResultCacheIfNeeded)
+	const maxConcurrentRequests = 4
+	vmstorage.Init(maxConcurrentRequests, promql.ResetRollupResultCacheIfNeeded)
 	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
 	defer cancel()
 	readyCheckFunc := func() bool {
@@ -384,7 +387,7 @@ func (tg *testGroup) test(evalInterval time.Duration, groupOrderMap map[string]i
 				}
 			}
 			// flush series after each group evaluation
-			vmstorage.Storage.DebugFlush()
+			vmstorage.DebugFlush()
 		}
 
 		// check alert_rule_test case at every eval time

app/vmalert/config/config.go

@@ -113,15 +113,15 @@ func (g *Group) Validate(validateTplFn ValidateTplFn, validateExpressions bool)
 			// because correct types must be inherited after unmarshalling.
 			exprValidator := g.Type.ValidateExpr
 			if err := exprValidator(r.Expr); err != nil {
-				return fmt.Errorf("invalid expression for rule  %q: %w", ruleName, err)
+				return fmt.Errorf("invalid expression for rule %q: %w", ruleName, err)
 			}
 		}
 		if validateTplFn != nil {
 			if err := validateTplFn(r.Annotations); err != nil {
-				return fmt.Errorf("invalid annotations for rule  %q: %w", ruleName, err)
+				return fmt.Errorf("invalid annotations for rule %q: %w", ruleName, err)
 			}
 			if err := validateTplFn(r.Labels); err != nil {
-				return fmt.Errorf("invalid labels for rule  %q: %w", ruleName, err)
+				return fmt.Errorf("invalid labels for rule %q: %w", ruleName, err)
 			}
 		}
 	}
@@ -222,6 +222,9 @@ func (r *Rule) Validate() error {
 	if r.Expr == "" {
 		return fmt.Errorf("expression can't be empty")
 	}
+	if _, ok := r.Labels["__name__"]; ok {
+		return fmt.Errorf("invalid rule label __name__")
+	}
 	return checkOverflow(r.XXX, "rule")
 }
 

app/vmalert/config/config_test.go

@@ -121,7 +121,7 @@ func TestParse_Failure(t *testing.T) {
 	f([]string{"testdata/dir/rules2-bad.rules"}, "function \"unknown\" not defined")
 	f([]string{"testdata/dir/rules3-bad.rules"}, "either `record` or `alert` must be set")
 	f([]string{"testdata/dir/rules4-bad.rules"}, "either `record` or `alert` must be set")
-	f([]string{"testdata/rules/rules1-bad.rules"}, "bad graphite expr")
+	f([]string{"testdata/rules/rules1-bad.rules"}, "bad GraphiteQL expr")
 	f([]string{"testdata/rules/vlog-rules0-bad.rules"}, "bad LogsQL expr")
 	f([]string{"testdata/dir/rules6-bad.rules"}, "missing ':' in header")
 	f([]string{"testdata/rules/rules-multi-doc-bad.rules"}, "unknown fields")
@@ -136,6 +136,9 @@ func TestRuleValidate(t *testing.T) {
 	if err := (&Rule{Alert: "alert"}).Validate(); err == nil {
 		t.Fatalf("expected empty expr error")
 	}
+	if err := (&Rule{Record: "record", Expr: "sum(test)", Labels: map[string]string{"__name__": "test"}}).Validate(); err == nil {
+		t.Fatalf("invalid rule label; got %s", err)
+	}
 	if err := (&Rule{Alert: "alert", Expr: "test>0"}).Validate(); err != nil {
 		t.Fatalf("expected valid rule; got %s", err)
 	}
@@ -280,7 +283,7 @@ func TestGroupValidate_Failure(t *testing.T) {
 				Expr:   "up | 0",
 			},
 		},
-	}, true, "bad prometheus expr")
+	}, true, "bad MetricsQL expr")
 
 	f(&Group{
 		Name: "test graphite expr",
@@ -290,7 +293,7 @@ func TestGroupValidate_Failure(t *testing.T) {
 				"description": "some-description",
 			}},
 		},
-	}, true, "bad graphite expr")
+	}, true, "bad GraphiteQL expr")
 
 	f(&Group{
 		Name: "test vlogs expr",
@@ -324,7 +327,7 @@ func TestGroupValidate_Failure(t *testing.T) {
 				Expr:   "sum(up == 0 ) by (host)",
 			},
 		},
-	}, true, "bad graphite expr")
+	}, true, "bad GraphiteQL expr")
 
 	f(&Group{
 		Name: "test vlogs with prometheus exp",
@@ -348,7 +351,7 @@ func TestGroupValidate_Failure(t *testing.T) {
 				For:    promutil.NewDuration(10 * time.Millisecond),
 			},
 		},
-	}, true, "bad prometheus expr")
+	}, true, "bad MetricsQL expr")
 }
 
 func TestGroupValidate_Success(t *testing.T) {

app/vmalert/config/types.go

@@ -66,11 +66,11 @@ func (t *Type) ValidateExpr(expr string) error {
 	switch t.String() {
 	case "graphite":
 		if _, err := graphiteql.Parse(expr); err != nil {
-			return fmt.Errorf("bad graphite expr: %q, err: %w", expr, err)
+			return fmt.Errorf("bad GraphiteQL expr: %q, err: %w", expr, err)
 		}
 	case "prometheus":
 		if _, err := metricsql.Parse(expr); err != nil {
-			return fmt.Errorf("bad prometheus expr: %q, err: %w", expr, err)
+			return fmt.Errorf("bad MetricsQL expr: %q, err: %w", expr, err)
 		}
 	case "vlogs":
 		q, err := logstorage.ParseStatsQuery(expr, 0)

app/vmalert/datasource/client_test.go

@@ -772,7 +772,7 @@ func TestHeaders(t *testing.T) {
 
 	// basic auth
 	f(func() *Client {
-		cfg, err := vmalertutil.AuthConfig(vmalertutil.WithBasicAuth("foo", "bar", ""))
+		cfg, err := vmalertutil.AuthConfig(vmalertutil.WithBasicAuth("foo", "", "bar", ""))
 		if err != nil {
 			t.Fatalf("Error get auth config: %s", err)
 		}
@@ -817,7 +817,7 @@ func TestHeaders(t *testing.T) {
 
 	// custom header overrides basic auth
 	f(func() *Client {
-		cfg, err := vmalertutil.AuthConfig(vmalertutil.WithBasicAuth("foo", "bar", ""))
+		cfg, err := vmalertutil.AuthConfig(vmalertutil.WithBasicAuth("foo", "", "bar", ""))
 		if err != nil {
 			t.Fatalf("Error get auth config: %s", err)
 		}

app/vmalert/datasource/datasource.go

@@ -87,6 +87,7 @@ func (m *Metric) DelLabel(key string) {
 	for i, l := range m.Labels {
 		if l.Name == key {
 			m.Labels = append(m.Labels[:i], m.Labels[i+1:]...)
+			break
 		}
 	}
 }

app/vmalert/datasource/init.go

@@ -27,6 +27,7 @@ var (
 		"Multiple headers must be delimited by '^^': -datasource.headers='header1:value1^^header2:value2'")
 
 	basicAuthUsername     = flag.String("datasource.basicAuth.username", "", "Optional basic auth username for -datasource.url")
+	basicAuthUsernameFile = flag.String("datasource.basicAuth.usernameFile", "", "Optional path to basic auth username to use for -datasource.url")
 	basicAuthPassword     = flag.String("datasource.basicAuth.password", "", "Optional basic auth password for -datasource.url")
 	basicAuthPasswordFile = flag.String("datasource.basicAuth.passwordFile", "", "Optional path to basic auth password to use for -datasource.url")
 
@@ -63,6 +64,7 @@ func InitSecretFlags() {
 	if !*showDatasourceURL {
 		flagutil.RegisterSecretFlag("datasource.url")
 	}
+	flagutil.RegisterSecretFlag("datasource.headers")
 }
 
 // ShowDatasourceURL whether to show -datasource.url with sensitive information
@@ -105,7 +107,7 @@ func Init(extraParams url.Values) (QuerierBuilder, error) {
 		return nil, fmt.Errorf("cannot parse JSON for -datasource.oauth2.endpointParams=%s: %w", *oauth2EndpointParams, err)
 	}
 	authCfg, err := vmalertutil.AuthConfig(
-		vmalertutil.WithBasicAuth(*basicAuthUsername, *basicAuthPassword, *basicAuthPasswordFile),
+		vmalertutil.WithBasicAuth(*basicAuthUsername, *basicAuthUsernameFile, *basicAuthPassword, *basicAuthPasswordFile),
 		vmalertutil.WithBearer(*bearerToken, *bearerTokenFile),
 		vmalertutil.WithOAuth(*oauth2ClientID, *oauth2ClientSecret, *oauth2ClientSecretFile, *oauth2TokenURL, *oauth2Scopes, endpointParams),
 		vmalertutil.WithHeaders(*headers))

app/vmalert/notifier/alertmanager.go

@@ -191,7 +191,7 @@ func NewAlertManager(alertManagerURL string, fn AlertURLGenerator, authCfg proma
 	}
 
 	aCfg, err := vmalertutil.AuthConfig(
-		vmalertutil.WithBasicAuth(ba.Username, ba.Password.String(), ba.PasswordFile),
+		vmalertutil.WithBasicAuth(ba.Username, ba.UsernameFile, ba.Password.String(), ba.PasswordFile),
 		vmalertutil.WithBearer(authCfg.BearerToken.String(), authCfg.BearerTokenFile),
 		vmalertutil.WithOAuth(oauth.ClientID, oauth.ClientSecret.String(), oauth.ClientSecretFile, oauth.TokenURL, strings.Join(oauth.Scopes, ";"), oauth.EndpointParams),
 		vmalertutil.WithHeaders(strings.Join(authCfg.Headers, "^^")),

app/vmalert/notifier/config_watcher.go

@@ -105,7 +105,7 @@ func (cw *configWatcher) add(typeK TargetType, interval time.Duration, targetsFn
 			}
 			targetMetadata, errors := getTargetMetadata(targetsFn, cw.cfg)
 			for _, err := range errors {
-				logger.Errorf("failed to init notifier for %q: %w", typeK, err)
+				logger.Errorf("failed to init notifier for %q: %s", typeK, err)
 			}
 			cw.updateTargets(typeK, targetMetadata, cw.cfg, cw.genFn)
 		}
@@ -274,7 +274,7 @@ func (cw *configWatcher) updateTargets(key TargetType, targetMts map[string]targ
 	for addr, metadata := range targetMts {
 		am, err := NewAlertManager(addr, genFn, cfg.HTTPClientConfig, metadata.alertRelabelConfigs, cfg.Timeout.Duration())
 		if err != nil {
-			logger.Errorf("failed to init %s notifier with addr %q: %w", key, addr, err)
+			logger.Errorf("failed to init %s notifier with addr %q: %s", key, addr, err)
 			continue
 		}
 		updatedTargets = append(updatedTargets, Target{

app/vmalert/notifier/init.go

@@ -36,6 +36,7 @@ var (
 		"For example, -remoteWrite.headers='My-Auth:foobar' would send 'My-Auth: foobar' HTTP header with every request to the corresponding -notifier.url. "+
 		"Multiple headers must be delimited by '^^': -notifier.headers='header1:value1^^header2:value2,header3:value3'")
 	basicAuthUsername     = flagutil.NewArrayString("notifier.basicAuth.username", "Optional basic auth username for -notifier.url")
+	basicAuthUsernameFile = flagutil.NewArrayString("notifier.basicAuth.usernameFile", "Optional path to basic auth username file for -notifier.url")
 	basicAuthPassword     = flagutil.NewArrayString("notifier.basicAuth.password", "Optional basic auth password for -notifier.url")
 	basicAuthPasswordFile = flagutil.NewArrayString("notifier.basicAuth.passwordFile", "Optional path to basic auth password file for -notifier.url")
 
@@ -193,6 +194,7 @@ func InitSecretFlags() {
 	if !*showNotifierURL {
 		flagutil.RegisterSecretFlag("notifier.url")
 	}
+	flagutil.RegisterSecretFlag("notifier.headers")
 }
 
 func notifiersFromFlags(gen AlertURLGenerator) ([]Notifier, error) {
@@ -213,6 +215,7 @@ func notifiersFromFlags(gen AlertURLGenerator) ([]Notifier, error) {
 			},
 			BasicAuth: &promauth.BasicAuthConfig{
 				Username:     basicAuthUsername.GetOptionalArg(i),
+				UsernameFile: basicAuthUsernameFile.GetOptionalArg(i),
 				Password:     promauth.NewSecret(basicAuthPassword.GetOptionalArg(i)),
 				PasswordFile: basicAuthPasswordFile.GetOptionalArg(i),
 			},

app/vmalert/notifier/notifier.go

@@ -14,7 +14,7 @@ type Notifier interface {
 	Send(ctx context.Context, alerts []Alert, alertLabels [][]prompb.Label, notifierHeaders map[string]string) error
 	// Addr returns address where alerts are sent.
 	Addr() string
-	// LastError returns error, that occured during last attempt to send data
+	// LastError returns error, that occurred during last attempt to send data
 	LastError() string
 	// Close is a destructor for the Notifier
 	Close()

app/vmalert/remoteread/init.go

@@ -28,6 +28,7 @@ var (
 		"Multiple headers must be delimited by '^^': -remoteRead.headers='header1:value1^^header2:value2'")
 
 	basicAuthUsername     = flag.String("remoteRead.basicAuth.username", "", "Optional basic auth username for -remoteRead.url")
+	basicAuthUsernameFile = flag.String("remoteRead.basicAuth.usernameFile", "", "Optional path to basic auth username to use for -remoteRead.url")
 	basicAuthPassword     = flag.String("remoteRead.basicAuth.password", "", "Optional basic auth password for -remoteRead.url")
 	basicAuthPasswordFile = flag.String("remoteRead.basicAuth.passwordFile", "", "Optional path to basic auth password to use for -remoteRead.url")
 
@@ -58,6 +59,7 @@ func InitSecretFlags() {
 	if !*showRemoteReadURL {
 		flagutil.RegisterSecretFlag("remoteRead.url")
 	}
+	flagutil.RegisterSecretFlag("remoteRead.headers")
 }
 
 // Init creates a Querier from provided flag values.
@@ -80,7 +82,7 @@ func Init() (datasource.QuerierBuilder, error) {
 		return nil, fmt.Errorf("cannot parse JSON for -remoteRead.oauth2.endpointParams=%s: %w", *oauth2EndpointParams, err)
 	}
 	authCfg, err := vmalertutil.AuthConfig(
-		vmalertutil.WithBasicAuth(*basicAuthUsername, *basicAuthPassword, *basicAuthPasswordFile),
+		vmalertutil.WithBasicAuth(*basicAuthUsername, *basicAuthUsernameFile, *basicAuthPassword, *basicAuthPasswordFile),
 		vmalertutil.WithBearer(*bearerToken, *bearerTokenFile),
 		vmalertutil.WithOAuth(*oauth2ClientID, *oauth2ClientSecret, *oauth2ClientSecretFile, *oauth2TokenURL, *oauth2Scopes, endpointParams),
 		vmalertutil.WithHeaders(*headers))

app/vmalert/remotewrite/client.go

@@ -11,6 +11,7 @@ import (
 	"path"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/cespare/xxhash/v2"
@@ -18,6 +19,8 @@ import (
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/bytesutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding/zstd"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httputil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
@@ -57,6 +60,11 @@ type Client struct {
 
 	wg     sync.WaitGroup
 	doneCh chan struct{}
+
+	// Whether to encode the write request with VictoriaMetrics remote write protocol.
+	// It is set to true by default, and will be switched to false if the client
+	// receives specific errors indicating that the remote storage doesn't support VictoriaMetrics remote write protocol.
+	isVMRemoteWrite atomic.Bool
 }
 
 // Config is config for remote write client.
@@ -116,6 +124,7 @@ func NewClient(ctx context.Context, cfg Config) (*Client, error) {
 		doneCh:        make(chan struct{}),
 		input:         make(chan prompb.TimeSeries, cfg.MaxQueueSize),
 	}
+	c.isVMRemoteWrite.Store(true)
 
 	for i := 0; i < cc; i++ {
 		c.wg.Go(func() {
@@ -265,8 +274,16 @@ func (c *Client) flush(ctx context.Context, wr *prompb.WriteRequest) {
 	defer wr.Reset()
 	defer bufferFlushDuration.UpdateDuration(time.Now())
 
-	data := wr.MarshalProtobuf(nil)
-	b := snappy.Encode(nil, data)
+	bb := writeRequestBufPool.Get()
+	bb.B = wr.MarshalProtobuf(bb.B[:0])
+	zb := compressBufPool.Get()
+	defer compressBufPool.Put(zb)
+	if c.isVMRemoteWrite.Load() {
+		zb.B = zstd.CompressLevel(zb.B[:0], bb.B, 0)
+	} else {
+		zb.B = snappy.Encode(zb.B[:cap(zb.B)], bb.B)
+	}
+	writeRequestBufPool.Put(bb)
 
 	maxRetryInterval := *retryMaxTime
 	bt := timeutil.NewBackoffTimer(*retryMinInterval, maxRetryInterval)
@@ -278,17 +295,17 @@ func (c *Client) flush(ctx context.Context, wr *prompb.WriteRequest) {
 	attempts := 0
 L:
 	for {
-		err := c.send(ctx, b)
+		err := c.send(ctx, zb.B)
 		if err != nil && (errors.Is(err, io.EOF) || netutil.IsTrivialNetworkError(err)) {
 			// Something in the middle between client and destination might be closing
 			// the connection. So we do a one more attempt in hope request will succeed.
-			err = c.send(ctx, b)
+			err = c.send(ctx, zb.B)
 		}
 		if err == nil {
 			sentRows.Add(len(wr.Timeseries))
-			sentBytes.Add(len(b))
+			sentBytes.Add(len(zb.B))
 			flushedRows.Update(float64(len(wr.Timeseries)))
-			flushedBytes.Update(float64(len(b)))
+			flushedBytes.Update(float64(len(zb.B)))
 			return
 		}
 
@@ -340,12 +357,16 @@ func (c *Client) send(ctx context.Context, data []byte) error {
 		return fmt.Errorf("failed to create new HTTP request: %w", err)
 	}
 
-	// RFC standard compliant headers
-	req.Header.Set("Content-Encoding", "snappy")
+	req.Header.Set("User-Agent", "vmalert")
 	req.Header.Set("Content-Type", "application/x-protobuf")
 
-	// Prometheus compliant headers
-	req.Header.Set("X-Prometheus-Remote-Write-Version", "0.1.0")
+	if encoding.IsZstd(data) {
+		req.Header.Set("Content-Encoding", "zstd")
+		req.Header.Set("X-VictoriaMetrics-Remote-Write-Version", "1")
+	} else {
+		req.Header.Set("Content-Encoding", "snappy")
+		req.Header.Set("X-Prometheus-Remote-Write-Version", "0.1.0")
+	}
 
 	if c.authCfg != nil {
 		err = c.authCfg.SetHeaders(req, true)
@@ -374,6 +395,29 @@ func (c *Client) send(ctx context.Context, data []byte) error {
 		// respond with HTTP 2xx status code when write is successful.
 		return nil
 	case 4:
+		// - Remote Write v1 specification implicitly expects a `400 Bad Request` when the encoding is not supported.
+		// - Remote Write v2 specification explicitly specifies a `415 Unsupported Media Type` for unsupported encodings.
+		// - Real-world implementations of v1 use both 400 and 415 status codes.
+		// See more in research: https://github.com/VictoriaMetrics/VictoriaMetrics/pull/8462#issuecomment-2786918054
+		if resp.StatusCode == http.StatusUnsupportedMediaType || resp.StatusCode == http.StatusBadRequest {
+			if encoding.IsZstd(data) {
+				logger.Infof("received unsupported media type or bad request from remote storage at %q. Re-packing the block to Prometheus remote write and retrying."+
+					"See https://docs.victoriametrics.com/victoriametrics/vmagent/#victoriametrics-remote-write-protocol", req.URL.Redacted())
+				zstdBlockLen := len(data)
+				data, err = repackBlockFromZstdToSnappy(data)
+				if err == nil {
+					logger.Infof("received unsupported media type or bad request from remote storage at %q. Downgrading protocol from VictoriaMetrics to Prometheus remote write for all future requests. "+
+						"See https://docs.victoriametrics.com/victoriametrics/vmagent/#victoriametrics-remote-write-protocol", req.URL.Redacted())
+					c.isVMRemoteWrite.Store(false)
+					return c.send(ctx, data)
+				}
+
+				logger.Warnf("failed to repack zstd block (%d bytes) to snappy: %s; The block will be rejected. "+
+					"Possible cause: ungraceful shutdown leading to persisted queue corruption.",
+					zstdBlockLen, err)
+			}
+		}
+
 		if resp.StatusCode != http.StatusTooManyRequests {
 			// MUST NOT retry write requests on HTTP 4xx responses other than 429
 			return &nonRetriableError{
@@ -394,3 +438,19 @@ type nonRetriableError struct {
 func (e *nonRetriableError) Error() string {
 	return e.err.Error()
 }
+
+var (
+	writeRequestBufPool bytesutil.ByteBufferPool
+	compressBufPool     bytesutil.ByteBufferPool
+)
+
+// repackBlockFromZstdToSnappy repacks the given zstd-compressed block to snappy-compressed block.
+func repackBlockFromZstdToSnappy(zstdBlock []byte) ([]byte, error) {
+	plainBlock := make([]byte, 0, len(zstdBlock)*2)
+	plainBlock, err := encoding.DecompressZSTD(plainBlock, zstdBlock)
+	if err != nil {
+		return nil, err
+	}
+
+	return snappy.Encode(nil, plainBlock), nil
+}

app/vmalert/remotewrite/client_test.go

@@ -12,8 +12,7 @@ import (
 	"testing"
 	"time"
 
-	"github.com/golang/snappy"
-
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding/zstd"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompb"
 )
 
@@ -103,7 +102,10 @@ func TestClient_run_maxBatchSizeDuringShutdown(t *testing.T) {
 
 		// push time series to the client.
 		for range pushCnt {
-			if err = rwClient.Push(prompb.TimeSeries{}); err != nil {
+			if err = rwClient.Push(prompb.TimeSeries{
+				Labels:  []prompb.Label{{Name: "__name__", Value: "m"}},
+				Samples: []prompb.Sample{{Value: 1, Timestamp: 1000}},
+			}); err != nil {
 				t.Fatalf("cannot time series to the client: %s", err)
 			}
 		}
@@ -156,8 +158,8 @@ func (rw *rwServer) handler(w http.ResponseWriter, r *http.Request) {
 	}
 
 	h := r.Header.Get("Content-Encoding")
-	if h != "snappy" {
-		rw.err(w, fmt.Errorf("header read error: Content-Encoding is not snappy (%q)", h))
+	if h != "zstd" {
+		rw.err(w, fmt.Errorf("header read error: Content-Encoding is not zstd (%q)", h))
 	}
 
 	h = r.Header.Get("Content-Type")
@@ -165,9 +167,9 @@ func (rw *rwServer) handler(w http.ResponseWriter, r *http.Request) {
 		rw.err(w, fmt.Errorf("header read error: Content-Type is not x-protobuf (%q)", h))
 	}
 
-	h = r.Header.Get("X-Prometheus-Remote-Write-Version")
-	if h != "0.1.0" {
-		rw.err(w, fmt.Errorf("header read error: X-Prometheus-Remote-Write-Version is not 0.1.0 (%q)", h))
+	h = r.Header.Get("X-VictoriaMetrics-Remote-Write-Version")
+	if h != "1" {
+		rw.err(w, fmt.Errorf("header read error: X-VictoriaMetrics-Remote-Write-Version is not 1 (%q)", h))
 	}
 
 	data, err := io.ReadAll(r.Body)
@@ -177,7 +179,7 @@ func (rw *rwServer) handler(w http.ResponseWriter, r *http.Request) {
 	}
 	defer func() { _ = r.Body.Close() }()
 
-	b, err := snappy.Decode(nil, data)
+	b, err := zstd.Decompress(nil, data)
 	if err != nil {
 		rw.err(w, fmt.Errorf("decode err: %w", err))
 		return

app/vmalert/remotewrite/debug_client.go

@@ -9,8 +9,7 @@ import (
 	"strings"
 	"sync"
 
-	"github.com/golang/snappy"
-
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding/zstd"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httputil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/promauth"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/prompb"
@@ -64,19 +63,17 @@ func (c *DebugClient) Close() error {
 }
 
 func (c *DebugClient) send(data []byte) error {
-	b := snappy.Encode(nil, data)
+	b := zstd.CompressLevel(nil, data, 0)
 	r := bytes.NewReader(b)
 	req, err := http.NewRequest(http.MethodPost, c.addr, r)
 	if err != nil {
 		return fmt.Errorf("failed to create new HTTP request: %w", err)
 	}
 
-	// RFC standard compliant headers
-	req.Header.Set("Content-Encoding", "snappy")
+	req.Header.Set("Content-Encoding", "zstd")
 	req.Header.Set("Content-Type", "application/x-protobuf")
 
-	// Prometheus compliant headers
-	req.Header.Set("X-Prometheus-Remote-Write-Version", "0.1.0")
+	req.Header.Set("X-VictoriaMetrics-Remote-Write-Version", "1")
 
 	if !*disablePathAppend {
 		req.URL.Path = path.Join(req.URL.Path, "/api/v1/write")

app/vmalert/remotewrite/init.go

@@ -13,8 +13,8 @@ import (
 )
 
 var (
-	addr = flag.String("remoteWrite.url", "", "Optional URL to VictoriaMetrics or vminsert where to persist alerts state "+
-		"and recording rules results in form of timeseries. "+
+	addr = flag.String("remoteWrite.url", "", "Optional URL to persist alerts state and recording rules results in form of timeseries. "+
+		"It must support either VictoriaMetrics remote write protocol or Prometheus remote_write protocol. "+
 		"Supports address in the form of IP address with a port (e.g., http://127.0.0.1:8428) or DNS SRV record. "+
 		"For example, if -remoteWrite.url=http://127.0.0.1:8428 is specified, "+
 		"then the alerts state will be written to http://127.0.0.1:8428/api/v1/write . See also -remoteWrite.disablePathAppend, '-remoteWrite.showURL'.")
@@ -26,6 +26,7 @@ var (
 		"Multiple headers must be delimited by '^^': -remoteWrite.headers='header1:value1^^header2:value2'")
 
 	basicAuthUsername     = flag.String("remoteWrite.basicAuth.username", "", "Optional basic auth username for -remoteWrite.url")
+	basicAuthUsernameFile = flag.String("remoteWrite.basicAuth.usernameFile", "", "Optional path to basic auth username to use for -remoteWrite.url")
 	basicAuthPassword     = flag.String("remoteWrite.basicAuth.password", "", "Optional basic auth password for -remoteWrite.url")
 	basicAuthPasswordFile = flag.String("remoteWrite.basicAuth.passwordFile", "", "Optional path to basic auth password to use for -remoteWrite.url")
 
@@ -61,6 +62,7 @@ func InitSecretFlags() {
 	if !*showRemoteWriteURL {
 		flagutil.RegisterSecretFlag("remoteWrite.url")
 	}
+	flagutil.RegisterSecretFlag("remoteWrite.headers")
 }
 
 // Init creates Client object from given flags.
@@ -83,7 +85,7 @@ func Init(ctx context.Context) (*Client, error) {
 		return nil, fmt.Errorf("cannot parse JSON for -remoteWrite.oauth2.endpointParams=%s: %w", *oauth2EndpointParams, err)
 	}
 	authCfg, err := vmalertutil.AuthConfig(
-		vmalertutil.WithBasicAuth(*basicAuthUsername, *basicAuthPassword, *basicAuthPasswordFile),
+		vmalertutil.WithBasicAuth(*basicAuthUsername, *basicAuthUsernameFile, *basicAuthPassword, *basicAuthPasswordFile),
 		vmalertutil.WithBearer(*bearerToken, *bearerTokenFile),
 		vmalertutil.WithOAuth(*oauth2ClientID, *oauth2ClientSecret, *oauth2ClientSecretFile, *oauth2TokenURL, *oauth2Scopes, endpointParams),
 		vmalertutil.WithHeaders(*headers))

app/vmalert/rule/alerting.go

@@ -312,9 +312,11 @@ type labelSet struct {
 // On k conflicts in origin set, the original value is preferred and copied
 // to processed with `exported_%k` key. The copy happens only if passed v isn't equal to origin[k] value.
 func (ls *labelSet) add(k, v string) {
-	// do not add label with empty value, since it has no meaning.
-	// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9984
+	// do not add label with empty value to the result, as it has no meaning:
+	// if the label already exists in the original query result, remove it to preserve compatibility with relabeling, see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/10766.
+	// otherwise, ignore the label, see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9984.
 	if v == "" {
+		delete(ls.processed, k)
 		return
 	}
 	ls.processed[k] = v

app/vmalert/rule/alerting_test.go

@@ -1363,6 +1363,7 @@ func TestAlertingRule_ToLabels(t *testing.T) {
 			{Name: "instance", Value: "0.0.0.0:8800"},
 			{Name: "group", Value: "vmalert"},
 			{Name: "alertname", Value: "ConfigurationReloadFailure"},
+			{Name: "pod", Value: "vmalert-0"},
 		},
 		Values:     []float64{1},
 		Timestamps: []int64{time.Now().UnixNano()},
@@ -1374,6 +1375,7 @@ func TestAlertingRule_ToLabels(t *testing.T) {
 			"group":         "vmalert",  // this shouldn't have effect since value in metric is equal
 			"invalid_label": "{{ .Values.mustRuntimeFail }}",
 			"empty_label":   "", // this should be dropped
+			"pod":           "", // this should remove the pod label from query result
 		},
 		Expr:      "sum(vmalert_alerting_rules_error) by(instance, group, alertname) > 0",
 		Name:      "AlertingRulesError",
@@ -1385,6 +1387,7 @@ func TestAlertingRule_ToLabels(t *testing.T) {
 		"group":         "vmalert",
 		"alertname":     "ConfigurationReloadFailure",
 		"alertgroup":    "vmalert",
+		"pod":           "vmalert-0",
 		"invalid_label": `error evaluating template: template: :1:298: executing "" at <.Values.mustRuntimeFail>: can't evaluate field Values in type notifier.tplData`,
 	}
 

app/vmalert/rule/group.go

@@ -8,6 +8,7 @@ import (
 	"hash/fnv"
 	"maps"
 	"net/url"
+	"path"
 	"sync"
 	"time"
 
@@ -42,6 +43,9 @@ var (
 		"For example, if lookback=1h then range from now() to now()-1h will be scanned.")
 	maxStartDelay = flag.Duration("group.maxStartDelay", 5*time.Minute, "Defines the max delay before starting the group evaluation. Group's start is artificially delayed for random duration on interval"+
 		" [0..min(--group.maxStartDelay, group.interval)]. This helps smoothing out the load on the configured datasource, so evaluations aren't executed too close to each other.")
+	ruleStripFilePath = flag.Bool("rule.stripFilePath", false, "Whether to strip rule file paths in logs and all API responses, including /metrics. "+
+		"For example, file path '/path/to/tenant_id/rules.yml' will be stripped to 'groupHashID/rules.yml'. "+
+		"This flag may be useful for hiding sensitive information in file paths, such as S3 bucket details.")
 )
 
 // Group is an entity for grouping rules
@@ -147,6 +151,12 @@ func NewGroup(cfg config.Group, qb datasource.QuerierBuilder, defaultInterval ti
 		g.EvalDelay = &cfg.EvalDelay.D
 	}
 	g.id = g.CreateID()
+	// strip file path from group.File after generated group ID when ruleStripFilePath is set,
+	// so it won't be exposed in logs and api responses
+	if *ruleStripFilePath {
+		_, filename := path.Split(g.File)
+		g.File = fmt.Sprintf("%d/%s", g.id, filename)
+	}
 	for _, h := range cfg.Headers {
 		g.Headers[h.Key] = h.Value
 	}
@@ -409,6 +419,9 @@ func (g *Group) Start(ctx context.Context, rw remotewrite.RWClient, rr datasourc
 	g.mu.Unlock()
 	defer g.evalCancel()
 
+	// start the interval ticker before the first evaluation,
+	// so that the evaluation timestamps of groups with the `eval_offset` option are also aligned,
+	// see https://github.com/VictoriaMetrics/VictoriaMetrics/pull/10773
 	t := time.NewTicker(g.Interval)
 	defer t.Stop()
 

app/vmalert/rule/group_test.go

@@ -742,3 +742,64 @@ func parseTime(t *testing.T, s string) time.Time {
 	}
 	return tt
 }
+
+func TestRuleStripFilePath(t *testing.T) {
+	configG := config.Group{
+		Name:        "group",
+		File:        "/var/local/test/rules.yaml",
+		Type:        config.NewRawType("prometheus"),
+		Concurrency: 1,
+		Rules: []config.Rule{
+			{
+				ID:    0,
+				Alert: "alert",
+			},
+			{
+				ID:     1,
+				Record: "record",
+			},
+		}}
+	qb := &datasource.FakeQuerier{}
+	g := NewGroup(configG, qb, 1*time.Minute, nil)
+
+	gID := g.id
+	if g.File != "/var/local/test/rules.yaml" {
+		t.Fatalf("expected file path to be unchanged; got %q instead", g.File)
+	}
+
+	for _, r := range g.Rules {
+		if ar, ok := r.(*AlertingRule); ok {
+			if ar.File != "/var/local/test/rules.yaml" {
+				t.Fatalf("expected rule file path to be unchanged; got %q instead", ar.File)
+			}
+		}
+		if rr, ok := r.(*RecordingRule); ok {
+			if rr.File != "/var/local/test/rules.yaml" {
+				t.Fatalf("expected rule file path to be unchanged; got %q instead", rr.File)
+			}
+		}
+	}
+
+	oldRuleStripFilePath := *ruleStripFilePath
+	*ruleStripFilePath = true
+	defer func() {
+		*ruleStripFilePath = oldRuleStripFilePath
+	}()
+	g = NewGroup(configG, qb, 1*time.Minute, nil)
+
+	if g.File != fmt.Sprintf("%d/rules.yaml", gID) {
+		t.Fatalf("expected file path to be stripped to %q; got %q instead", fmt.Sprintf("%d/rules.yaml", gID), g.File)
+	}
+	for _, r := range g.Rules {
+		if ar, ok := r.(*AlertingRule); ok {
+			if ar.File != fmt.Sprintf("%d/rules.yaml", gID) {
+				t.Fatalf("expected rule file path to be unchanged; got %q instead", ar.File)
+			}
+		}
+		if rr, ok := r.(*RecordingRule); ok {
+			if rr.File != fmt.Sprintf("%d/rules.yaml", gID) {
+				t.Fatalf("expected rule file path to be unchanged; got %q instead", rr.File)
+			}
+		}
+	}
+}

app/vmalert/rule/recording.go

@@ -293,9 +293,11 @@ func (rr *RecordingRule) toTimeSeries(m datasource.Metric) prompb.TimeSeries {
 	}
 	// add extra labels configured by user
 	for k := range rr.Labels {
-		// do not add label with empty value, since it has no meaning.
-		// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9984
+		// do not add label with empty value to the result, as it has no meaning:
+		// if the label already exists in the original query result, remove it to preserve compatibility with relabeling, see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/10766.
+		// otherwise, ignore the label, see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9984.
 		if rr.Labels[k] == "" {
+			m.DelLabel(k)
 			continue
 		}
 		existingLabel := promrelabel.GetLabelByName(m.Labels, k)

app/vmalert/rule/recording_test.go

@@ -163,11 +163,13 @@ func TestRecordingRule_Exec(t *testing.T) {
 	f(&RecordingRule{
 		Name: "job:foo",
 		Labels: map[string]string{
-			"source": "test",
+			"source":      "test",
+			"empty_label": "", // this should be dropped
+			"pod":         "", // this should remove the pod label from query result
 		},
 	}, [][]datasource.Metric{{
-		metricWithValueAndLabels(t, 2, "__name__", "foo", "job", "foo"),
-		metricWithValueAndLabels(t, 1, "__name__", "bar", "job", "bar", "source", "origin"),
+		metricWithValueAndLabels(t, 2, "__name__", "foo", "job", "foo", "pod", "vmalert-0"),
+		metricWithValueAndLabels(t, 1, "__name__", "bar", "job", "bar", "source", "origin", "pod", "vmalert-1"),
 		metricWithValueAndLabels(t, 1, "__name__", "baz", "job", "baz", "source", "test"),
 	}}, [][]prompb.TimeSeries{{
 		newTimeSeries([]float64{2}, []int64{ts.UnixNano()}, []prompb.Label{

app/vmalert/rule/web.go

@@ -252,6 +252,9 @@ func (r *ApiRule) ExtendState() {
 
 // ToAPI returns ApiGroup representation of g
 func (g *Group) ToAPI() *ApiGroup {
+	if g == nil {
+		return &ApiGroup{}
+	}
 	g.mu.RLock()
 	defer g.mu.RUnlock()
 	ag := ApiGroup{

app/vmalert/templates/template.go

@@ -402,6 +402,20 @@ func templateFuncs() textTpl.FuncMap {
 			return t, nil
 		},
 
+		// formatTime formats the given Unix timestamp with the provided layout.
+		// For example: {{ now | formatTime "2006-01-02T15:04:05Z07:00" }}
+		"formatTime": func(layout string, i any) (string, error) {
+			v, err := toFloat64(i)
+			if err != nil {
+				return "", fmt.Errorf("formatTime: %w", err)
+			}
+			if math.IsNaN(v) || math.IsInf(v, 0) {
+				return "", fmt.Errorf("formatTime: cannot convert %v to time", v)
+			}
+			t := timeFromUnixTimestamp(v).Time().UTC()
+			return t.Format(layout), nil
+		},
+
 		/* URLs */
 
 		// externalURL returns value of `external.url` flag

app/vmalert/templates/template_test.go

@@ -6,6 +6,7 @@ import (
 	"strings"
 	"testing"
 	textTpl "text/template"
+	"time"
 )
 
 func TestTemplateFuncs_StringConversion(t *testing.T) {
@@ -103,6 +104,26 @@ func TestTemplateFuncs_Formatting(t *testing.T) {
 	f("humanizeTimestamp", 1679055557, "2023-03-17 12:19:17 +0000 UTC")
 }
 
+func TestTemplateFuncs_FormatTime(t *testing.T) {
+	funcs := templateFuncs()
+	formatTime := funcs["formatTime"].(func(layout string, i any) (string, error))
+
+	f := func(layout string, input any, expected string) {
+		t.Helper()
+		result, err := formatTime(layout, input)
+		if err != nil {
+			t.Fatalf("unexpected error for formatTime(%q, %v): %s", layout, input, err)
+		}
+		if result != expected {
+			t.Fatalf("unexpected result for formatTime(%q, %v); got\n%s\nwant\n%s", layout, input, result, expected)
+		}
+	}
+
+	f(time.RFC3339, float64(1679055557), "2023-03-17T12:19:17Z")
+	f("2006-01-02T15:04:05", int64(1679055557), "2023-03-17T12:19:17")
+	f(time.RFC822, int(1679055557), "17 Mar 23 12:19 UTC")
+}
+
 func mkTemplate(current, replacement any) textTemplate {
 	tmpl := textTemplate{}
 	if current != nil {

app/vmalert/vmalertutil/auth.go

@@ -20,11 +20,12 @@ func AuthConfig(filterOptions ...AuthConfigOptions) (*promauth.Config, error) {
 }
 
 // WithBasicAuth returns AuthConfigOptions and initialized promauth.BasicAuthConfig based on given params
-func WithBasicAuth(username, password, passwordFile string) AuthConfigOptions {
+func WithBasicAuth(username, usernameFile, password, passwordFile string) AuthConfigOptions {
 	return func(config *promauth.HTTPClientConfig) {
-		if username != "" || password != "" || passwordFile != "" {
+		if username != "" || usernameFile != "" || password != "" || passwordFile != "" {
 			config.BasicAuth = &promauth.BasicAuthConfig{
 				Username:     username,
+				UsernameFile: usernameFile,
 				Password:     promauth.NewSecret(password),
 				PasswordFile: passwordFile,
 			}

app/vmauth/auth_config.go

@@ -362,40 +362,62 @@ func (up *URLPrefix) setLoadBalancingPolicy(loadBalancingPolicy string) error {
 }
 
 type backendURLs struct {
-	healthChecksContext context.Context
-	healthChecksCancel  func()
-	healthChecksWG      sync.WaitGroup
-
+	bhc backendHealthCheck
 	bus []*backendURL
 }
 
+type backendHealthCheck struct {
+	ctx context.Context
+	// mu protects fields below
+	cancel    func()
+	mu        sync.Mutex
+	isStopped bool
+	wg        sync.WaitGroup
+}
+
+func (bhc *backendHealthCheck) run(hc func()) {
+	bhc.mu.Lock()
+	defer bhc.mu.Unlock()
+	if bhc.isStopped {
+		return
+	}
+	bhc.wg.Go(hc)
+}
+
+func (bhc *backendHealthCheck) stop() {
+	bhc.mu.Lock()
+	bhc.cancel()
+	bhc.isStopped = true
+	bhc.mu.Unlock()
+	bhc.wg.Wait()
+}
+
 func newBackendURLs() *backendURLs {
 	ctx, cancel := context.WithCancel(context.Background())
 	return &backendURLs{
-		healthChecksContext: ctx,
-		healthChecksCancel:  cancel,
+		bhc: backendHealthCheck{
+			ctx:    ctx,
+			cancel: cancel,
+		},
 	}
 }
 
 func (bus *backendURLs) add(u *url.URL) {
 	bus.bus = append(bus.bus, &backendURL{
-		url:                u,
-		healthCheckContext: bus.healthChecksContext,
-		healthCheckWG:      &bus.healthChecksWG,
-		hasPlaceHolders:    hasAnyPlaceholders(u),
+		url:             u,
+		bhc:             &bus.bhc,
+		hasPlaceHolders: hasAnyPlaceholders(u),
 	})
 }
 
 func (bus *backendURLs) stopHealthChecks() {
-	bus.healthChecksCancel()
-	bus.healthChecksWG.Wait()
+	bus.bhc.stop()
 }
 
 type backendURL struct {
 	broken atomic.Bool
 
-	healthCheckContext context.Context
-	healthCheckWG      *sync.WaitGroup
+	bhc *backendHealthCheck
 
 	concurrentRequests atomic.Int32
 
@@ -410,7 +432,7 @@ func (bu *backendURL) isBroken() bool {
 
 func (bu *backendURL) setBroken() {
 	if bu.broken.CompareAndSwap(false, true) {
-		bu.healthCheckWG.Go(func() {
+		bu.bhc.run(func() {
 			bu.runHealthCheck()
 			bu.broken.Store(false)
 		})
@@ -432,11 +454,11 @@ func (bu *backendURL) runHealthCheck() {
 		case <-t.C:
 			// Verify network connectivity via TCP dial before marking backend healthy.
 			// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9997
-			ctx, cancel := context.WithTimeout(bu.healthCheckContext, time.Second)
+			ctx, cancel := context.WithTimeout(bu.bhc.ctx, time.Second)
 			c, err := netutil.Dialer.DialContext(ctx, "tcp", addr)
 			cancel()
 			if err != nil {
-				if errors.Is(bu.healthCheckContext.Err(), context.Canceled) {
+				if errors.Is(bu.bhc.ctx.Err(), context.Canceled) {
 					return
 				}
 				logger.Warnf("ignoring the backend at %s for %s because of dial error: %s", addr, *failTimeout, err)
@@ -445,7 +467,7 @@ func (bu *backendURL) runHealthCheck() {
 
 			_ = c.Close()
 			return
-		case <-bu.healthCheckContext.Done():
+		case <-bu.bhc.ctx.Done():
 			return
 		}
 	}
@@ -588,6 +610,7 @@ func areEqualBackendURLs(a, b []*backendURL) bool {
 }
 
 // getFirstAvailableBackendURL returns the first available backendURL, which isn't broken.
+// If all backendURLs are broken, then returns the first backendURL.
 //
 // backendURL.put() must be called on the returned backendURL after the request is complete.
 func getFirstAvailableBackendURL(bus []*backendURL) *backendURL {
@@ -606,21 +629,22 @@ func getFirstAvailableBackendURL(bus []*backendURL) *backendURL {
 			return bu
 		}
 	}
-	return nil
+
+	// All backend urls are unavailable, then returning a first one, it could help increase the success rate of the requests。
+	// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/10837#issuecomment-4307050980.
+	bu.get()
+	return bu
 }
 
 // getLeastLoadedBackendURL returns a non-broken backendURL with the lowest number of concurrent requests.
+// If all backendURLs are broken, then returns the first backendURL.
 //
 // backendURL.put() must be called on the returned backendURL after the request is complete.
 func getLeastLoadedBackendURL(bus []*backendURL, atomicCounter *atomic.Uint32) *backendURL {
+	firstBu := bus[0]
 	if len(bus) == 1 {
-		// Fast path - return the only backend url.
-		bu := bus[0]
-		if bu.isBroken() {
-			return nil
-		}
-		bu.get()
-		return bu
+		firstBu.get()
+		return firstBu
 	}
 
 	// Slow path - select other backend urls.
@@ -658,7 +682,10 @@ func getLeastLoadedBackendURL(bus []*backendURL, atomicCounter *atomic.Uint32) *
 	}
 	buMin := bus[buMinIdx]
 	if buMin.isBroken() {
-		return nil
+		// If all backendURLs are broken, then returns the first backendURL.
+		// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/10837#issuecomment-4307050980.
+		firstBu.get()
+		return firstBu
 	}
 	buMin.get()
 	atomicCounter.CompareAndSwap(n+1, buMinIdx+1)

app/vmauth/auth_config_test.go

@@ -1031,6 +1031,33 @@ func TestLogRequest(t *testing.T) {
 	f("foo", 404, 10*time.Millisecond, `access_log request_host="localhost:8080" request_uri="" status_code=404 remote_addr="" user_agent="" referer="" duration_ms=10 username="foo"`)
 }
 
+func TestGetFirstAvailableBackend(t *testing.T) {
+	f := func(broken []bool, expectedIdx int) {
+		t.Helper()
+		bus := make([]*backendURL, len(broken))
+		for i := range broken {
+			bus[i] = &backendURL{
+				url: &url.URL{Host: fmt.Sprintf("server-%d", i)},
+			}
+			bus[i].broken.Store(broken[i])
+		}
+		bu := getFirstAvailableBackendURL(bus)
+		if bu == nil {
+			t.Fatalf("unexpected nil backend")
+		}
+		if bu.url.Host != fmt.Sprintf("server-%d", expectedIdx) {
+			t.Fatalf("unexpected backend, expected server-%d, got %s", expectedIdx, bu.url.Host)
+		}
+	}
+
+	f([]bool{false, false, false}, 0)
+	f([]bool{true, true, false}, 2)
+	// all backend are broken, then return the first one.
+	f([]bool{true, true, true}, 0)
+	f([]bool{true}, 0)
+
+}
+
 func getRegexs(paths []string) []*Regex {
 	var sps []*Regex
 	for _, path := range paths {

app/vmauth/example_config.yml

@@ -130,6 +130,16 @@ users:
   - "http://vmselect1:8481/select/{{.MetricsTenant}}/prometheus"
   - "http://vmselect2:8481/select/{{.MetricsTenant}}/prometheus"
 
+# JWT-based routing using header-based tenant identification (VictoriaMetrics cluster)
+# The AccountID and ProjectID from JWT vm_access claims are injected as HTTP headers.
+- name: jwt-header-tenant
+  jwt:
+    skip_verify: true
+  headers:
+  - "AccountID: {{.MetricsAccountID}}"
+  - "ProjectID: {{.MetricsProjectID}}"
+  url_prefix: "http://vminsert:8480/insert/prometheus"
+
 # Requests without Authorization header are proxied according to `unauthorized_user` section.
 # Requests are proxied in round-robin fashion between `url_prefix` backends.
 # The deny_partial_response query arg is added to all the proxied requests.

app/vmauth/jwt.go

@@ -17,6 +17,8 @@ import (
 
 const (
 	metricsTenantPlaceholder       = `{{.MetricsTenant}}`
+	metricsAccountIDPlaceholder    = `{{.MetricsAccountID}}`
+	metricsProjectIDPlaceholder    = `{{.MetricsProjectID}}`
 	metricsExtraLabelsPlaceholder  = `{{.MetricsExtraLabels}}`
 	metricsExtraFiltersPlaceholder = `{{.MetricsExtraFilters}}`
 
@@ -30,6 +32,8 @@ const (
 
 var allPlaceholders = []string{
 	metricsTenantPlaceholder,
+	metricsAccountIDPlaceholder,
+	metricsProjectIDPlaceholder,
 	metricsExtraLabelsPlaceholder,
 	metricsExtraFiltersPlaceholder,
 	logsAccountIDPlaceholder,
@@ -40,6 +44,8 @@ var allPlaceholders = []string{
 
 var urlPathPlaceHolders = []string{
 	metricsTenantPlaceholder,
+	metricsAccountIDPlaceholder,
+	metricsProjectIDPlaceholder,
 	logsAccountIDPlaceholder,
 	logsProjectIDPlaceholder,
 }
@@ -371,6 +377,8 @@ func jwtClaimsData(vma *jwt.VMAccessClaim) map[string][]string {
 	data := map[string][]string{
 		// TODO: optimize at parsing stage
 		metricsTenantPlaceholder:       {fmt.Sprintf("%d:%d", vma.MetricsAccountID, vma.MetricsProjectID)},
+		metricsAccountIDPlaceholder:    {fmt.Sprintf("%d", vma.MetricsAccountID)},
+		metricsProjectIDPlaceholder:    {fmt.Sprintf("%d", vma.MetricsProjectID)},
 		metricsExtraLabelsPlaceholder:  vma.MetricsExtraLabels,
 		metricsExtraFiltersPlaceholder: vma.MetricsExtraFilters,
 

app/vmauth/jwt_test.go

@@ -170,13 +170,13 @@ users:
   url_prefix: http://foo.bar
 `, "cannot parse public key from file \""+publicKeyFile+"\": failed to parse key \"invalidPEM\": failed to decode PEM block containing public key")
 
-	// unsupported placeholder in a header
+	// unsupported placeholder in a URL path
 	f(`
 users:
 - jwt: 
     skip_verify: true
   url_prefix: http://foo.bar/{{.UnsupportedPlaceholder}}/foo`,
-		"invalid placeholder found in URL request path: \"/{{.UnsupportedPlaceholder}}/foo\", supported values are: {{.MetricsTenant}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
+		"invalid placeholder found in URL request path: \"/{{.UnsupportedPlaceholder}}/foo\", supported values are: {{.MetricsTenant}}, {{.MetricsAccountID}}, {{.MetricsProjectID}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
 	)
 	// unsupported placeholder in a header
 	f(`
@@ -187,7 +187,7 @@ users:
     - "AccountID: {{.UnsupportedPlaceholder}}"
   url_prefix: http://foo.bar
 `,
-		"request header: \"AccountID\" has unsupported placeholder: \"{{.UnsupportedPlaceholder}}\", supported values are: {{.MetricsTenant}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
+		"request header: \"AccountID\" has unsupported placeholder: \"{{.UnsupportedPlaceholder}}\", supported values are: {{.MetricsTenant}}, {{.MetricsAccountID}}, {{.MetricsProjectID}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
 	)
 
 	// spaces in templating not allowed
@@ -199,7 +199,19 @@ users:
     - "AccountID: {{ .LogsAccountID }}"
   url_prefix: http://foo.bar
 `,
-		"request header: \"AccountID\" has unsupported placeholder: \"{{ .LogsAccountID }}\", supported values are: {{.MetricsTenant}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
+		"request header: \"AccountID\" has unsupported placeholder: \"{{ .LogsAccountID }}\", supported values are: {{.MetricsTenant}}, {{.MetricsAccountID}}, {{.MetricsProjectID}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
+	)
+
+	// placeholder must match the entire header value
+	f(`
+users:
+- jwt: 
+    skip_verify: true
+  headers:
+    - "AccountID: foo {{.MetricsAccountID}}"
+  url_prefix: http://foo.bar
+`,
+		"request header: \"AccountID\" has unsupported placeholder: \"foo {{.MetricsAccountID}}\", supported values are: {{.MetricsTenant}}, {{.MetricsAccountID}}, {{.MetricsProjectID}}, {{.MetricsExtraLabels}}, {{.MetricsExtraFilters}}, {{.LogsAccountID}}, {{.LogsProjectID}}, {{.LogsExtraFilters}}, {{.LogsExtraStreamFilters}}",
 	)
 
 	// oidc is not an object
@@ -364,10 +376,25 @@ users:
   url_prefix: http://foo.bar
 `, validRSAPublicKey, validECDSAPublicKey))
 
+	// metrics header placeholders
 	f(`
 users:
 - jwt:
     skip_verify: true
+  headers:
+  - "MetricsAccountID: {{.MetricsAccountID}}"
+  - "MetricsProjectID: {{.MetricsProjectID}}"
+  url_prefix: http://foo.bar
+`)
+
+	// logs header placeholders
+	f(`
+users:
+- jwt:
+    skip_verify: true
+  headers:
+  - "LogsAccountID: {{.LogsAccountID}}"
+  - "LogsProjectID: {{.LogsProjectID}}"
   url_prefix: http://foo.bar
 `)
 

app/vmauth/main.go

@@ -51,7 +51,7 @@ var (
 		"This allows reducing the consumption of backend resources when processing requests from clients connected via slow networks. "+
 		"Set to 0 to disable request buffering. See https://docs.victoriametrics.com/victoriametrics/vmauth/#request-body-buffering")
 	maxRequestBodySizeToRetry = flagutil.NewBytes("maxRequestBodySizeToRetry", 16*1024, "The maximum request body size to buffer in memory for potential retries at other backends. "+
-		"Request bodies larger than this size cannot be retried if the backend fails. Zero or negative value disables request body buffering and retries. "+
+		"Request bodies larger than this size cannot be retried if the backend fails. Zero or negative value disables retries. "+
 		"See also -requestBufferSize")
 
 	maxConcurrentRequests = flag.Int("maxConcurrentRequests", 1000, "The maximum number of concurrent requests vmauth can process simultaneously. "+
@@ -481,6 +481,9 @@ func tryProcessingRequest(w http.ResponseWriter, r *http.Request, targetURL *url
 	canRetry := !bbOK || bb.canRetry()
 
 	res, err := ui.rt.RoundTrip(req)
+	if err == nil {
+		defer func() { _ = res.Body.Close() }()
+	}
 
 	if errors.Is(r.Context().Err(), context.Canceled) {
 		// Do not retry canceled requests.
@@ -550,7 +553,6 @@ func tryProcessingRequest(w http.ResponseWriter, r *http.Request, targetURL *url
 	w.WriteHeader(res.StatusCode)
 
 	err = copyStreamToClient(w, res.Body)
-	_ = res.Body.Close()
 
 	if errors.Is(r.Context().Err(), context.Canceled) {
 		// Do not retry canceled requests.
@@ -848,14 +850,18 @@ func (bb *bufferedBody) Read(p []byte) (int, error) {
 }
 
 func (bb *bufferedBody) canRetry() bool {
-	return bb.r == nil
+	if bb.r != nil {
+		return false
+	}
+	maxRetrySize := maxRequestBodySizeToRetry.IntN()
+	return len(bb.buf) == 0 || (maxRetrySize > 0 && len(bb.buf) <= maxRetrySize)
 }
 
 // Close implements io.Closer interface.
 func (bb *bufferedBody) Close() error {
 	bb.resetReader()
+	bb.cannotRetry = !bb.canRetry()
 	if bb.r != nil {
-		bb.cannotRetry = true
 		return bb.r.Close()
 	}
 	return nil

app/vmauth/main_test.go

@@ -19,6 +19,7 @@ import (
 	"os"
 	"path/filepath"
 	"sort"
+	"strconv"
 	"strings"
 	"sync/atomic"
 	"testing"
@@ -850,6 +851,30 @@ users:
 		responseExpected,
 	)
 
+	// test header injection and URL templating with individual placeholders
+	request = httptest.NewRequest(`GET`, "http://some-host.com/api/v1/query", nil)
+	request.Header.Set(`Authorization`, `Bearer `+fullToken)
+	responseExpected = `
+statusCode=200
+path: /select/123/234/api/v1/query
+query:
+headers:
+    AccountID=123
+    ProjectID=234`
+	f(fmt.Sprintf(
+		`
+users:
+- jwt:
+    public_keys:
+    - %q
+  url_prefix: {BACKEND}/select/{{.MetricsAccountID}}/{{.MetricsProjectID}}
+  headers:
+  - "AccountID: {{.MetricsAccountID}}"
+  - "ProjectID: {{.MetricsProjectID}}"`, string(publicKeyPEM)),
+		request,
+		responseExpected,
+	)
+
 	// extra_label and extra_filters from vm_access claim merged with statically defined
 	request = httptest.NewRequest(`GET`, "http://some-host.com/api/v1/query", nil)
 	request.Header.Set(`Authorization`, `Bearer `+fullToken)
@@ -1831,7 +1856,7 @@ func (r *mockBody) Read(p []byte) (n int, err error) {
 }
 
 func TestBufferedBody_RetrySuccess(t *testing.T) {
-	f := func(s string, maxBodySize int) {
+	f := func(s string, maxSizeToRetry, bufferSize int) {
 		t.Helper()
 
 		defaultRequestBufferSize := requestBufferSize.String()
@@ -1840,7 +1865,7 @@ func TestBufferedBody_RetrySuccess(t *testing.T) {
 				t.Fatalf("cannot reset requestBufferSize: %s", err)
 			}
 		}()
-		if err := requestBufferSize.Set(fmt.Sprintf("%d", maxBodySize)); err != nil {
+		if err := requestBufferSize.Set(strconv.Itoa(bufferSize)); err != nil {
 			t.Fatalf("cannot set requestBufferSize: %s", err)
 		}
 
@@ -1850,7 +1875,7 @@ func TestBufferedBody_RetrySuccess(t *testing.T) {
 				t.Fatalf("cannot reset maxRequestBodySizeToRetry: %s", err)
 			}
 		}()
-		if err := maxRequestBodySizeToRetry.Set("0"); err != nil {
+		if err := maxRequestBodySizeToRetry.Set(strconv.Itoa(maxSizeToRetry)); err != nil {
 			t.Fatalf("cannot set maxRequestBodySizeToRetry: %s", err)
 		}
 
@@ -1879,16 +1904,20 @@ func TestBufferedBody_RetrySuccess(t *testing.T) {
 		}
 	}
 
-	f("", 0)
-	f("", -1)
-	f("", 100)
-	f("foo", 100)
-	f("foobar", 100)
-	f(newTestString(1000), 1001)
+	f("", 0, 2000)
+	f("", 0, 0)
+	f("", -1, 2000)
+	f("", 100, 2000)
+	f("foo", 100, 2000)
+	f("foobar", 100, 2000)
+	f("foobar", 100, 0)
+	f("foobar", 100, -1)
+	f(newTestString(1000), 1001, 2000)
+	f(newTestString(1000), 1001, 500)
 }
 
 func TestBufferedBody_RetrySuccessPartialRead(t *testing.T) {
-	f := func(s string, maxBodySize int) {
+	f := func(s string, maxSizeToRetry, bufferSize int) {
 		t.Helper()
 
 		// Check the case with partial read
@@ -1898,7 +1927,7 @@ func TestBufferedBody_RetrySuccessPartialRead(t *testing.T) {
 				t.Fatalf("cannot reset requestBufferSize: %s", err)
 			}
 		}()
-		if err := requestBufferSize.Set(fmt.Sprintf("%d", maxBodySize)); err != nil {
+		if err := requestBufferSize.Set(strconv.Itoa(bufferSize)); err != nil {
 			t.Fatalf("cannot set requestBufferSize: %s", err)
 		}
 
@@ -1908,7 +1937,7 @@ func TestBufferedBody_RetrySuccessPartialRead(t *testing.T) {
 				t.Fatalf("cannot reset maxRequestBodySizeToRetry: %s", err)
 			}
 		}()
-		if err := maxRequestBodySizeToRetry.Set("0"); err != nil {
+		if err := maxRequestBodySizeToRetry.Set(strconv.Itoa(maxSizeToRetry)); err != nil {
 			t.Fatalf("cannot set maxRequestBodySizeToRetry: %s", err)
 		}
 
@@ -1952,16 +1981,20 @@ func TestBufferedBody_RetrySuccessPartialRead(t *testing.T) {
 		}
 	}
 
-	f("", 0)
-	f("", -1)
-	f("", 100)
-	f("foo", 100)
-	f("foobar", 100)
-	f(newTestString(1000), 1001)
+	f("", 0, 2000)
+	f("", 0, 0)
+	f("", -1, 2000)
+	f("", 100, 2000)
+	f("foo", 100, 2000)
+	f("foobar", 100, 2000)
+	f("foobar", 100, 0)
+	f("foobar", 100, -1)
+	f(newTestString(1000), 1001, 2000)
+	f(newTestString(1000), 1001, 500)
 }
 
 func TestBufferedBody_RetryFailureTooBigBody(t *testing.T) {
-	f := func(s string, maxBodySize int) {
+	f := func(s string, maxSizeToRetry, bufferSize int) {
 		t.Helper()
 
 		defaultRequestBufferSize := requestBufferSize.String()
@@ -1970,7 +2003,7 @@ func TestBufferedBody_RetryFailureTooBigBody(t *testing.T) {
 				t.Fatalf("cannot reset requestBufferSize: %s", err)
 			}
 		}()
-		if err := requestBufferSize.Set("0"); err != nil {
+		if err := requestBufferSize.Set(strconv.Itoa(bufferSize)); err != nil {
 			t.Fatalf("cannot set requestBufferSize: %s", err)
 		}
 
@@ -1980,7 +2013,7 @@ func TestBufferedBody_RetryFailureTooBigBody(t *testing.T) {
 				t.Fatalf("cannot reset maxRequestBodySizeToRetry: %s", err)
 			}
 		}()
-		if err := maxRequestBodySizeToRetry.Set(fmt.Sprintf("%d", maxBodySize)); err != nil {
+		if err := maxRequestBodySizeToRetry.Set(strconv.Itoa(maxSizeToRetry)); err != nil {
 			t.Fatalf("cannot set maxRequestBodySizeToRetry: %s", err)
 		}
 
@@ -2025,12 +2058,17 @@ func TestBufferedBody_RetryFailureTooBigBody(t *testing.T) {
 	}
 
 	const maxBodySize = 1000
-	f(newTestString(maxBodySize+1), maxBodySize)
-	f(newTestString(2*maxBodySize), maxBodySize)
+	f(newTestString(maxBodySize+1), 0, 2*maxBodySize)
+	f(newTestString(maxBodySize+1), -1, 2*maxBodySize)
+	f(newTestString(maxBodySize+1), maxBodySize, 0)
+	f(newTestString(maxBodySize+1), maxBodySize, -1)
+	f(newTestString(maxBodySize+1), maxBodySize, maxBodySize)
+	f(newTestString(maxBodySize+1), maxBodySize, 2*maxBodySize)
+	f(newTestString(2*maxBodySize), maxBodySize, 0)
 }
 
-func TestBufferedBody_RetryFailureZeroOrNegativeMaxBodySize(t *testing.T) {
-	f := func(s string, maxBodySize int) {
+func TestBufferedBody_RetryDisabledByMaxRequestBodySizeToRetry(t *testing.T) {
+	f := func(s string, maxSizeToRetry, bufferSize int) {
 		t.Helper()
 
 		defaultRequestBufferSize := requestBufferSize.String()
@@ -2039,10 +2077,20 @@ func TestBufferedBody_RetryFailureZeroOrNegativeMaxBodySize(t *testing.T) {
 				t.Fatalf("cannot reset requestBufferSize: %s", err)
 			}
 		}()
-		if err := requestBufferSize.Set(fmt.Sprintf("%d", maxBodySize)); err != nil {
+		if err := requestBufferSize.Set(strconv.Itoa(bufferSize)); err != nil {
 			t.Fatalf("cannot set requestBufferSize: %s", err)
 		}
 
+		defaultMaxRequestBodySizeToRetry := maxRequestBodySizeToRetry.String()
+		defer func() {
+			if err := maxRequestBodySizeToRetry.Set(defaultMaxRequestBodySizeToRetry); err != nil {
+				t.Fatalf("cannot reset maxRequestBodySizeToRetry: %s", err)
+			}
+		}()
+		if err := maxRequestBodySizeToRetry.Set(strconv.Itoa(maxSizeToRetry)); err != nil {
+			t.Fatalf("cannot set maxRequestBodySizeToRetry: %s", err)
+		}
+
 		ctx := context.Background()
 		rb, err := bufferRequestBody(ctx, io.NopCloser(bytes.NewBufferString(s)), "foo")
 		if err != nil {
@@ -2051,8 +2099,8 @@ func TestBufferedBody_RetryFailureZeroOrNegativeMaxBodySize(t *testing.T) {
 		bb, ok := rb.(*bufferedBody)
 		canRetry := !ok || bb.canRetry()
 
-		if !canRetry {
-			t.Fatalf("canRetry() must return true before reading anything")
+		if canRetry {
+			t.Fatalf("canRetry() must return false before reading anything")
 		}
 		data, err := io.ReadAll(rb)
 		if err != nil {
@@ -2066,19 +2114,19 @@ func TestBufferedBody_RetryFailureZeroOrNegativeMaxBodySize(t *testing.T) {
 		}
 
 		data, err = io.ReadAll(rb)
-		if err != nil {
-			t.Fatalf("unexpected error in io.ReadAll: %s", err)
+		if err == nil {
+			t.Fatalf("expecting non-nil error")
 		}
-		if string(data) != s {
-			t.Fatalf("unexpected data read\ngot\n%s\nwant\n%s", data, s)
+		if len(data) != 0 {
+			t.Fatalf("unexpected non-empty data read: %q", data)
 		}
 	}
 
-	f("foobar", 0)
-	f(newTestString(1000), 0)
+	f("foobar", 0, 2048)
+	f(newTestString(1000), 0, 2048)
 
-	f("foobar", -1)
-	f(newTestString(1000), -1)
+	f("foobar", -1, 2048)
+	f(newTestString(1000), -1, 2048)
 }
 
 func newTestString(sLen int) string {

app/vmctl/opentsdb.go

@@ -8,10 +8,10 @@ import (
 	"time"
 
 	vmetrics "github.com/VictoriaMetrics/metrics"
+	"github.com/cheggaaa/pb/v3"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmctl/opentsdb"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmctl/vm"
-	"github.com/cheggaaa/pb/v3"
 )
 
 type otsdbProcessor struct {
@@ -89,9 +89,6 @@ func (op *otsdbProcessor) run(ctx context.Context) error {
 		// we're going to make serieslist * queryRanges queries, so we should represent that in the progress bar
 		otsdbSeriesTotal.Add(len(serieslist) * queryRanges)
 		bar := pb.StartNew(len(serieslist) * queryRanges)
-		defer func(bar *pb.ProgressBar) {
-			bar.Finish()
-		}(bar)
 		var wg sync.WaitGroup
 		for range op.otsdbcc {
 			wg.Go(func() {
@@ -106,41 +103,22 @@ func (op *otsdbProcessor) run(ctx context.Context) error {
 				}
 			})
 		}
-		/*
-			Loop through all series for this metric, processing all retentions and time ranges
-			requested. This loop is our primary "collect data from OpenTSDB loop" and should
-			be async, sending data to VictoriaMetrics over time.
+		runErr := op.sendQueries(ctx, serieslist, seriesCh, errCh, startTime)
 
-			The idea with having the select at the inner-most loop is to ensure quick
-			short-circuiting on error.
-		*/
-		for _, series := range serieslist {
-			for _, rt := range op.oc.Retentions {
-				for _, tr := range rt.QueryRanges {
-					select {
-					case otsdbErr := <-errCh:
-						return fmt.Errorf("opentsdb error: %s", otsdbErr)
-					case vmErr := <-op.im.Errors():
-						otsdbErrorsTotal.Inc()
-						return fmt.Errorf("import process failed: %s", wrapErr(vmErr, op.isVerbose))
-					case seriesCh <- queryObj{
-						Tr: tr, StartTime: startTime,
-						Series: series, Rt: opentsdb.RetentionMeta{
-							FirstOrder: rt.FirstOrder, SecondOrder: rt.SecondOrder, AggTime: rt.AggTime}}:
-					}
-				}
-			}
-		}
-
-		// Drain channels per metric
+		// Always drain channels and wait for workers to prevent goroutine leaks
 		close(seriesCh)
 		wg.Wait()
 		close(errCh)
 		// check for any lingering errors on the query side
 		for otsdbErr := range errCh {
-			return fmt.Errorf("import process failed: \n%s", otsdbErr)
+			if runErr == nil {
+				runErr = fmt.Errorf("import process failed: \n%s", otsdbErr)
+			}
 		}
 		bar.Finish()
+		if runErr != nil {
+			return runErr
+		}
 		log.Print(op.im.Stats())
 	}
 	op.im.Close()
@@ -155,6 +133,34 @@ func (op *otsdbProcessor) run(ctx context.Context) error {
 	return nil
 }
 
+// sendQueries iterates over all series and retention ranges, sending queries to workers.
+// It returns early if ctx is canceled or an error is received.
+func (op *otsdbProcessor) sendQueries(ctx context.Context, serieslist []opentsdb.Meta, seriesCh chan<- queryObj, errCh <-chan error, startTime int64) error {
+	for _, series := range serieslist {
+		for _, rt := range op.oc.Retentions {
+			for _, tr := range rt.QueryRanges {
+				select {
+				case <-ctx.Done():
+					return fmt.Errorf("context canceled: %s", ctx.Err())
+				case otsdbErr := <-errCh:
+					otsdbErrorsTotal.Inc()
+					return fmt.Errorf("opentsdb error: %s", otsdbErr)
+				case vmErr := <-op.im.Errors():
+					return fmt.Errorf("import process failed: %s", wrapErr(vmErr, op.isVerbose))
+				case seriesCh <- queryObj{
+					Tr: tr, StartTime: startTime,
+					Series: series, Rt: opentsdb.RetentionMeta{
+						FirstOrder:  rt.FirstOrder,
+						SecondOrder: rt.SecondOrder,
+						AggTime:     rt.AggTime,
+					}}:
+				}
+			}
+		}
+	}
+	return nil
+}
+
 func (op *otsdbProcessor) do(s queryObj) error {
 	start := s.StartTime - s.Tr.Start
 	end := s.StartTime - s.Tr.End
@@ -163,6 +169,7 @@ func (op *otsdbProcessor) do(s queryObj) error {
 		return fmt.Errorf("failed to collect data for %v in %v:%v :: %v", s.Series, s.Rt, s.Tr, err)
 	}
 	if len(data.Timestamps) < 1 || len(data.Values) < 1 {
+		log.Printf("no data found for %v in %v:%v...skipping", s.Series, s.Rt, s.Tr)
 		return nil
 	}
 	labels := make([]vm.LabelPair, 0, len(data.Tags))

app/vmctl/opentsdb/opentsdb.go

@@ -108,10 +108,10 @@ func (c Client) FindMetrics(q string) ([]string, error) {
 	if err != nil {
 		return nil, fmt.Errorf("failed to send GET request to %q: %s", q, err)
 	}
+	defer func() { _ = resp.Body.Close() }()
 	if resp.StatusCode != 200 {
 		return nil, fmt.Errorf("bad return from OpenTSDB: %d: %v", resp.StatusCode, resp)
 	}
-	defer func() { _ = resp.Body.Close() }()
 	body, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return nil, fmt.Errorf("could not retrieve metric data from %q: %s", q, err)
@@ -130,12 +130,12 @@ func (c Client) FindSeries(metric string) ([]Meta, error) {
 	q := fmt.Sprintf("%s/api/search/lookup?m=%s&limit=%d", c.Addr, metric, c.Limit)
 	resp, err := c.c.Get(q)
 	if err != nil {
-		return nil, fmt.Errorf("failed to set GET request to %q: %s", q, err)
+		return nil, fmt.Errorf("failed to send GET request to %q: %s", q, err)
 	}
+	defer func() { _ = resp.Body.Close() }()
 	if resp.StatusCode != 200 {
 		return nil, fmt.Errorf("bad return from OpenTSDB: %d: %v", resp.StatusCode, resp)
 	}
-	defer func() { _ = resp.Body.Close() }()
 	body, err := io.ReadAll(resp.Body)
 	if err != nil {
 		return nil, fmt.Errorf("could not retrieve series data from %q: %s", q, err)
@@ -185,6 +185,7 @@ func (c Client) GetData(series Meta, rt RetentionMeta, start int64, end int64, m
 	if err != nil {
 		return Metric{}, fmt.Errorf("failed to send GET request to %q: %s", q, err)
 	}
+	defer func() { _ = resp.Body.Close() }()
 	/*
 		There are three potential failures here, none of which should kill the entire
 		migration run:
@@ -196,7 +197,6 @@ func (c Client) GetData(series Meta, rt RetentionMeta, start int64, end int64, m
 		log.Printf("bad response code from OpenTSDB query %v for %q...skipping", resp.StatusCode, q)
 		return Metric{}, nil
 	}
-	defer func() { _ = resp.Body.Close() }()
 	body, err := io.ReadAll(resp.Body)
 	if err != nil {
 		log.Println("couldn't read response body from OpenTSDB query...skipping")
@@ -239,27 +239,20 @@ func (c Client) GetData(series Meta, rt RetentionMeta, start int64, end int64, m
 		In all "bad" cases, we don't end the migration, we just don't process that particular message
 	*/
 	if len(output) < 1 {
-		// no results returned...return an empty object without error
 		return Metric{}, nil
 	}
 	if len(output) > 1 {
-		// multiple series returned for a single query. We can't process this right, so...
-		return Metric{}, nil
+		return Metric{}, fmt.Errorf("unexpected number of series returned: %d for query %q; expected 1", len(output), q)
 	}
 	if len(output[0].AggregateTags) > 0 {
-		// This failure means we've suppressed potential series somehow...
-		return Metric{}, nil
+		return Metric{}, fmt.Errorf("aggregate tags %v present in response for query %q; series may be suppressed", output[0].AggregateTags, q)
 	}
 	data := Metric{}
 	data.Metric = output[0].Metric
 	data.Tags = output[0].Tags
-	/*
-		We evaluate data for correctness before formatting the actual values
-		to skip a little bit of time if the series has invalid formatting
-	*/
 	data, err = modifyData(data, c.Normalize)
 	if err != nil {
-		return Metric{}, nil
+		return Metric{}, fmt.Errorf("failed to convert metric data for query %q: %w", q, err)
 	}
 
 	/*

app/vmctl/opentsdb/parser.go

@@ -32,7 +32,7 @@ func convertDuration(duration string) (time.Duration, error) {
 	var err error
 	var timeValue int
 	if strings.HasSuffix(duration, "y") {
-		timeValue, err = strconv.Atoi(strings.Trim(duration, "y"))
+		timeValue, err = strconv.Atoi(strings.TrimSuffix(duration, "y"))
 		if err != nil {
 			return 0, fmt.Errorf("invalid time range: %q", duration)
 		}
@@ -42,7 +42,7 @@ func convertDuration(duration string) (time.Duration, error) {
 			return 0, fmt.Errorf("invalid time range: %q", duration)
 		}
 	} else if strings.HasSuffix(duration, "w") {
-		timeValue, err = strconv.Atoi(strings.Trim(duration, "w"))
+		timeValue, err = strconv.Atoi(strings.TrimSuffix(duration, "w"))
 		if err != nil {
 			return 0, fmt.Errorf("invalid time range: %q", duration)
 		}
@@ -52,7 +52,7 @@ func convertDuration(duration string) (time.Duration, error) {
 			return 0, fmt.Errorf("invalid time range: %q", duration)
 		}
 	} else if strings.HasSuffix(duration, "d") {
-		timeValue, err = strconv.Atoi(strings.Trim(duration, "d"))
+		timeValue, err = strconv.Atoi(strings.TrimSuffix(duration, "d"))
 		if err != nil {
 			return 0, fmt.Errorf("invalid time range: %q", duration)
 		}
@@ -95,6 +95,9 @@ func convertRetention(retention string, offset int64, msecTime bool) (Retention,
 	if !msecTime {
 		queryLength = queryLength / 1000
 	}
+	if queryLength <= 0 {
+		return Retention{}, fmt.Errorf("ttl %q resolves to non-positive query range %d; use a larger duration", chunks[2], queryLength)
+	}
 	queryRange := queryLength
 	// bump by the offset so we don't look at empty ranges any time offset > ttl
 	queryLength += offset
@@ -138,16 +141,29 @@ func convertRetention(retention string, offset int64, msecTime bool) (Retention,
 			2. we discover the actual size of each "chunk"
 			   This is second division step
 		*/
-		querySize = int64(queryRange / (queryRange / (rowLength * 4)))
+		divisor := queryRange / (rowLength * 4)
+		if divisor == 0 {
+			querySize = queryRange
+		} else {
+			querySize = queryRange / divisor
+		}
 	} else {
 		/*
 			Unless the aggTime (how long a range of data we're requesting per individual point)
 			is greater than the row size. Then we'll need to use that to determine
 			how big each individual query should be
 		*/
-		querySize = int64(queryRange / (queryRange / (aggTime * 4)))
+		divisor := queryRange / (aggTime * 4)
+		if divisor == 0 {
+			querySize = queryRange
+		} else {
+			querySize = queryRange / divisor
+		}
 	}
 
+	if querySize <= 0 {
+		return Retention{}, fmt.Errorf("computed non-positive querySize=%d for retention %q; check parameters", querySize, retention)
+	}
 	var timeChunks []TimeRange
 	var i int64
 	for i = offset; i <= queryLength; i = i + querySize {

app/vmctl/vmctlutil/time_test.go

@@ -20,6 +20,9 @@ func TestGetTime_Failure(t *testing.T) {
 
 	// negative time
 	f("-292273086-05-16T16:47:06Z")
+
+	// relative duration that resolves to a timestamp before 1970
+	f("-9223372036.855")
 }
 
 func TestGetTime_Success(t *testing.T) {
@@ -77,9 +80,6 @@ func TestGetTime_Success(t *testing.T) {
 	// float timestamp representation",
 	f("1562529662.324", time.Date(2019, 7, 7, 20, 01, 02, 324e6, time.UTC))
 
-	// negative timestamp
-	f("-9223372036.855", time.Date(1970, 01, 01, 00, 00, 00, 00, time.UTC))
-
 	// big timestamp
 	f("1223372036855", time.Date(2008, 10, 7, 9, 33, 56, 855e6, time.UTC))
 

app/vminsert/common/insert_ctx.go

@@ -184,7 +184,7 @@ func (ctx *InsertCtx) WriteMetadata(mmpbs []prompb.MetricMetadata) error {
 	}
 	ctx.mms = mms
 
-	err := vmstorage.AddMetadataRows(mms)
+	err := vmstorage.VMInsertAPI.WriteMetadata(mms)
 	if err != nil {
 		return &httpserver.ErrorWithStatusCode{
 			Err:        fmt.Errorf("cannot store metrics metadata: %w", err),
@@ -209,7 +209,7 @@ func (ctx *InsertCtx) WritePromMetadata(mmps []prometheus.Metadata) error {
 	}
 	ctx.mms = mms
 
-	err := vmstorage.AddMetadataRows(mms)
+	err := vmstorage.VMInsertAPI.WriteMetadata(mms)
 	if err != nil {
 		return &httpserver.ErrorWithStatusCode{
 			Err:        fmt.Errorf("cannot store prometheus metrics metadata: %w", err),
@@ -278,7 +278,7 @@ func (ctx *InsertCtx) FlushBufs() error {
 	// since the number of concurrent FlushBufs() calls should be already limited via writeconcurrencylimiter
 	// used at every stream.Parse() call under lib/protoparser/*
 
-	err := vmstorage.AddRows(ctx.mrs)
+	err := vmstorage.VMInsertAPI.WriteRows(ctx.mrs)
 	ctx.Reset(0)
 	if err == nil {
 		return nil

app/vminsert/common/streamaggr.go

@@ -283,7 +283,7 @@ func pushAggregateSeries(tss []prompb.TimeSeries) {
 	}
 	// There is no need in limiting the number of concurrent calls to vmstorage.AddRows() here,
 	// since the number of concurrent pushAggregateSeries() calls should be already limited by lib/streamaggr.
-	if err := vmstorage.AddRows(ctx.mrs); err != nil {
+	if err := vmstorage.VMInsertAPI.WriteRows(ctx.mrs); err != nil {
 		logger.Errorf("cannot flush aggregate series: %s", err)
 	}
 }

app/vminsert/main.go

@@ -89,6 +89,7 @@ var staticServer = http.FileServer(http.FS(staticFiles))
 func Init() {
 	relabel.Init()
 	common.InitStreamAggr()
+	opentelemetry.Init()
 	protoparserutil.StartUnmarshalWorkers()
 	if len(*graphiteListenAddr) > 0 {
 		graphiteServer = graphiteserver.MustStart(*graphiteListenAddr, *graphiteUseProxyProtocol, graphite.InsertHandler)

app/vminsert/opentelemetry/request_handler.go

@@ -20,6 +20,11 @@ var (
 	metadataInserted = metrics.NewCounter(`vm_metadata_rows_inserted_total{type="opentelemetry"}`)
 )
 
+// Init must be called after flag.Parse and before using the opentelemetry package.
+func Init() {
+	stream.InitDecodeOptions()
+}
+
 // InsertHandler processes opentelemetry metrics.
 func InsertHandler(req *http.Request) error {
 	extraLabels, err := protoparserutil.GetExtraLabels(req)

app/vmselect/graphite/metrics_api.go

@@ -1,7 +1,6 @@
 package graphite
 
 import (
-	"flag"
 	"fmt"
 	"math"
 	"net/http"
@@ -21,8 +20,6 @@ import (
 	"github.com/VictoriaMetrics/metricsql"
 )
 
-var maxTagValueSuffixes = flag.Int("search.maxTagValueSuffixesPerSearch", 100e3, "The maximum number of tag value suffixes returned from /metrics/find")
-
 // MetricsFindHandler implements /metrics/find handler.
 //
 // See https://graphite-api.readthedocs.io/en/latest/api.html#metrics-find
@@ -225,7 +222,7 @@ func metricsFind(tr storage.TimeRange, label, qHead, qTail string, delimiter byt
 	n := strings.IndexAny(qTail, "*{[")
 	if n < 0 {
 		query := qHead + qTail
-		suffixes, err := netstorage.TagValueSuffixes(nil, tr, label, query, delimiter, *maxTagValueSuffixes, deadline)
+		suffixes, err := netstorage.TagValueSuffixes(nil, tr, label, query, delimiter, 0, deadline)
 		if err != nil {
 			return nil, err
 		}
@@ -245,7 +242,7 @@ func metricsFind(tr storage.TimeRange, label, qHead, qTail string, delimiter byt
 	}
 	if n == len(qTail)-1 && strings.HasSuffix(qTail, "*") {
 		query := qHead + qTail[:len(qTail)-1]
-		suffixes, err := netstorage.TagValueSuffixes(nil, tr, label, query, delimiter, *maxTagValueSuffixes, deadline)
+		suffixes, err := netstorage.TagValueSuffixes(nil, tr, label, query, delimiter, 0, deadline)
 		if err != nil {
 			return nil, err
 		}

app/vmselect/graphite/tags_api.go

@@ -138,7 +138,9 @@ func registerMetrics(startTime time.Time, w http.ResponseWriter, r *http.Request
 		mr.MetricNameRaw = storage.MarshalMetricNameRaw(mr.MetricNameRaw[:0], labels)
 		mr.Timestamp = ct
 	}
-	vmstorage.RegisterMetricNames(nil, mrs)
+	if err := vmstorage.VMSelectAPI.RegisterMetricNames(nil, mrs, 0); err != nil {
+		return err
+	}
 
 	// Return response
 	contentType := "text/plain; charset=utf-8"

app/vmselect/main.go

@@ -21,7 +21,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/stats"
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmstorage"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/buildinfo"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/cgroup"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
@@ -36,12 +35,6 @@ var (
 	deleteAuthKey                = flagutil.NewPassword("deleteAuthKey", "authKey for metrics' deletion via /api/v1/admin/tsdb/delete_series and /tags/delSeries. It could be passed via authKey query arg. It overrides -httpAuth.*")
 	metricNamesStatsResetAuthKey = flagutil.NewPassword("metricNamesStatsResetAuthKey", "authKey for resetting metric names usage cache via /api/v1/admin/status/metric_names_stats/reset. It overrides -httpAuth.*. "+
 		"See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#track-ingested-metrics-usage")
-
-	maxConcurrentRequests = flag.Int("search.maxConcurrentRequests", getDefaultMaxConcurrentRequests(), "The maximum number of concurrent search requests. "+
-		"It shouldn't be high, since a single request can saturate all the CPU cores, while many concurrently executed requests may require high amounts of memory. "+
-		"See also -search.maxQueueDuration and -search.maxMemoryPerQuery")
-	maxQueueDuration = flag.Duration("search.maxQueueDuration", 10*time.Second, "The maximum time the request waits for execution when -search.maxConcurrentRequests "+
-		"limit is reached; see also -search.maxQueryDuration")
 	resetCacheAuthKey    = flagutil.NewPassword("search.resetCacheAuthKey", "Optional authKey for resetting rollup cache via /internal/resetRollupResultCache call. It could be passed via authKey query arg. It overrides -httpAuth.*")
 	logSlowQueryDuration = flag.Duration("search.logSlowQueryDuration", 5*time.Second, "Log queries with execution time exceeding this value. Zero disables slow query logging. "+
 		"See also -search.logQueryMemoryUsage")
@@ -50,25 +43,22 @@ var (
 
 var slowQueries = metrics.NewCounter(`vm_slow_queries_total`)
 
-func getDefaultMaxConcurrentRequests() int {
-	// A single request can saturate all the CPU cores, so there is no sense
-	// in allowing higher number of concurrent requests - they will just contend
-	// for unavailable CPU time.
-	n := min(cgroup.AvailableCPUs()*2, 16)
-	return n
-}
-
 // Init initializes vmselect
-func Init() {
-	tmpDirPath := *vmstorage.DataPath + "/tmp"
+func Init(maxConcurrentRequests int, maxQueueDuration time.Duration) {
+	tmpDirPath := vmstorage.DataPath() + "/tmp"
 	fs.MustRemoveDirContents(tmpDirPath)
 	netstorage.InitTmpBlocksDir(tmpDirPath)
-	promql.InitRollupResultCache(*vmstorage.DataPath + "/cache/rollupResult")
-	prometheus.InitMaxUniqueTimeseries(*maxConcurrentRequests)
+	promql.InitRollupResultCache(vmstorage.DataPath() + "/cache/rollupResult")
 
-	concurrencyLimitCh = make(chan struct{}, *maxConcurrentRequests)
+	concurrencyLimitCh = make(chan struct{}, maxConcurrentRequests)
 	initVMUIConfig()
 	initVMAlertProxy()
+
+	flagutil.RegisterSecretFlag("vmalert.proxyURL")
+
+	RequestHandler = func(w http.ResponseWriter, r *http.Request) bool {
+		return requestHandler(w, r, maxConcurrentRequests, maxQueueDuration)
+	}
 }
 
 // Stop stops vmselect
@@ -88,9 +78,6 @@ var (
 	_ = metrics.NewGauge(`vm_concurrent_select_current`, func() float64 {
 		return float64(len(concurrencyLimitCh))
 	})
-	_ = metrics.NewGauge(`vm_search_max_unique_timeseries`, func() float64 {
-		return float64(prometheus.GetMaxUniqueTimeSeries())
-	})
 )
 
 //go:embed vmui
@@ -98,8 +85,10 @@ var vmuiFiles embed.FS
 
 var vmuiFileServer = http.FileServer(http.FS(vmuiFiles))
 
-// RequestHandler handles remote read API requests
-func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
+var RequestHandler func(w http.ResponseWriter, r *http.Request) bool
+
+// requestHandler handles remote read API requests
+func requestHandler(w http.ResponseWriter, r *http.Request, maxConcurrentRequests int, maxQueueDuration time.Duration) bool {
 	path := strings.ReplaceAll(r.URL.Path, "//", "/")
 
 	// Strip /prometheus and /graphite prefixes in order to provide path compatibility with cluster version
@@ -129,12 +118,12 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 	default:
 		// Sleep for a while until giving up. This should resolve short bursts in requests.
 		concurrencyLimitReached.Inc()
-		d := min(searchutil.GetMaxQueryDuration(r), *maxQueueDuration)
+		d := min(searchutil.GetMaxQueryDuration(r), maxQueueDuration)
 		t := timerpool.Get(d)
 		select {
 		case concurrencyLimitCh <- struct{}{}:
 			timerpool.Put(t)
-			qt.Printf("wait in queue because -search.maxConcurrentRequests=%d concurrent requests are executed", *maxConcurrentRequests)
+			qt.Printf("wait in queue because -search.maxConcurrentRequests=%d concurrent requests are executed", maxConcurrentRequests)
 			defer func() { <-concurrencyLimitCh }()
 		case <-r.Context().Done():
 			timerpool.Put(t)
@@ -150,7 +139,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 				Err: fmt.Errorf("couldn't start executing the request in %.3f seconds, since -search.maxConcurrentRequests=%d concurrent requests "+
 					"are executed. Possible solutions: to reduce query load; to add more compute resources to the server; "+
 					"to increase -search.maxQueueDuration=%s; to increase -search.maxQueryDuration; to increase -search.maxConcurrentRequests",
-					d.Seconds(), *maxConcurrentRequests, maxQueueDuration),
+					d.Seconds(), maxConcurrentRequests, maxQueueDuration),
 				StatusCode: http.StatusTooManyRequests,
 			}
 			w.Header().Add("Retry-After", "10")
@@ -262,6 +251,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 		return true
 	case "/api/v1/export":
 		exportRequests.Inc()
+		httpserver.EnableCORS(w, r)
 		if err := prometheus.ExportHandler(startTime, w, r); err != nil {
 			exportErrors.Inc()
 			httpserver.Errorf(w, r, "%s", err)
@@ -270,6 +260,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 		return true
 	case "/api/v1/export/csv":
 		exportCSVRequests.Inc()
+		httpserver.EnableCORS(w, r)
 		if err := prometheus.ExportCSVHandler(startTime, w, r); err != nil {
 			exportCSVErrors.Inc()
 			httpserver.Errorf(w, r, "%s", err)
@@ -278,6 +269,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 		return true
 	case "/api/v1/export/native":
 		exportNativeRequests.Inc()
+		httpserver.EnableCORS(w, r)
 		if err := prometheus.ExportNativeHandler(startTime, w, r); err != nil {
 			exportNativeErrors.Inc()
 			httpserver.Errorf(w, r, "%s", err)

app/vmselect/netstorage/netstorage.go

@@ -27,10 +27,6 @@ import (
 )
 
 var (
-	maxTagKeysPerSearch = flag.Int("search.maxTagKeys", 100e3, "The maximum number of tag keys returned from /api/v1/labels . "+
-		"See also -search.maxLabelsAPISeries and -search.maxLabelsAPIDuration")
-	maxTagValuesPerSearch = flag.Int("search.maxTagValues", 100e3, "The maximum number of tag values returned from /api/v1/label/<label_name>/values . "+
-		"See also -search.maxLabelsAPISeries and -search.maxLabelsAPIDuration")
 	maxSamplesPerSeries = flag.Int("search.maxSamplesPerSeries", 30e6, "The maximum number of raw samples a single query can scan per each time series. This option allows limiting memory usage")
 	maxSamplesPerQuery  = flag.Int("search.maxSamplesPerQuery", 1e9, "The maximum number of raw samples a single query can process across all time series. "+
 		"This protects from heavy queries, which select unexpectedly high number of raw samples. See also -search.maxSamplesPerSeries")
@@ -80,7 +76,7 @@ func (rss *Results) Cancel() {
 }
 
 func (rss *Results) mustClose() {
-	putStorageSearch(rss.sr)
+	vmstorage.PutSearch(rss.sr)
 	rss.sr = nil
 	putTmpBlocksFile(rss.tbf)
 	rss.tbf = nil
@@ -758,12 +754,7 @@ var sbhPool sync.Pool
 func DeleteSeries(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline searchutil.Deadline) (int, error) {
 	qt = qt.NewChild("delete series: %s", sq)
 	defer qt.Done()
-	tr := sq.GetTimeRange()
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
-	if err != nil {
-		return 0, err
-	}
-	return vmstorage.DeleteSeries(qt, tfss, sq.MaxMetrics)
+	return vmstorage.VMSelectAPI.DeleteSeries(qt, sq, deadline.Deadline())
 }
 
 // LabelNames returns label names matching the given sq until the given deadline.
@@ -773,15 +764,7 @@ func LabelNames(qt *querytracer.Tracer, sq *storage.SearchQuery, maxLabelNames i
 	if deadline.Exceeded() {
 		return nil, fmt.Errorf("timeout exceeded before starting the query processing: %s", deadline.String())
 	}
-	if maxLabelNames > *maxTagKeysPerSearch || maxLabelNames <= 0 {
-		maxLabelNames = *maxTagKeysPerSearch
-	}
-	tr := sq.GetTimeRange()
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
-	if err != nil {
-		return nil, err
-	}
-	labels, err := vmstorage.SearchLabelNames(qt, tfss, tr, maxLabelNames, sq.MaxMetrics, deadline.Deadline())
+	labels, err := vmstorage.VMSelectAPI.LabelNames(qt, sq, maxLabelNames, deadline.Deadline())
 	if err != nil {
 		return nil, fmt.Errorf("error during labels search on time range: %w", err)
 	}
@@ -841,15 +824,7 @@ func LabelValues(qt *querytracer.Tracer, labelName string, sq *storage.SearchQue
 	if deadline.Exceeded() {
 		return nil, fmt.Errorf("timeout exceeded before starting the query processing: %s", deadline.String())
 	}
-	if maxLabelValues > *maxTagValuesPerSearch || maxLabelValues <= 0 {
-		maxLabelValues = *maxTagValuesPerSearch
-	}
-	tr := sq.GetTimeRange()
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
-	if err != nil {
-		return nil, err
-	}
-	labelValues, err := vmstorage.SearchLabelValues(qt, labelName, tfss, tr, maxLabelValues, sq.MaxMetrics, deadline.Deadline())
+	labelValues, err := vmstorage.VMSelectAPI.LabelValues(qt, sq, labelName, maxLabelValues, deadline.Deadline())
 	if err != nil {
 		return nil, fmt.Errorf("error during label values search on time range for labelName=%q: %w", labelName, err)
 	}
@@ -864,7 +839,10 @@ func GetMetricsMetadata(qt *querytracer.Tracer, limit int, metricName string) ([
 	qt = qt.NewChild("get metrics metadata: limit=%d, metric_name=%q", limit, metricName)
 	defer qt.Done()
 
-	metadata := vmstorage.Storage.GetMetadataRows(qt, limit, metricName)
+	metadata, err := vmstorage.VMSelectAPI.GetMetadataRecords(qt, nil, limit, metricName, 0)
+	if err != nil {
+		return nil, err
+	}
 
 	sort.Slice(metadata, func(i, j int) bool {
 		return string(metadata[i].MetricFamilyName) < string(metadata[j].MetricFamilyName)
@@ -912,16 +890,11 @@ func TagValueSuffixes(qt *querytracer.Tracer, tr storage.TimeRange, tagKey, tagV
 	if deadline.Exceeded() {
 		return nil, fmt.Errorf("timeout exceeded before starting the query processing: %s", deadline.String())
 	}
-	suffixes, err := vmstorage.SearchTagValueSuffixes(qt, tr, tagKey, tagValuePrefix, delimiter, maxSuffixes, deadline.Deadline())
+	suffixes, err := vmstorage.VMSelectAPI.TagValueSuffixes(qt, 0, 0, tr, tagKey, tagValuePrefix, delimiter, maxSuffixes, deadline.Deadline())
 	if err != nil {
 		return nil, fmt.Errorf("error during search for suffixes for tagKey=%q, tagValuePrefix=%q, delimiter=%c on time range %s: %w",
 			tagKey, tagValuePrefix, delimiter, tr.String(), err)
 	}
-	if len(suffixes) >= maxSuffixes {
-		return nil, fmt.Errorf("more than -search.maxTagValueSuffixesPerSearch=%d tag value suffixes found for tagKey=%q, tagValuePrefix=%q, delimiter=%c on time range %s; "+
-			"either narrow down the query or increase -search.maxTagValueSuffixesPerSearch command-line flag value",
-			maxSuffixes, tagKey, tagValuePrefix, delimiter, tr.String())
-	}
 	return suffixes, nil
 }
 
@@ -934,13 +907,7 @@ func TSDBStatus(qt *querytracer.Tracer, sq *storage.SearchQuery, focusLabel stri
 	if deadline.Exceeded() {
 		return nil, fmt.Errorf("timeout exceeded before starting the query processing: %s", deadline.String())
 	}
-	tr := sq.GetTimeRange()
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
-	if err != nil {
-		return nil, err
-	}
-	date := uint64(tr.MinTimestamp) / (3600 * 24 * 1000)
-	status, err := vmstorage.GetTSDBStatus(qt, tfss, date, focusLabel, topN, sq.MaxMetrics, deadline.Deadline())
+	status, err := vmstorage.VMSelectAPI.TSDBStatus(qt, sq, focusLabel, topN, deadline.Deadline())
 	if err != nil {
 		return nil, fmt.Errorf("error during tsdb status request: %w", err)
 	}
@@ -954,28 +921,13 @@ func SeriesCount(qt *querytracer.Tracer, deadline searchutil.Deadline) (uint64,
 	if deadline.Exceeded() {
 		return 0, fmt.Errorf("timeout exceeded before starting the query processing: %s", deadline.String())
 	}
-	n, err := vmstorage.GetSeriesCount(deadline.Deadline())
+	n, err := vmstorage.VMSelectAPI.SeriesCount(qt, 0, 0, deadline.Deadline())
 	if err != nil {
 		return 0, fmt.Errorf("error during series count request: %w", err)
 	}
 	return n, nil
 }
 
-func getStorageSearch() *storage.Search {
-	v := ssPool.Get()
-	if v == nil {
-		return &storage.Search{}
-	}
-	return v.(*storage.Search)
-}
-
-func putStorageSearch(sr *storage.Search) {
-	sr.MustClose()
-	ssPool.Put(sr)
-}
-
-var ssPool sync.Pool
-
 // ExportBlocks searches for time series matching sq and calls f for each found block.
 //
 // f is called in parallel from multiple goroutines.
@@ -989,21 +941,12 @@ func ExportBlocks(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline sear
 	if deadline.Exceeded() {
 		return fmt.Errorf("timeout exceeded before starting data export: %s", deadline.String())
 	}
-	tr := sq.GetTimeRange()
-	if err := vmstorage.CheckTimeRange(tr); err != nil {
-		return err
-	}
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
+
+	sr, _, err := vmstorage.GetSearch(qt, sq, deadline.Deadline())
 	if err != nil {
 		return err
 	}
-
-	vmstorage.WG.Add(1)
-	defer vmstorage.WG.Done()
-
-	sr := getStorageSearch()
-	defer putStorageSearch(sr)
-	sr.Init(qt, vmstorage.Storage, tfss, tr, sq.MaxMetrics, deadline.Deadline())
+	defer vmstorage.PutSearch(sr)
 
 	// Start workers that call f in parallel on available CPU cores.
 	workCh := make(chan *exportWork, gomaxprocs*8)
@@ -1013,6 +956,7 @@ func ExportBlocks(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline sear
 		mustStop      atomic.Bool
 	)
 	var wg sync.WaitGroup
+	tr := sq.GetTimeRange()
 	for workerID := range gomaxprocs {
 		wg.Go(func() {
 			for xw := range workCh {
@@ -1096,17 +1040,7 @@ func SearchMetricNames(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline
 		return nil, fmt.Errorf("timeout exceeded before starting to search metric names: %s", deadline.String())
 	}
 
-	// Setup search.
-	tr := sq.GetTimeRange()
-	if err := vmstorage.CheckTimeRange(tr); err != nil {
-		return nil, err
-	}
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
-	if err != nil {
-		return nil, err
-	}
-
-	metricNames, err := vmstorage.SearchMetricNames(qt, tfss, tr, sq.MaxMetrics, deadline.Deadline())
+	metricNames, err := vmstorage.VMSelectAPI.SearchMetricNames(qt, sq, deadline.Deadline())
 	if err != nil {
 		return nil, fmt.Errorf("cannot find metric names: %w", err)
 	}
@@ -1125,21 +1059,11 @@ func ProcessSearchQuery(qt *querytracer.Tracer, sq *storage.SearchQuery, deadlin
 		return nil, fmt.Errorf("timeout exceeded before starting the query processing: %s", deadline.String())
 	}
 
-	// Setup search.
-	tr := sq.GetTimeRange()
-	if err := vmstorage.CheckTimeRange(tr); err != nil {
-		return nil, err
-	}
-	tfss, err := setupTfss(qt, tr, sq.TagFilterss, sq.MaxMetrics, deadline)
+	sr, maxSeriesCount, err := vmstorage.GetSearch(qt, sq, deadline.Deadline())
 	if err != nil {
 		return nil, err
 	}
 
-	vmstorage.WG.Add(1)
-	defer vmstorage.WG.Done()
-
-	sr := getStorageSearch()
-	maxSeriesCount := sr.Init(qt, vmstorage.Storage, tfss, tr, sq.MaxMetrics, deadline.Deadline())
 	type blockRefs struct {
 		brs []blockRef
 	}
@@ -1177,7 +1101,7 @@ func ProcessSearchQuery(qt *querytracer.Tracer, sq *storage.SearchQuery, deadlin
 		blocksRead++
 		if deadline.Exceeded() {
 			putTmpBlocksFile(tbf)
-			putStorageSearch(sr)
+			vmstorage.PutSearch(sr)
 			return nil, fmt.Errorf("timeout exceeded while fetching data block #%d from storage: %s", blocksRead, deadline.String())
 		}
 		br := sr.MetricBlockRef.BlockRef
@@ -1189,7 +1113,7 @@ func ProcessSearchQuery(qt *querytracer.Tracer, sq *storage.SearchQuery, deadlin
 		samples += br.RowsCount()
 		if *maxSamplesPerQuery > 0 && samples > *maxSamplesPerQuery {
 			putTmpBlocksFile(tbf)
-			putStorageSearch(sr)
+			vmstorage.PutSearch(sr)
 			return nil, fmt.Errorf("cannot select more than -search.maxSamplesPerQuery=%d samples; possible solutions: increase the -search.maxSamplesPerQuery; "+
 				"reduce time range for the query; use more specific label filters in order to select fewer series", *maxSamplesPerQuery)
 		}
@@ -1198,7 +1122,7 @@ func ProcessSearchQuery(qt *querytracer.Tracer, sq *storage.SearchQuery, deadlin
 		addr, err := tbf.WriteBlockRefData(buf)
 		if err != nil {
 			putTmpBlocksFile(tbf)
-			putStorageSearch(sr)
+			vmstorage.PutSearch(sr)
 			return nil, fmt.Errorf("cannot write %d bytes to temporary file: %w", len(buf), err)
 		}
 
@@ -1256,7 +1180,7 @@ func ProcessSearchQuery(qt *querytracer.Tracer, sq *storage.SearchQuery, deadlin
 
 	if err := sr.Error(); err != nil {
 		putTmpBlocksFile(tbf)
-		putStorageSearch(sr)
+		vmstorage.PutSearch(sr)
 		if errors.Is(err, storage.ErrDeadlineExceeded) {
 			return nil, fmt.Errorf("timeout exceeded during the query: %s", deadline.String())
 		}
@@ -1264,13 +1188,13 @@ func ProcessSearchQuery(qt *querytracer.Tracer, sq *storage.SearchQuery, deadlin
 	}
 	if err := tbf.Finalize(); err != nil {
 		putTmpBlocksFile(tbf)
-		putStorageSearch(sr)
+		vmstorage.PutSearch(sr)
 		return nil, fmt.Errorf("cannot finalize temporary file: %w", err)
 	}
 	qt.Printf("fetch unique series=%d, blocks=%d, samples=%d, bytes=%d", len(m), blocksRead, samples, tbf.Len())
 
 	var rss Results
-	rss.tr = tr
+	rss.tr = sq.GetTimeRange()
 	rss.deadline = deadline
 	pts := make([]packedTimeseries, len(orderedMetricNames))
 	for i, metricName := range orderedMetricNames {
@@ -1311,35 +1235,6 @@ func getBlockRefsEnd(a []blockRef) uintptr {
 	return uintptr(unsafe.Pointer(unsafe.SliceData(a))) + uintptr(len(a))*unsafe.Sizeof(blockRef{})
 }
 
-func setupTfss(qt *querytracer.Tracer, tr storage.TimeRange, tagFilterss [][]storage.TagFilter, maxMetrics int, deadline searchutil.Deadline) ([]*storage.TagFilters, error) {
-	tfss := make([]*storage.TagFilters, 0, len(tagFilterss))
-	for _, tagFilters := range tagFilterss {
-		tfs := storage.NewTagFilters()
-		for i := range tagFilters {
-			tf := &tagFilters[i]
-			if string(tf.Key) == "__graphite__" {
-				query := tf.Value
-				paths, err := vmstorage.SearchGraphitePaths(qt, tr, query, maxMetrics, deadline.Deadline())
-				if err != nil {
-					return nil, fmt.Errorf("error when searching for Graphite paths for query %q: %w", query, err)
-				}
-				if len(paths) >= maxMetrics {
-					return nil, fmt.Errorf("more than %d time series match Graphite query %q; "+
-						"either narrow down the query or increase the corresponding -search.max* command-line flag value; "+
-						"see https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#resource-usage-limits", maxMetrics, query)
-				}
-				tfs.AddGraphiteQuery(query, paths, tf.IsNegative)
-				continue
-			}
-			if err := tfs.Add(tf.Key, tf.Value, tf.IsNegative, tf.IsRegexp); err != nil {
-				return nil, fmt.Errorf("cannot parse tag filter %s: %w", tf, err)
-			}
-		}
-		tfss = append(tfss, tfs)
-	}
-	return tfss, nil
-}
-
 func applyGraphiteRegexpFilter(filter string, ss []string) ([]string, error) {
 	// Anchor filter regexp to the beginning of the string as Graphite does.
 	// See https://github.com/graphite-project/graphite-web/blob/3ad279df5cb90b211953e39161df416e54a84948/webapp/graphite/tags/localdatabase.py#L157
@@ -1366,13 +1261,12 @@ const maxFastAllocBlockSize = 32 * 1024
 func GetMetricNamesStats(qt *querytracer.Tracer, limit, le int, matchPattern string) (metricnamestats.StatsResult, error) {
 	qt = qt.NewChild("get metric names usage statistics with limit: %d, less or equal to: %d, match pattern=%q", limit, le, matchPattern)
 	defer qt.Done()
-	return vmstorage.GetMetricNamesStats(qt, limit, le, matchPattern)
+	return vmstorage.VMSelectAPI.GetMetricNamesUsageStats(qt, nil, limit, le, matchPattern, 0)
 }
 
 // ResetMetricNamesStats resets state of metric names usage
 func ResetMetricNamesStats(qt *querytracer.Tracer) error {
 	qt = qt.NewChild("reset metric names usage stats")
 	defer qt.Done()
-	vmstorage.ResetMetricNamesStats(qt)
-	return nil
+	return vmstorage.VMSelectAPI.ResetMetricNamesUsageStats(qt, 0)
 }

app/vmselect/prometheus/prometheus.go

@@ -28,8 +28,6 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httpserver"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/httputil"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/memory"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/netutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/querytracer"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
@@ -50,9 +48,6 @@ var (
 		"If set to true, the query model becomes closer to InfluxDB data model. If set to true, then -search.maxLookback and -search.maxStalenessInterval are ignored")
 	maxStepForPointsAdjustment = flag.Duration("search.maxStepForPointsAdjustment", time.Minute, "The maximum step when /api/v1/query_range handler adjusts "+
 		"points with timestamps closer than -search.latencyOffset to the current time. The adjustment is needed because such points may contain incomplete data")
-
-	maxUniqueTimeseries = flag.Int("search.maxUniqueTimeseries", 0, "The maximum number of unique time series, which can be selected during /api/v1/query and /api/v1/query_range queries. This option allows limiting memory usage. "+
-		"When set to zero, the limit is automatically calculated based on -search.maxConcurrentRequests (inversely proportional) and memory available to the process (proportional).")
 	maxFederateSeries       = flag.Int("search.maxFederateSeries", 1e6, "The maximum number of time series, which can be returned from /federate. This option allows limiting memory usage")
 	maxExportSeries         = flag.Int("search.maxExportSeries", 10e6, "The maximum number of time series, which can be returned from /api/v1/export* APIs. This option allows limiting memory usage")
 	maxTSDBStatusSeries     = flag.Int("search.maxTSDBStatusSeries", 10e6, "The maximum number of time series, which can be processed during the call to /api/v1/status/tsdb. This option allows limiting memory usage")
@@ -853,7 +848,7 @@ func QueryHandler(qt *querytracer.Tracer, startTime time.Time, w http.ResponseWr
 		End:                 start,
 		Step:                step,
 		MaxPointsPerSeries:  *maxPointsPerTimeseries,
-		MaxSeries:           GetMaxUniqueTimeSeries(),
+		MaxSeries:           0, // let vmstorage use maxUniqueTimeseries by default
 		QuotedRemoteAddr:    httpserver.GetQuotedRemoteAddr(r),
 		Deadline:            deadline,
 		MayCache:            mayCache,
@@ -964,7 +959,7 @@ func queryRangeHandler(qt *querytracer.Tracer, startTime time.Time, w http.Respo
 		End:                 end,
 		Step:                step,
 		MaxPointsPerSeries:  *maxPointsPerTimeseries,
-		MaxSeries:           GetMaxUniqueTimeSeries(),
+		MaxSeries:           0, // let vmstorage use maxUniqueTimeseries by default
 		QuotedRemoteAddr:    httpserver.GetQuotedRemoteAddr(r),
 		Deadline:            deadline,
 		MayCache:            mayCache,
@@ -1223,11 +1218,7 @@ func getCommonParamsInternal(r *http.Request, startTime time.Time, requireNonEmp
 	if err != nil {
 		return nil, err
 	}
-	// Limit the `end` arg to the current time +2 days in the same way
-	// as it is limited during data ingestion.
-	// See https://github.com/VictoriaMetrics/VictoriaMetrics/blob/ea06d2fd3ccbbb6aa4480ab3b04f7b671408be2a/lib/storage/table.go#L378
-	// This should fix possible timestamp overflow - see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2669
-	maxTS := startTime.UnixNano()/1e6 + 2*24*3600*1000
+	maxTS := int64(math.MaxInt64 / 1_000_000)
 	if end > maxTS {
 		end = maxTS
 	}
@@ -1304,43 +1295,6 @@ func (sw *scalableWriter) flush() error {
 	return sw.bw.Flush()
 }
 
-var (
-	maxUniqueTimeseriesValueOnce sync.Once
-	maxUniqueTimeseriesValue     int
-)
-
-// InitMaxUniqueTimeseries init the max metrics limit calculated by available resources.
-// The calculation is split into calculateMaxUniqueTimeSeriesForResource for unit testing.
-func InitMaxUniqueTimeseries(maxConcurrentRequests int) {
-	maxUniqueTimeseriesValueOnce.Do(func() {
-		maxUniqueTimeseriesValue = *maxUniqueTimeseries
-		if maxUniqueTimeseriesValue <= 0 {
-			maxUniqueTimeseriesValue = calculateMaxUniqueTimeSeriesForResource(maxConcurrentRequests, memory.Remaining())
-		}
-	})
-}
-
-// calculateMaxUniqueTimeSeriesForResource calculate the max metrics limit calculated by available resources.
-func calculateMaxUniqueTimeSeriesForResource(maxConcurrentRequests, remainingMemory int) int {
-	if maxConcurrentRequests <= 0 {
-		// This line should NOT be reached unless the user has set an incorrect `search.maxConcurrentRequests`.
-		// In such cases, fallback to unlimited.
-		logger.Warnf("limiting -search.maxUniqueTimeseries to %v because -search.maxConcurrentRequests=%d.", 2e9, maxConcurrentRequests)
-		return 2e9
-	}
-
-	// Calculate the max metrics limit for a single request in the worst-case concurrent scenario.
-	// The approximate size of 1 unique series that could occupy in the vmstorage is 200 bytes.
-	mts := remainingMemory / 200 / maxConcurrentRequests
-	logger.Infof("limiting -search.maxUniqueTimeseries to %d according to -search.maxConcurrentRequests=%d and remaining memory=%d bytes. To increase the limit, reduce -search.maxConcurrentRequests or increase memory available to the process.", mts, maxConcurrentRequests, remainingMemory)
-	return mts
-}
-
-// GetMaxUniqueTimeSeries returns the max metrics limit calculated by available resources.
-func GetMaxUniqueTimeSeries() int {
-	return maxUniqueTimeseriesValue
-}
-
 // copied from https://github.com/prometheus/common/blob/adea6285c1c7447fcb7bfdeb6abfc6eff893e0a7/model/metric.go#L483
 // it's not possible to use direct import due to increased binary size
 func unescapePrometheusLabelName(name string) string {

app/vmselect/prometheus/prometheus_test.go

@@ -4,7 +4,6 @@ import (
 	"math"
 	"net/http"
 	"reflect"
-	"runtime"
 	"testing"
 
 	"github.com/VictoriaMetrics/VictoriaMetrics/app/vmselect/netstorage"
@@ -230,29 +229,3 @@ func TestGetLatencyOffsetMillisecondsFailure(t *testing.T) {
 	}
 	f("http://localhost?latency_offset=foobar")
 }
-
-func TestCalculateMaxMetricsLimitByResource(t *testing.T) {
-	f := func(maxConcurrentRequest, remainingMemory, expect int) {
-		t.Helper()
-		maxMetricsLimit := calculateMaxUniqueTimeSeriesForResource(maxConcurrentRequest, remainingMemory)
-		if maxMetricsLimit != expect {
-			t.Fatalf("unexpected max metrics limit: got %d, want %d", maxMetricsLimit, expect)
-		}
-	}
-
-	// Skip when GOARCH=386
-	if runtime.GOARCH != "386" {
-		// 8 CPU & 32 GiB
-		f(16, int(math.Round(32*1024*1024*1024*0.4)), 4294967)
-		// 4 CPU & 32 GiB
-		f(8, int(math.Round(32*1024*1024*1024*0.4)), 8589934)
-	}
-
-	// 2 CPU & 4 GiB
-	f(4, int(math.Round(4*1024*1024*1024*0.4)), 2147483)
-
-	// other edge cases
-	f(0, int(math.Round(4*1024*1024*1024*0.4)), 2e9)
-	f(4, 0, 0)
-
-}

app/vmselect/searchutil/searchutil.go

@@ -132,9 +132,20 @@ func (d *Deadline) String() string {
 //
 //	{env="prod",team="devops",t1="v1",t2="v2"}
 //	{env=~"dev|staging",team!="devops",t1="v1",t2="v2"}
+//
+//	Query args from URL path have precedence over post form args.
 func GetExtraTagFilters(r *http.Request) ([][]storage.TagFilter, error) {
 	var tagFilters []storage.TagFilter
-	for _, match := range r.Form["extra_label"] {
+	urlQueryValues := r.URL.Query()
+	getRequestParam := func(key string) []string {
+		// query request param must always take precedence over form values
+		// in order to simplify security enforcement policy for extra_label and extra_filters
+		if uv, ok := urlQueryValues[key]; ok {
+			return uv
+		}
+		return r.Form[key]
+	}
+	for _, match := range getRequestParam("extra_label") {
 		tmp := strings.SplitN(match, "=", 2)
 		if len(tmp) != 2 {
 			return nil, fmt.Errorf("`extra_label` query arg must have the format `name=value`; got %q", match)
@@ -148,8 +159,8 @@ func GetExtraTagFilters(r *http.Request) ([][]storage.TagFilter, error) {
 			Value: []byte(tmp[1]),
 		})
 	}
-	extraFilters := append([]string{}, r.Form["extra_filters"]...)
-	extraFilters = append(extraFilters, r.Form["extra_filters[]"]...)
+	extraFilters := append([]string{}, getRequestParam("extra_filters")...)
+	extraFilters = append(extraFilters, getRequestParam("extra_filters[]")...)
 	if len(extraFilters) == 0 {
 		if len(tagFilters) == 0 {
 			return nil, nil

app/vmselect/searchutil/searchutil_test.go

@@ -20,6 +20,7 @@ func TestGetExtraTagFilters(t *testing.T) {
 		}
 		return &http.Request{
 			Form: q,
+			URL:  &url.URL{RawQuery: q.Encode()},
 		}
 	}
 	f := func(t *testing.T, r *http.Request, want []string, wantErr bool) {
@@ -79,6 +80,24 @@ func TestGetExtraTagFilters(t *testing.T) {
 		nil,
 		false,
 	)
+
+	formValues, err := url.ParseQuery(`extra_label=env=prod&extra_label=job=vmsingle&extra_label=tenant=prod&extra_filters[]={foo="bar"}&extra_filters[]={tenant="prod"}`)
+	if err != nil {
+		t.Fatalf("BUG: cannot parse query: %s", err)
+	}
+	urlValues, err := url.ParseQuery(`extra_label=job=vmagent&extra_label=env=dev&extra_filters[]={tenant="dev"}`)
+	if err != nil {
+		t.Fatalf("BUG: cannot parse query: %s", err)
+	}
+	httpReqWithBothFormAndURLParams := &http.Request{
+		Form: formValues,
+		URL: &url.URL{
+			RawQuery: urlValues.Encode(),
+		},
+	}
+	f(t, httpReqWithBothFormAndURLParams,
+		[]string{`{tenant="dev",job="vmagent",env="dev"}`},
+		false)
 }
 
 func TestParseMetricSelectorSuccess(t *testing.T) {

app/vmselect/vmui/index.html

@@ -37,11 +37,11 @@
   <meta property="og:title" content="UI for VictoriaMetrics">
   <meta property="og:url" content="https://victoriametrics.com/">
   <meta property="og:description" content="Explore and troubleshoot your VictoriaMetrics data">
-  <script type="module" crossorigin src="./assets/index-C24BPpD_.js"></script>
+  <script type="module" crossorigin src="./assets/index-BjJ7fDL7.js"></script>
   <link rel="modulepreload" crossorigin href="./assets/rolldown-runtime-COnpUsM8.js">
-  <link rel="modulepreload" crossorigin href="./assets/vendor-BWBgVCcr.js">
+  <link rel="modulepreload" crossorigin href="./assets/vendor-C8Kwp93_.js">
   <link rel="stylesheet" crossorigin href="./assets/vendor-CnsZ1jie.css">
-  <link rel="stylesheet" crossorigin href="./assets/index-D2OEy8Ra.css">
+  <link rel="stylesheet" crossorigin href="./assets/index-BL7jEFBa.css">
 </head>
 <body>
 <noscript>You need to enable JavaScript to run this app.</noscript>

app/vmstorage/main.go

@@ -1,7 +1,6 @@
 package vmstorage
 
 import (
-	"errors"
 	"flag"
 	"fmt"
 	"io"
@@ -9,12 +8,10 @@ import (
 	"net/http"
 	"strconv"
 	"strings"
-	"sync"
 	"time"
 
 	"github.com/VictoriaMetrics/metrics"
 
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fasttime"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/fs"
@@ -23,27 +20,22 @@ import (
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/mergeset"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/querytracer"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage/metricnamestats"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage/metricsmetadata"
 	"github.com/VictoriaMetrics/VictoriaMetrics/lib/stringsutil"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/syncwg"
-	"github.com/VictoriaMetrics/VictoriaMetrics/lib/timeutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/vminsertapi"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/vmselectapi"
 )
 
 var (
+	storageDataPath = flag.String("storageDataPath", "victoria-metrics-data", "Path to storage data")
 	retentionPeriod = flagutil.NewRetentionDuration("retentionPeriod", "1M", "Data with timestamps outside the retentionPeriod is automatically deleted. The minimum retentionPeriod is 24h or 1d. "+
 		"See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#retention. See also -retentionFilter")
+	futureRetention = flagutil.NewRetentionDuration("futureRetention", "2d", "Data with timestamps bigger than now+futureRetention is automatically deleted. "+
+		"The minimum futureRetention is 2 days. See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#retention")
 	snapshotAuthKey   = flagutil.NewPassword("snapshotAuthKey", "authKey, which must be passed in query string to /snapshot* pages. It overrides -httpAuth.*")
 	forceMergeAuthKey = flagutil.NewPassword("forceMergeAuthKey", "authKey, which must be passed in query string to /internal/force_merge pages. It overrides -httpAuth.*")
 	forceFlushAuthKey = flagutil.NewPassword("forceFlushAuthKey", "authKey, which must be passed in query string to /internal/force_flush pages. It overrides -httpAuth.*")
-	snapshotsMaxAge   = flagutil.NewRetentionDuration("snapshotsMaxAge", "3d", "Automatically delete snapshots older than -snapshotsMaxAge if it is set to non-zero duration. Make sure that backup process has enough time to finish the backup before the corresponding snapshot is automatically deleted")
 	_                 = flag.Duration("snapshotCreateTimeout", 0, "Deprecated: this flag does nothing")
 
-	precisionBits = flag.Int("precisionBits", 64, "The number of precision bits to store per each value. Lower precision bits improves data compression at the cost of precision loss")
-
-	// DataPath is a path to storage data.
-	DataPath = flag.String("storageDataPath", "victoria-metrics-data", "Path to storage data")
-
 	_ = flag.Duration("finalMergeDelay", 0, "Deprecated: this flag does nothing")
 	_ = flag.Int("bigMergeConcurrency", 0, "Deprecated: this flag does nothing")
 	_ = flag.Int("smallMergeConcurrency", 0, "Deprecated: this flag does nothing")
@@ -51,11 +43,17 @@ var (
 	retentionTimezoneOffset = flag.Duration("retentionTimezoneOffset", 0, "The offset for performing indexdb rotation. "+
 		"If set to 0, then the indexdb rotation is performed at 4am UTC time per each -retentionPeriod. "+
 		"If set to 2h, then the indexdb rotation is performed at 4am EET time (the timezone with +2h offset)")
+	minScrapeInterval = flag.Duration("dedup.minScrapeInterval", 0, "Leave only the last sample in every time series per each discrete interval "+
+		"equal to -dedup.minScrapeInterval > 0. See also -streamAggr.dedupInterval and https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#deduplication")
+	inmemoryDataFlushInterval = flag.Duration("inmemoryDataFlushInterval", 5*time.Second, "The interval for guaranteed saving of in-memory data to disk. "+
+		"The saved data survives unclean shutdowns such as OOM crash, hardware reset, SIGKILL, etc. "+
+		"Bigger intervals may help increase the lifetime of flash storage with limited write cycles (e.g. Raspberry PI). "+
+		"Smaller intervals increase disk IO load. Minimum supported value is 1s")
 
 	logNewSeries = flag.Bool("logNewSeries", false, "Whether to log new series. This option is for debug purposes only. It can lead to performance issues "+
 		"when big number of new series are ingested into VictoriaMetrics")
-	denyQueriesOutsideRetention = flag.Bool("denyQueriesOutsideRetention", false, "Whether to deny queries outside the configured -retentionPeriod. "+
-		"When set, then /api/v1/query_range would return '503 Service Unavailable' error for queries with 'from' value outside -retentionPeriod. "+
+	denyQueriesOutsideRetention = flag.Bool("denyQueriesOutsideRetention", false, "Whether to deny queries outside the configured -retentionPeriod and -futureRetention. "+
+		"When set, then /api/v1/query_range will return an error for queries with 'from' value outside -retentionPeriod or 'to' value beyond -futureRetention. "+
 		"This may be useful when multiple data sources with distinct retentions are hidden behind query-tee")
 	maxHourlySeries = flag.Int64("storage.maxHourlySeries", 0, "The maximum number of unique series can be added to the storage during the last hour. "+
 		"Excess series are logged and dropped. This can be useful for limiting series cardinality. See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#cardinality-limiter . "+
@@ -68,6 +66,11 @@ var (
 
 	minFreeDiskSpaceBytes = flagutil.NewBytes("storage.minFreeDiskSpaceBytes", 100e6, "The minimum free disk space at -storageDataPath after which the storage stops accepting new data")
 
+	finalDedupScheduleInterval = flag.Duration("storage.finalDedupScheduleCheckInterval", time.Hour, "The interval for checking when final deduplication process should be started."+
+		"Storage unconditionally adds 25% jitter to the interval value on each check evaluation."+
+		" Changing the interval to the bigger values may delay downsampling, deduplication for historical data."+
+		" See also https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#deduplication")
+
 	cacheSizeStorageTSID = flagutil.NewBytes("storage.cacheSizeStorageTSID", 0, "Overrides max size for storage/tsid cache. "+
 		"See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#cache-tuning")
 	cacheSizeStorageMetricName = flagutil.NewBytes("storage.cacheSizeStorageMetricName", 0, "Overrides max size for storage/metricName cache. "+
@@ -101,32 +104,22 @@ var (
 		"If set to 0 or a negative value, defaults to 1% of allowed memory.")
 )
 
-// CheckTimeRange returns true if the given tr is denied for querying.
-func CheckTimeRange(tr storage.TimeRange) error {
-	if !*denyQueriesOutsideRetention {
-		return nil
-	}
-	minAllowedTimestamp := int64(fasttime.UnixTimestamp()*1000) - retentionPeriod.Milliseconds()
-	if tr.MinTimestamp > minAllowedTimestamp {
-		return nil
-	}
-	return &httpserver.ErrorWithStatusCode{
-		Err:        fmt.Errorf("the given time range %s is outside the allowed -retentionPeriod=%s according to -denyQueriesOutsideRetention", &tr, retentionPeriod),
-		StatusCode: http.StatusServiceUnavailable,
-	}
+func DataPath() string {
+	return *storageDataPath
 }
 
 // Init initializes vmstorage.
-func Init(resetCacheIfNeeded func(mrs []storage.MetricRow)) {
-	if err := encoding.CheckPrecisionBits(uint8(*precisionBits)); err != nil {
-		logger.Fatalf("invalid `-precisionBits`: %s", err)
-	}
-
-	resetResponseCacheIfNeeded = resetCacheIfNeeded
+func Init(maxConcurrentRequests int, resetCacheIfNeeded func(mrs []storage.MetricRow)) {
+	storage.SetDedupInterval(*minScrapeInterval)
+	storage.SetDataFlushInterval(*inmemoryDataFlushInterval)
 	storage.LegacySetRetentionTimezoneOffset(*retentionTimezoneOffset)
 	storage.SetFreeDiskSpaceLimit(minFreeDiskSpaceBytes.N)
 	storage.SetTSIDCacheSize(cacheSizeStorageTSID.IntN())
 	storage.SetTagFiltersCacheSize(cacheSizeIndexDBTagFilters.IntN())
+	if *finalDedupScheduleInterval < time.Hour {
+		logger.Fatalf("-storage.finalDedupScheduleCheckInterval cannot be smaller than 1 hour; got %s", *finalDedupScheduleInterval)
+	}
+	storage.SetFinalDedupScheduleInterval(*finalDedupScheduleInterval)
 	storage.SetMetricNamesStatsCacheSize(cacheSizeMetricNamesStats.IntN())
 	storage.SetMetricNameCacheSize(cacheSizeStorageMetricName.IntN())
 	storage.SetMetadataStorageSize(metadataStorageSize.IntN())
@@ -135,26 +128,32 @@ func Init(resetCacheIfNeeded func(mrs []storage.MetricRow)) {
 	mergeset.SetDataBlocksSparseCacheSize(cacheSizeIndexDBDataBlocksSparse.IntN())
 
 	if retentionPeriod.Duration() < 24*time.Hour {
-		logger.Fatalf("-retentionPeriod cannot be smaller than a day; got %s", retentionPeriod)
+		logger.Fatalf("-retentionPeriod cannot be smaller than a day; got %s. "+
+			"See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#retention", retentionPeriod)
+	}
+	if futureRetention.Duration() < 2*24*time.Hour {
+		logger.Fatalf("-futureRetention cannot be smaller than 2 days; got %s. "+
+			"See https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#retention", futureRetention)
 	}
 	if *idbPrefillStart > 23*time.Hour {
 		logger.Panicf("-storage.idbPrefillStart cannot exceed 23 hours; got %s", idbPrefillStart)
 	}
-	logger.Infof("opening storage at %q with -retentionPeriod=%s", *DataPath, retentionPeriod)
+	fs.RegisterPathFsMetrics(*storageDataPath)
+	logger.Infof("opening storage at %q with -retentionPeriod=%s", *storageDataPath, retentionPeriod)
 	startTime := time.Now()
-	WG = syncwg.WaitGroup{}
 	opts := storage.OpenOptions{
-		Retention:             retentionPeriod.Duration(),
-		MaxHourlySeries:       getMaxHourlySeries(),
-		MaxDailySeries:        getMaxDailySeries(),
-		DisablePerDayIndex:    *disablePerDayIndex,
-		TrackMetricNamesStats: *trackMetricNamesStats,
-		IDBPrefillStart:       *idbPrefillStart,
-		LogNewSeries:          *logNewSeries,
+		Retention:                   retentionPeriod.Duration(),
+		FutureRetention:             futureRetention.Duration(),
+		DenyQueriesOutsideRetention: *denyQueriesOutsideRetention,
+		MaxHourlySeries:             getMaxHourlySeries(),
+		MaxDailySeries:              getMaxDailySeries(),
+		DisablePerDayIndex:          *disablePerDayIndex,
+		TrackMetricNamesStats:       *trackMetricNamesStats,
+		IDBPrefillStart:             *idbPrefillStart,
+		LogNewSeries:                *logNewSeries,
 	}
-	strg := storage.MustOpenStorage(*DataPath, opts)
-	Storage = strg
-	initStaleSnapshotsRemover(strg)
+	strg := storage.MustOpenStorage(*storageDataPath, opts)
+	vmStorage = newVMStorageSingleNode(strg, maxConcurrentRequests, resetCacheIfNeeded)
 
 	var m storage.Metrics
 	strg.UpdateMetrics(&m)
@@ -164,151 +163,46 @@ func Init(resetCacheIfNeeded func(mrs []storage.MetricRow)) {
 	rowsCount := tm.SmallRowsCount + tm.BigRowsCount
 	sizeBytes := tm.SmallSizeBytes + tm.BigSizeBytes
 	logger.Infof("successfully opened storage %q in %.3f seconds; partsCount: %d; blocksCount: %d; rowsCount: %d; sizeBytes: %d",
-		*DataPath, time.Since(startTime).Seconds(), partsCount, blocksCount, rowsCount, sizeBytes)
+		*storageDataPath, time.Since(startTime).Seconds(), partsCount, blocksCount, rowsCount, sizeBytes)
 
 	// register storage metrics
 	storageMetrics = metrics.NewSet()
 	storageMetrics.RegisterMetricsWriter(func(w io.Writer) {
-		writeStorageMetrics(w, strg)
+		vmStorage.writeStorageMetrics(w)
 	})
 	metrics.RegisterSet(storageMetrics)
+
+	VMInsertAPI = vmStorage
+	VMSelectAPI = vmStorage
+	GetSearch = func(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) (*storage.Search, int, error) {
+		return vmStorage.GetSearch(qt, sq, deadline)
+	}
+	PutSearch = func(sr *storage.Search) {
+		vmStorage.PutSearch(sr)
+	}
+	RequestHandler = func(w http.ResponseWriter, r *http.Request) bool {
+		return vmStorage.requestHandler(w, r)
+	}
+	DebugFlush = func() {
+		vmStorage.s.DebugFlush()
+	}
 }
 
 var storageMetrics *metrics.Set
 
-// Storage is a storage.
-//
-// Every storage call must be wrapped into WG.Add(1) ... WG.Done()
-// for proper graceful shutdown when Stop is called.
-var Storage *storage.Storage
+var (
+	// vmStorageSingleNode is an instance of vmstorage used by vminsert and
+	// vmselect for writing and reading data.
+	vmStorage      *VMStorageSingleNode
+	VMInsertAPI    vminsertapi.API
+	VMSelectAPI    vmselectapi.API
+	GetSearch      func(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) (*storage.Search, int, error)
+	PutSearch      func(sr *storage.Search)
+	RequestHandler func(w http.ResponseWriter, r *http.Request) bool
 
-// WG must be incremented before Storage call.
-//
-// Use syncwg instead of sync, since Add is called from concurrent goroutines.
-var WG syncwg.WaitGroup
-
-// resetResponseCacheIfNeeded is a callback for automatic resetting of response cache if needed.
-var resetResponseCacheIfNeeded func(mrs []storage.MetricRow)
-
-// AddRows adds mrs to the storage.
-//
-// The caller should limit the number of concurrent calls to AddRows() in order to limit memory usage.
-func AddRows(mrs []storage.MetricRow) error {
-	if Storage.IsReadOnly() {
-		return errReadOnly
-	}
-	resetResponseCacheIfNeeded(mrs)
-	WG.Add(1)
-	Storage.AddRows(mrs, uint8(*precisionBits))
-	WG.Done()
-	return nil
-}
-
-// AddMetadataRows adds mrs to the storage.
-//
-// The caller should limit the number of concurrent calls to AddMetadataRows() in order to limit memory usage.
-func AddMetadataRows(mms []metricsmetadata.Row) error {
-	if Storage.IsReadOnly() {
-		return errReadOnly
-	}
-	WG.Add(1)
-	Storage.AddMetadataRows(mms)
-	WG.Done()
-	return nil
-}
-
-var errReadOnly = errors.New("the storage is in read-only mode; check -storage.minFreeDiskSpaceBytes command-line flag value")
-
-// RegisterMetricNames registers all the metrics from mrs in the storage.
-func RegisterMetricNames(qt *querytracer.Tracer, mrs []storage.MetricRow) {
-	WG.Add(1)
-	Storage.RegisterMetricNames(qt, mrs)
-	WG.Done()
-}
-
-// DeleteSeries deletes series matching tfss.
-//
-// Returns the number of deleted series.
-func DeleteSeries(qt *querytracer.Tracer, tfss []*storage.TagFilters, maxMetrics int) (int, error) {
-	WG.Add(1)
-	n, err := Storage.DeleteSeries(qt, tfss, maxMetrics)
-	WG.Done()
-	return n, err
-}
-
-// GetMetricNamesStats returns metric names usage stats with give limit and lte predicate
-func GetMetricNamesStats(qt *querytracer.Tracer, limit, le int, matchPattern string) (metricnamestats.StatsResult, error) {
-	WG.Add(1)
-	r := Storage.GetMetricNamesStats(qt, limit, le, matchPattern)
-	WG.Done()
-	return r, nil
-}
-
-// ResetMetricNamesStats resets state for metric names usage tracker
-func ResetMetricNamesStats(qt *querytracer.Tracer) {
-	WG.Add(1)
-	Storage.ResetMetricNamesStats(qt)
-	WG.Done()
-}
-
-// SearchMetricNames returns metric names for the given tfss on the given tr.
-func SearchMetricNames(qt *querytracer.Tracer, tfss []*storage.TagFilters, tr storage.TimeRange, maxMetrics int, deadline uint64) ([]string, error) {
-	WG.Add(1)
-	metricNames, err := Storage.SearchMetricNames(qt, tfss, tr, maxMetrics, deadline)
-	WG.Done()
-	return metricNames, err
-}
-
-// SearchLabelNames searches for tag keys matching the given tfss on tr.
-func SearchLabelNames(qt *querytracer.Tracer, tfss []*storage.TagFilters, tr storage.TimeRange, maxTagKeys, maxMetrics int, deadline uint64) ([]string, error) {
-	WG.Add(1)
-	labelNames, err := Storage.SearchLabelNames(qt, tfss, tr, maxTagKeys, maxMetrics, deadline)
-	WG.Done()
-	return labelNames, err
-}
-
-// SearchLabelValues searches for label values for the given labelName, tfss and
-// tr.
-func SearchLabelValues(qt *querytracer.Tracer, labelName string, tfss []*storage.TagFilters, tr storage.TimeRange, maxLabelValues, maxMetrics int, deadline uint64) ([]string, error) {
-	WG.Add(1)
-	labelValues, err := Storage.SearchLabelValues(qt, labelName, tfss, tr, maxLabelValues, maxMetrics, deadline)
-	WG.Done()
-	return labelValues, err
-}
-
-// SearchTagValueSuffixes returns all the tag value suffixes for the given tagKey and tagValuePrefix on the given tr.
-//
-// This allows implementing https://graphite-api.readthedocs.io/en/latest/api.html#metrics-find or similar APIs.
-func SearchTagValueSuffixes(qt *querytracer.Tracer, tr storage.TimeRange, tagKey, tagValuePrefix string, delimiter byte, maxTagValueSuffixes int, deadline uint64) ([]string, error) {
-	WG.Add(1)
-	suffixes, err := Storage.SearchTagValueSuffixes(qt, tr, tagKey, tagValuePrefix, delimiter, maxTagValueSuffixes, deadline)
-	WG.Done()
-	return suffixes, err
-}
-
-// SearchGraphitePaths returns all the metric names matching the given Graphite query.
-func SearchGraphitePaths(qt *querytracer.Tracer, tr storage.TimeRange, query []byte, maxPaths int, deadline uint64) ([]string, error) {
-	WG.Add(1)
-	paths, err := Storage.SearchGraphitePaths(qt, tr, query, maxPaths, deadline)
-	WG.Done()
-	return paths, err
-}
-
-// GetTSDBStatus returns TSDB status for given filters on the given date.
-func GetTSDBStatus(qt *querytracer.Tracer, tfss []*storage.TagFilters, date uint64, focusLabel string, topN, maxMetrics int, deadline uint64) (*storage.TSDBStatus, error) {
-	WG.Add(1)
-	status, err := Storage.GetTSDBStatus(qt, tfss, date, focusLabel, topN, maxMetrics, deadline)
-	WG.Done()
-	return status, err
-}
-
-// GetSeriesCount returns the number of time series in the storage.
-func GetSeriesCount(deadline uint64) (uint64, error) {
-	WG.Add(1)
-	n, err := Storage.GetSeriesCount(deadline)
-	WG.Done()
-	return n, err
-}
+	// TODO(@rtm0): Remove this dependency from vmalert-tool unit tests.
+	DebugFlush func()
+)
 
 // Stop stops the vmstorage
 func Stop() {
@@ -316,19 +210,24 @@ func Stop() {
 	metrics.UnregisterSet(storageMetrics, true)
 	storageMetrics = nil
 
-	logger.Infof("gracefully closing the storage at %s", *DataPath)
+	logger.Infof("gracefully closing the storage at %s", *storageDataPath)
 	startTime := time.Now()
-	WG.WaitAndBlock()
-	stopStaleSnapshotsRemover()
-	Storage.MustClose()
+	vmStorage.Stop()
 	logger.Infof("successfully closed the storage in %.3f seconds", time.Since(startTime).Seconds())
 
 	fs.MustStopDirRemover()
-	logger.Infof("the storage has been stopped")
+	logger.Infof("the vmstorage has been stopped")
 }
 
-// RequestHandler is a storage request handler.
-func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
+func (api *VMStorageSingleNode) requestHandler(w http.ResponseWriter, r *http.Request) bool {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.requestHandler(w, r)
+}
+
+// requestHandler is a storage request handler.
+// TODO(@rtm0): Move to a separate file, request_handler.go
+func (api *VMStorage) requestHandler(w http.ResponseWriter, r *http.Request) bool {
 	path := r.URL.Path
 	if path == "/internal/force_merge" {
 		if !httpserver.CheckAuthFlag(w, r, forceMergeAuthKey) {
@@ -341,8 +240,9 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 			defer activeForceMerges.Dec()
 			logger.Infof("forced merge for partition_prefix=%q has been started", partitionNamePrefix)
 			startTime := time.Now()
-			if err := Storage.ForceMergePartitions(partitionNamePrefix); err != nil {
+			if err := api.s.ForceMergePartitions(partitionNamePrefix); err != nil {
 				logger.Errorf("error in forced merge for partition_prefix=%q: %s", partitionNamePrefix, err)
+				return
 			}
 			logger.Infof("forced merge for partition_prefix=%q has been successfully finished in %.3f seconds", partitionNamePrefix, time.Since(startTime).Seconds())
 		}()
@@ -353,9 +253,10 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 			return true
 		}
 		logger.Infof("flushing storage to make pending data available for reading")
-		Storage.DebugFlush()
+		api.s.DebugFlush()
 		return true
 	}
+
 	if path == "/internal/log_new_series" {
 		if !httpserver.CheckAuthFlag(w, r, logNewSeriesAuthKey) {
 			return true
@@ -372,7 +273,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 		}
 		logger.Infof("enabling logging of new series for the next %s. This may increase resource usage during this period.", time.Duration(dealine)*time.Second)
 		endTime := fasttime.UnixTimestamp() + uint64(dealine)
-		Storage.SetLogNewSeriesUntil(endTime)
+		api.s.SetLogNewSeriesUntil(endTime)
 		fmt.Fprintf(w, `{"status":"success","data":{"logEndTime":%q}}`, time.Unix(int64(endTime), 0))
 		return true
 	}
@@ -394,13 +295,13 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 	case "/create":
 		snapshotsCreateTotal.Inc()
 		w.Header().Set("Content-Type", "application/json")
-		snapshotName := Storage.MustCreateSnapshot()
+		snapshotName := api.s.MustCreateSnapshot()
 
 		// Verify whether the client already closed the connection.
 		// In this case it is better to drop the created snapshot, since the client isn't interested in it.
 		if err := r.Context().Err(); err != nil {
 			logger.Infof("deleting already created snapshot at %s because the client canceled the request", snapshotName)
-			if err := deleteSnapshot(snapshotName); err != nil {
+			if err := api.deleteSnapshot(snapshotName); err != nil {
 				logger.Infof("cannot delete just created snapshot: %s", err)
 				return true
 			}
@@ -416,7 +317,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 	case "/list":
 		snapshotsListTotal.Inc()
 		w.Header().Set("Content-Type", "application/json")
-		snapshots := Storage.MustListSnapshots()
+		snapshots := api.s.MustListSnapshots()
 		fmt.Fprintf(w, `{"status":"ok","snapshots":[`)
 		if len(snapshots) > 0 {
 			for _, snapshot := range snapshots[:len(snapshots)-1] {
@@ -430,7 +331,7 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 		snapshotsDeleteTotal.Inc()
 		w.Header().Set("Content-Type", "application/json")
 		snapshotName := r.FormValue("snapshot")
-		if err := deleteSnapshot(snapshotName); err != nil {
+		if err := api.deleteSnapshot(snapshotName); err != nil {
 			jsonResponseError(w, err)
 			snapshotsDeleteErrorsTotal.Inc()
 			return true
@@ -440,9 +341,10 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 	case "/delete_all":
 		snapshotsDeleteAllTotal.Inc()
 		w.Header().Set("Content-Type", "application/json")
-		snapshots := Storage.MustListSnapshots()
+		snapshots := api.s.MustListSnapshots()
 		for _, snapshotName := range snapshots {
-			if err := Storage.DeleteSnapshot(snapshotName); err != nil {
+			// TODO(@rtm0): Use VMStorage.deleteSnapshot()?
+			if err := api.s.DeleteSnapshot(snapshotName); err != nil {
 				err = fmt.Errorf("cannot delete snapshot %q: %w", snapshotName, err)
 				jsonResponseError(w, err)
 				snapshotsDeleteAllErrorsTotal.Inc()
@@ -456,50 +358,6 @@ func RequestHandler(w http.ResponseWriter, r *http.Request) bool {
 	}
 }
 
-func deleteSnapshot(snapshotName string) error {
-	snapshots := Storage.MustListSnapshots()
-	for _, snName := range snapshots {
-		if snName == snapshotName {
-			if err := Storage.DeleteSnapshot(snName); err != nil {
-				return fmt.Errorf("cannot delete snapshot %q: %w", snName, err)
-			}
-			return nil
-		}
-	}
-	return fmt.Errorf("cannot find snapshot %q", snapshotName)
-}
-
-func initStaleSnapshotsRemover(strg *storage.Storage) {
-	staleSnapshotsRemoverCh = make(chan struct{})
-	if snapshotsMaxAge.Duration() <= 0 {
-		return
-	}
-	snapshotsMaxAgeDur := snapshotsMaxAge.Duration()
-	staleSnapshotsRemoverWG.Go(func() {
-		d := timeutil.AddJitterToDuration(time.Second * 11)
-		t := time.NewTicker(d)
-		defer t.Stop()
-		for {
-			select {
-			case <-staleSnapshotsRemoverCh:
-				return
-			case <-t.C:
-			}
-			strg.MustDeleteStaleSnapshots(snapshotsMaxAgeDur)
-		}
-	})
-}
-
-func stopStaleSnapshotsRemover() {
-	close(staleSnapshotsRemoverCh)
-	staleSnapshotsRemoverWG.Wait()
-}
-
-var (
-	staleSnapshotsRemoverCh chan struct{}
-	staleSnapshotsRemoverWG sync.WaitGroup
-)
-
 var (
 	activeForceMerges = metrics.NewCounter("vm_active_force_merges")
 
@@ -514,21 +372,23 @@ var (
 	snapshotsDeleteAllErrorsTotal = metrics.NewCounter(`vm_http_request_errors_total{path="/snapshot/delete_all"}`)
 )
 
-func writeStorageMetrics(w io.Writer, strg *storage.Storage) {
+// TODO(@rtm0): Move to metrics.go.
+func (api *VMStorage) writeStorageMetrics(w io.Writer) {
+	strg := api.s
 	var m storage.Metrics
 	strg.UpdateMetrics(&m)
 	tm := &m.TableMetrics
 	idbm := &m.TableMetrics.IndexDBMetrics
 
-	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_free_disk_space_bytes{path=%q}`, *DataPath), fs.MustGetFreeSpace(*DataPath))
-	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_free_disk_space_limit_bytes{path=%q}`, *DataPath), uint64(minFreeDiskSpaceBytes.N))
-	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_total_disk_space_bytes{path=%q}`, *DataPath), fs.MustGetTotalSpace(*DataPath))
+	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_free_disk_space_bytes{path=%q}`, *storageDataPath), fs.MustGetFreeSpace(*storageDataPath))
+	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_free_disk_space_limit_bytes{path=%q}`, *storageDataPath), uint64(minFreeDiskSpaceBytes.N))
+	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_total_disk_space_bytes{path=%q}`, *storageDataPath), fs.MustGetTotalSpace(*storageDataPath))
 
 	isReadOnly := 0
 	if strg.IsReadOnly() {
 		isReadOnly = 1
 	}
-	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_storage_is_read_only{path=%q}`, *DataPath), uint64(isReadOnly))
+	metrics.WriteGaugeUint64(w, fmt.Sprintf(`vm_storage_is_read_only{path=%q}`, *storageDataPath), uint64(isReadOnly))
 
 	metrics.WriteGaugeUint64(w, `vm_active_merges{type="storage/inmemory"}`, tm.ActiveInmemoryMerges)
 	metrics.WriteGaugeUint64(w, `vm_active_merges{type="storage/small"}`, tm.ActiveSmallMerges)
@@ -738,6 +598,8 @@ func writeStorageMetrics(w io.Writer, strg *storage.Storage) {
 	metrics.WriteGaugeUint64(w, `vm_downsampling_partitions_scheduled`, tm.ScheduledDownsamplingPartitions)
 	metrics.WriteGaugeUint64(w, `vm_downsampling_partitions_scheduled_size_bytes`, tm.ScheduledDownsamplingPartitionsSize)
 
+	metrics.WriteGaugeUint64(w, `vm_search_max_unique_timeseries`, uint64(api.maxUniqueTimeSeriesCalculated))
+
 	metrics.WriteGaugeUint64(w, `vm_metrics_metadata_storage_items`, m.MetadataStorageItemsCurrent)
 	metrics.WriteCounterUint64(w, `vm_metrics_metadata_storage_size_bytes`, m.MetadataStorageCurrentSizeBytes)
 	metrics.WriteCounterUint64(w, `vm_metrics_metadata_storage_max_size_bytes`, m.MetadataStorageMaxSizeBytes)

app/vmstorage/vmstorage.go

@@ -0,0 +1,396 @@
+package vmstorage
+
+import (
+	"flag"
+	"fmt"
+	"sync"
+	"time"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/encoding"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/flagutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/logger"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/memory"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/querytracer"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage/metricnamestats"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage/metricsmetadata"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/timeutil"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/vmselectapi"
+)
+
+var (
+	precisionBits = flag.Int("precisionBits", 64, "The number of precision bits to store per each value. Lower precision bits improves data compression "+
+		"at the cost of precision loss")
+	maxUniqueTimeseries = flag.Int("search.maxUniqueTimeseries", 0, "The maximum number of unique time series, which can be scanned during every query. "+
+		"This allows protecting against heavy queries, which select unexpectedly high number of series. When set to zero, the limit is automatically calculated based on -search.maxConcurrentRequests (inversely proportional) and memory available to the process (proportional). See also -search.max* command-line flags at vmselect")
+	maxTagKeys = flag.Int("search.maxTagKeys", 100e3, "The maximum number of tag keys returned per search. "+
+		"See also -search.maxLabelsAPISeries and -search.maxLabelsAPIDuration")
+	maxTagValues = flag.Int("search.maxTagValues", 100e3, "The maximum number of tag values returned per search. "+
+		"See also -search.maxLabelsAPISeries and -search.maxLabelsAPIDuration")
+	maxTagValueSuffixesPerSearch = flag.Int("search.maxTagValueSuffixesPerSearch", 100e3, "The maximum number of tag value suffixes returned from /metrics/find")
+	snapshotsMaxAge              = flagutil.NewRetentionDuration("snapshotsMaxAge", "3d", "Automatically delete snapshots older than -snapshotsMaxAge if it is set to non-zero duration. Make sure that backup process has enough time to finish the backup before the corresponding snapshot is automatically deleted")
+)
+
+// newVMStorage creates a new instance of of VMStorage.
+//
+// The created VMStorage instance takes ownership of s.
+func newVMStorage(s *storage.Storage, maxConcurrentRequests int) *VMStorage {
+	if err := encoding.CheckPrecisionBits(uint8(*precisionBits)); err != nil {
+		logger.Fatalf("invalid -precisionBits: %d", err)
+	}
+
+	maxUniqueTimeseriesCalculated := *maxUniqueTimeseries
+	if maxUniqueTimeseriesCalculated <= 0 {
+		maxUniqueTimeseriesCalculated = calculateMaxUniqueTimeseries(maxConcurrentRequests, memory.Remaining())
+	}
+
+	vms := &VMStorage{
+		s:                             s,
+		maxUniqueTimeseries:           *maxUniqueTimeseries,
+		maxUniqueTimeSeriesCalculated: maxUniqueTimeseriesCalculated,
+		staleSnapshotsRemoverCh:       make(chan struct{}),
+	}
+	vms.initStaleSnapshotsRemover()
+	return vms
+}
+
+// calculateMaxUniqueTimeseries calculates the maxUniqueTimeseries based on the
+// available system resources.
+func calculateMaxUniqueTimeseries(maxConcurrentRequests, remainingMemory int) int {
+	if maxConcurrentRequests <= 0 {
+		// This line should NOT be reached unless the user has set an incorrect `search.maxConcurrentRequests`.
+		// In such cases, fallback to unlimited.
+		logger.Warnf("limiting -search.maxUniqueTimeseries to %v because -search.maxConcurrentRequests=%d.", 2e9, maxConcurrentRequests)
+		return 2e9
+	}
+
+	// Calculate the max metrics limit for a single request in the worst-case concurrent scenario.
+	// The approximate size of 1 unique series that could occupy in the vmstorage is 200 bytes.
+	mts := remainingMemory / 200 / maxConcurrentRequests
+	logger.Infof("limiting -search.maxUniqueTimeseries to %d according to -search.maxConcurrentRequests=%d and remaining memory=%d bytes. To increase the limit, reduce -search.maxConcurrentRequests or increase memory available to the process.", mts, maxConcurrentRequests, remainingMemory)
+	return mts
+}
+
+// VMStorage impelements vmselectapi.API and vminsertapi.API.
+type VMStorage struct {
+	s                             *storage.Storage
+	maxUniqueTimeseries           int
+	maxUniqueTimeSeriesCalculated int
+	staleSnapshotsRemoverCh       chan struct{}
+	staleSnapshotsRemoverWG       sync.WaitGroup
+}
+
+func (api *VMStorage) initStaleSnapshotsRemover() {
+	if snapshotsMaxAge.Duration() <= 0 {
+		return
+	}
+	snapshotsMaxAgeDuration := snapshotsMaxAge.Duration()
+	api.staleSnapshotsRemoverWG.Go(func() {
+		d := timeutil.AddJitterToDuration(time.Second * 11)
+		t := time.NewTicker(d)
+		defer t.Stop()
+		for {
+			select {
+			case <-api.staleSnapshotsRemoverCh:
+				return
+			case <-t.C:
+			}
+			api.s.MustDeleteStaleSnapshots(snapshotsMaxAgeDuration)
+		}
+	})
+}
+
+func (api *VMStorage) Stop() {
+	close(api.staleSnapshotsRemoverCh)
+	api.staleSnapshotsRemoverWG.Wait()
+	api.s.MustClose()
+}
+
+// WriteRows writes metric rows to the storage.
+//
+// The caller should limit the number of concurrent calls to WriteRows() in
+// order to limit memory usage.
+func (api *VMStorage) WriteRows(rows []storage.MetricRow) error {
+	api.s.AddRows(rows, uint8(*precisionBits))
+	return nil
+}
+
+// WriteMetadata writes metrics metadata to storage.
+//
+// The caller should limit the number of concurrent calls to WriteMetadata() in
+// order to limit memory usage.
+func (api *VMStorage) WriteMetadata(rows []metricsmetadata.Row) error {
+	api.s.AddMetadataRows(rows)
+	return nil
+}
+
+// IsReadOnly returns true is the storage is in read-only mode.
+func (api *VMStorage) IsReadOnly() bool {
+	return api.s.IsReadOnly()
+}
+
+func (api *VMStorage) InitSearch(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) (vmselectapi.BlockIterator, error) {
+	tr := sq.GetTimeRange()
+	maxMetrics := api.getMaxMetrics(sq.MaxMetrics)
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return nil, err
+	}
+	if len(tfss) == 0 {
+		return nil, fmt.Errorf("missing tag filters")
+	}
+	bi := getBlockIterator()
+	bi.sr.Init(qt, api.s, tfss, tr, maxMetrics, deadline)
+	if err := bi.sr.Error(); err != nil {
+		bi.MustClose()
+		return nil, err
+	}
+	return bi, nil
+}
+
+func (api *VMStorage) getMaxMetrics(searchQueryLimit int) int {
+	if searchQueryLimit <= 0 {
+		return api.maxUniqueTimeSeriesCalculated
+	}
+	// searchQueryLimit cannot exceed `-search.maxUniqueTimeseries`
+	if api.maxUniqueTimeseries != 0 && searchQueryLimit > api.maxUniqueTimeseries {
+		searchQueryLimit = api.maxUniqueTimeseries
+	}
+	return searchQueryLimit
+}
+
+// blockIterator implements vmselectapi.BlockIterator
+type blockIterator struct {
+	sr storage.Search
+	mb storage.MetricBlock
+}
+
+var blockIteratorsPool sync.Pool
+
+func (bi *blockIterator) MustClose() {
+	bi.sr.MustClose()
+	bi.mb.MetricName = nil
+	bi.mb.Block.Reset()
+	blockIteratorsPool.Put(bi)
+}
+
+func getBlockIterator() *blockIterator {
+	v := blockIteratorsPool.Get()
+	if v == nil {
+		v = &blockIterator{}
+	}
+	return v.(*blockIterator)
+}
+
+func (bi *blockIterator) NextBlock(dst []byte) ([]byte, bool) {
+	if !bi.sr.NextMetricBlock() {
+		return dst, false
+	}
+	mb := bi.mb
+	mb.MetricName = bi.sr.MetricBlockRef.MetricName
+	bi.sr.MetricBlockRef.BlockRef.MustReadBlock(&mb.Block)
+	dst = mb.Marshal(dst[:0])
+	return dst, true
+}
+
+func (bi *blockIterator) Error() error {
+	return bi.sr.Error()
+}
+
+// SearchMetricNames returns metric names for the given tfss on the given tr.
+func (api *VMStorage) SearchMetricNames(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) ([]string, error) {
+	tr := sq.GetTimeRange()
+	maxMetrics := sq.MaxMetrics
+	if maxMetrics <= 0 {
+		// fallback to maxUniqueTimeSeries if no limit is provided,
+		// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7857
+		maxMetrics = api.maxUniqueTimeSeriesCalculated
+	}
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return nil, err
+	}
+	if len(tfss) == 0 {
+		return nil, fmt.Errorf("missing tag filters")
+	}
+	return api.s.SearchMetricNames(qt, tfss, tr, maxMetrics, deadline)
+}
+
+// SearchLabelValues searches for label values for the given labelName, tfss and
+// tr.
+func (api *VMStorage) LabelValues(qt *querytracer.Tracer, sq *storage.SearchQuery, labelName string, maxLabelValues int, deadline uint64) ([]string, error) {
+	tr := sq.GetTimeRange()
+	if maxLabelValues <= 0 || maxLabelValues > *maxTagValues {
+		maxLabelValues = *maxTagValues
+	}
+	maxMetrics := sq.MaxMetrics
+	if maxMetrics <= 0 {
+		// fallback to maxUniqueTimeSeries if no limit is provided,
+		// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7857
+		maxMetrics = api.maxUniqueTimeSeriesCalculated
+	}
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return nil, err
+	}
+	return api.s.SearchLabelValues(qt, labelName, tfss, tr, maxLabelValues, maxMetrics, deadline)
+}
+
+// TagValueSuffixes returns all the tag value suffixes for the given tagKey and
+// tagValuePrefix on the given tr.
+//
+// This allows implementing
+// https://graphite-api.readthedocs.io/en/latest/api.html#metrics-find or
+// similar APIs.
+func (api *VMStorage) TagValueSuffixes(qt *querytracer.Tracer, _, _ uint32, tr storage.TimeRange, tagKey, tagValuePrefix string, delimiter byte,
+	maxSuffixes int, deadline uint64) ([]string, error) {
+	if maxSuffixes <= 0 || maxSuffixes > *maxTagValueSuffixesPerSearch {
+		maxSuffixes = *maxTagValueSuffixesPerSearch
+	}
+	suffixes, err := api.s.SearchTagValueSuffixes(qt, tr, tagKey, tagValuePrefix, delimiter, maxSuffixes, deadline)
+	if err != nil {
+		return nil, err
+	}
+	if len(suffixes) >= maxSuffixes {
+		return nil, fmt.Errorf("more than -search.maxTagValueSuffixesPerSearch=%d suffixes returned; "+
+			"either narrow down the search or increase -search.maxTagValueSuffixesPerSearch command-line flag value", maxSuffixes)
+	}
+	return suffixes, nil
+}
+
+// SearchLabelNames searches for tag keys matching the given tfss on tr.
+func (api *VMStorage) LabelNames(qt *querytracer.Tracer, sq *storage.SearchQuery, maxLabelNames int, deadline uint64) ([]string, error) {
+	tr := sq.GetTimeRange()
+	if maxLabelNames <= 0 || maxLabelNames > *maxTagKeys {
+		maxLabelNames = *maxTagKeys
+	}
+	maxMetrics := sq.MaxMetrics
+	if maxMetrics <= 0 {
+		// fallback to maxUniqueTimeSeries if no limit is provided,
+		// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7857
+		maxMetrics = api.maxUniqueTimeSeriesCalculated
+	}
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return nil, err
+	}
+	return api.s.SearchLabelNames(qt, tfss, tr, maxLabelNames, maxMetrics, deadline)
+}
+
+func (api *VMStorage) SeriesCount(_ *querytracer.Tracer, _, _ uint32, deadline uint64) (uint64, error) {
+	return api.s.GetSeriesCount(deadline)
+}
+
+func (api *VMStorage) Tenants(_ *querytracer.Tracer, _ storage.TimeRange, _ uint64) ([]string, error) {
+	return nil, nil
+}
+
+// GetTSDBStatus returns TSDB status for given filters on the given date.
+func (api *VMStorage) TSDBStatus(qt *querytracer.Tracer, sq *storage.SearchQuery, focusLabel string, topN int, deadline uint64) (*storage.TSDBStatus, error) {
+	tr := sq.GetTimeRange()
+	maxMetrics := sq.MaxMetrics
+	if maxMetrics <= 0 {
+		// fallback to maxUniqueTimeSeries if no limit is provided,
+		// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7857
+		maxMetrics = api.maxUniqueTimeSeriesCalculated
+	}
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return nil, err
+	}
+	date := uint64(sq.MinTimestamp) / (24 * 3600 * 1000)
+	return api.s.GetTSDBStatus(qt, tfss, date, focusLabel, topN, maxMetrics, deadline)
+}
+
+// DeleteSeries deletes series matching tfss.
+//
+// Returns the number of deleted series.
+func (api *VMStorage) DeleteSeries(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) (int, error) {
+	// TODO(@rtm0): Return an error if the storage is in read-only mode?
+
+	tr := sq.GetTimeRange()
+	maxMetrics := sq.MaxMetrics
+	if maxMetrics <= 0 {
+		// fallback to maxUniqueTimeSeries if no limit is provided,
+		// see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7857
+		maxMetrics = api.maxUniqueTimeSeriesCalculated
+	}
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return 0, err
+	}
+	if len(tfss) == 0 {
+		return 0, fmt.Errorf("missing tag filters")
+	}
+	return api.s.DeleteSeries(qt, tfss, maxMetrics)
+}
+
+func (api *VMStorage) RegisterMetricNames(qt *querytracer.Tracer, mrs []storage.MetricRow, _ uint64) error {
+	// TODO(@rtm0): Return an error if the storage is in read-only mode?
+
+	api.s.RegisterMetricNames(qt, mrs)
+	return nil
+}
+
+// GetMetricNamesUsageStats returns metric name usage stats.
+func (api *VMStorage) GetMetricNamesUsageStats(qt *querytracer.Tracer, _ *storage.TenantToken, limit, le int, matchPattern string, _ uint64) (metricnamestats.StatsResult, error) {
+
+	return api.s.GetMetricNamesStats(qt, limit, le, matchPattern), nil
+}
+
+// ResetMetricNamesStats resets state for metric names usage tracker
+func (api *VMStorage) ResetMetricNamesUsageStats(qt *querytracer.Tracer, _ uint64) error {
+	api.s.ResetMetricNamesStats(qt)
+	return nil
+}
+
+func (api *VMStorage) setupTfss(qt *querytracer.Tracer, sq *storage.SearchQuery, tr storage.TimeRange, maxMetrics int, deadline uint64) ([]*storage.TagFilters, error) {
+	tfss := make([]*storage.TagFilters, 0, len(sq.TagFilterss))
+	for _, tagFilters := range sq.TagFilterss {
+		tfs := storage.NewTagFilters()
+		for i := range tagFilters {
+			tf := &tagFilters[i]
+			if string(tf.Key) == "__graphite__" {
+				query := tf.Value
+				qtChild := qt.NewChild("searching for series matching __graphite__=%q", query)
+				paths, err := api.s.SearchGraphitePaths(qtChild, tr, query, maxMetrics, deadline)
+				qtChild.Donef("found %d series", len(paths))
+				if err != nil {
+					return nil, fmt.Errorf("error when searching for Graphite paths for query %q: %w", query, err)
+				}
+				if len(paths) >= maxMetrics {
+					return nil, fmt.Errorf("more than %d time series match Graphite query %q; "+
+						"either narrow down the query or increase the corresponding -search.max* command-line flag value at vmselect nodes; "+
+						"see https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#resource-usage-limits", maxMetrics, query)
+				}
+				tfs.AddGraphiteQuery(query, paths, tf.IsNegative)
+				continue
+			}
+			if err := tfs.Add(tf.Key, tf.Value, tf.IsNegative, tf.IsRegexp); err != nil {
+				return nil, fmt.Errorf("cannot parse tag filter %s: %w", tf, err)
+			}
+		}
+		tfss = append(tfss, tfs)
+	}
+	return tfss, nil
+}
+
+func (api *VMStorage) GetMetadataRecords(qt *querytracer.Tracer, _ *storage.TenantToken, limit int, metricName string, _ uint64) ([]*metricsmetadata.Row, error) {
+	return api.s.GetMetadataRows(qt, limit, metricName), nil
+}
+
+// deleteSnapshot deletes a snapshot by its name.
+//
+// Callers must wrap the call with wg.Add(1)...wg.Done().
+func (api *VMStorage) deleteSnapshot(snapshotName string) error {
+	snapshots := api.s.MustListSnapshots()
+	for _, snName := range snapshots {
+		if snName == snapshotName {
+			if err := api.s.DeleteSnapshot(snName); err != nil {
+				return fmt.Errorf("cannot delete snapshot %q: %w", snName, err)
+			}
+			return nil
+		}
+	}
+	return fmt.Errorf("cannot find snapshot %q", snapshotName)
+}

app/vmstorage/vmstorage_single_node.go

@@ -0,0 +1,202 @@
+package vmstorage
+
+import (
+	"errors"
+	"sync"
+
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/querytracer"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage/metricnamestats"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/storage/metricsmetadata"
+	"github.com/VictoriaMetrics/VictoriaMetrics/lib/syncwg"
+)
+
+// newVMStorageSingleNode creates a new instance of of VMStorage for vmsingle.
+func newVMStorageSingleNode(s *storage.Storage, maxConcurrentRequests int, resetCacheIfNeeded func(mrs []storage.MetricRow)) *VMStorageSingleNode {
+	vms := newVMStorage(s, maxConcurrentRequests)
+	return &VMStorageSingleNode{
+		VMStorage:          vms,
+		wg:                 syncwg.WaitGroup{},
+		resetCacheIfNeeded: resetCacheIfNeeded,
+	}
+}
+
+type VMStorageSingleNode struct {
+	*VMStorage
+
+	// wg is used to wrap every storage call into wg.Add(1) ... wg.Done()
+	// for proper graceful shutdown when Stop is called.
+	//
+	// Use syncwg instead of sync, since Add is called from concurrent
+	// goroutines.
+	wg syncwg.WaitGroup
+
+	// resetCacheIfNeeded is a callback for automatic resetting of response
+	// cache if needed.
+	resetCacheIfNeeded func(mrs []storage.MetricRow)
+}
+
+func (api *VMStorageSingleNode) Stop() {
+	api.wg.WaitAndBlock()
+	api.VMStorage.Stop()
+}
+
+// WriteRows writes metric rows to the storage.
+//
+// Returns an error if the storage is in read-only mode.
+//
+// The caller should limit the number of concurrent calls to WriteRows() in
+// order to limit memory usage.
+func (api *VMStorageSingleNode) WriteRows(rows []storage.MetricRow) error {
+	api.wg.Add(1)
+	defer api.wg.Done()
+
+	if api.s.IsReadOnly() {
+		return errReadOnly
+	}
+	api.resetCacheIfNeeded(rows)
+	return api.VMStorage.WriteRows(rows)
+}
+
+// WriteMetadata writes metrics metadata to storage.
+//
+// Returns an error if the storage is in read-only mode.
+//
+// The caller should limit the number of concurrent calls to WriteMetadata() in
+// order to limit memory usage.
+func (api *VMStorageSingleNode) WriteMetadata(rows []metricsmetadata.Row) error {
+	api.wg.Add(1)
+	defer api.wg.Done()
+
+	if api.s.IsReadOnly() {
+		return errReadOnly
+	}
+	return api.VMStorage.WriteMetadata(rows)
+}
+
+var errReadOnly = errors.New("the storage is in read-only mode; check -storage.minFreeDiskSpaceBytes command-line flag value")
+
+func (api *VMStorageSingleNode) IsReadOnly() bool {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.IsReadOnly()
+}
+
+// GetSearch sets up an instance of storage search and returns it to the caller
+// along with the max series count that the search can return.
+//
+// This method is not part of the vmselectapi.API and must only be used by
+// vmsingle HTTP handlers.
+//
+// Callers of this method must call PutSearch() once the search instance is not
+// needed anymore.
+func (api *VMStorageSingleNode) GetSearch(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) (*storage.Search, int, error) {
+	api.wg.Add(1)
+
+	tr := sq.GetTimeRange()
+	maxMetrics := api.getMaxMetrics(sq.MaxMetrics)
+	tfss, err := api.setupTfss(qt, sq, tr, maxMetrics, deadline)
+	if err != nil {
+		return nil, 0, err
+	}
+
+	sr := getSearch()
+	maxSeriesCount := sr.Init(qt, api.s, tfss, tr, sq.MaxMetrics, deadline)
+	return sr, maxSeriesCount, nil
+}
+
+// PutSearch resets the search once it is not needed anymore and puts it aside
+// for future reuse.
+//
+// This method is not part of the vmselectapi.API and must only be used by
+// vmsingle HTTP handlers.
+//
+// The method must only be used on search instances that have been created with
+// GetSearch().
+func (api *VMStorageSingleNode) PutSearch(sr *storage.Search) {
+	api.wg.Done()
+	putSearch(sr)
+}
+
+func getSearch() *storage.Search {
+	v := ssPool.Get()
+	if v == nil {
+		return &storage.Search{}
+	}
+	return v.(*storage.Search)
+}
+
+func putSearch(sr *storage.Search) {
+	sr.MustClose()
+	ssPool.Put(sr)
+}
+
+var ssPool sync.Pool
+
+func (api *VMStorageSingleNode) SearchMetricNames(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) ([]string, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.SearchMetricNames(qt, sq, deadline)
+}
+
+func (api *VMStorageSingleNode) LabelValues(qt *querytracer.Tracer, sq *storage.SearchQuery, labelName string, maxLabelValues int, deadline uint64) ([]string, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.LabelValues(qt, sq, labelName, maxLabelValues, deadline)
+}
+
+func (api *VMStorageSingleNode) TagValueSuffixes(qt *querytracer.Tracer, accountID, projectID uint32, tr storage.TimeRange, tagKey, tagValuePrefix string, delimiter byte, maxSuffixes int, deadline uint64) ([]string, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.TagValueSuffixes(qt, accountID, projectID, tr, tagKey, tagValuePrefix, delimiter, maxSuffixes, deadline)
+}
+
+func (api *VMStorageSingleNode) LabelNames(qt *querytracer.Tracer, sq *storage.SearchQuery, maxLabelNames int, deadline uint64) ([]string, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.LabelNames(qt, sq, maxLabelNames, deadline)
+}
+
+func (api *VMStorageSingleNode) SeriesCount(qt *querytracer.Tracer, accountID, projectID uint32, deadline uint64) (uint64, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.SeriesCount(qt, accountID, projectID, deadline)
+}
+
+func (api *VMStorageSingleNode) TSDBStatus(qt *querytracer.Tracer, sq *storage.SearchQuery, focusLabel string, topN int, deadline uint64) (*storage.TSDBStatus, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.TSDBStatus(qt, sq, focusLabel, topN, deadline)
+}
+
+func (api *VMStorageSingleNode) DeleteSeries(qt *querytracer.Tracer, sq *storage.SearchQuery, deadline uint64) (int, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	// TODO(@rtm0): Return an error if the storage is in read-only mode?
+	return api.VMStorage.DeleteSeries(qt, sq, deadline)
+}
+
+func (api *VMStorageSingleNode) RegisterMetricNames(qt *querytracer.Tracer, mrs []storage.MetricRow, deadline uint64) error {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	// TODO(@rtm0): Return an error if the storage is in read-only mode?
+	return api.VMStorage.RegisterMetricNames(qt, mrs, deadline)
+}
+
+func (api *VMStorageSingleNode) GetMetricNamesUsageStats(qt *querytracer.Tracer, tt *storage.TenantToken, limit, le int, matchPattern string, deadline uint64) (metricnamestats.StatsResult, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.GetMetricNamesUsageStats(qt, tt, limit, le, matchPattern, deadline)
+}
+
+func (api *VMStorageSingleNode) ResetMetricNamesUsageStats(qt *querytracer.Tracer, deadline uint64) error {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.ResetMetricNamesUsageStats(qt, deadline)
+}
+
+func (api *VMStorageSingleNode) GetMetadataRecords(qt *querytracer.Tracer, tt *storage.TenantToken, limit int, metricName string, deadline uint64) ([]*metricsmetadata.Row, error) {
+	api.wg.Add(1)
+	defer api.wg.Done()
+	return api.VMStorage.GetMetadataRecords(qt, tt, limit, metricName, deadline)
+}

app/vmui/Dockerfile-web

@@ -1,4 +1,4 @@
-FROM golang:1.26.2 AS build-web-stage
+FROM golang:1.26.3 AS build-web-stage
 COPY build /build
 
 WORKDIR /build
@@ -6,7 +6,7 @@ COPY web/ /build/
 RUN GOOS=linux GOARCH=amd64 CGO_ENABLED=0 go build -o web-amd64 github.com/VictoriMetrics/vmui/ && \
     GOOS=windows GOARCH=amd64 CGO_ENABLED=0 go build -o web-windows github.com/VictoriMetrics/vmui/
 
-FROM alpine:3.23.3
+FROM alpine:3.23.4
 USER root
 
 COPY --from=build-web-stage /build/web-amd64 /app/web

app/vmui/packages/vmui/package.json

@@ -23,14 +23,14 @@
     "classnames": "^2.5.1",
     "dayjs": "^1.11.20",
     "lodash.debounce": "^4.0.8",
-    "marked": "^17.0.5",
-    "preact": "^10.29.0",
-    "qs": "^6.15.0",
+    "marked": "^18.0.2",
+    "preact": "^10.29.1",
+    "qs": "^6.15.1",
     "react-input-mask": "^2.0.4",
-    "react-router-dom": "^7.13.2",
+    "react-router-dom": "^7.14.1",
     "uplot": "^1.6.32",
-    "vite": "^8.0.7",
-    "web-vitals": "^5.1.0"
+    "vite": "^8.0.8",
+    "web-vitals": "^5.2.0"
   },
   "devDependencies": {
     "@eslint/eslintrc": "^3.3.5",
@@ -39,24 +39,24 @@
     "@testing-library/jest-dom": "^6.9.1",
     "@testing-library/preact": "^3.2.4",
     "@types/lodash.debounce": "^4.0.9",
-    "@types/node": "^25.5.0",
+    "@types/node": "^25.6.0",
     "@types/qs": "^6.15.0",
     "@types/react": "^19.2.14",
     "@types/react-input-mask": "^3.0.6",
     "@types/react-router-dom": "^5.3.3",
-    "@typescript-eslint/eslint-plugin": "^8.57.2",
-    "@typescript-eslint/parser": "^8.57.2",
+    "@typescript-eslint/eslint-plugin": "^8.58.2",
+    "@typescript-eslint/parser": "^8.58.2",
     "cross-env": "^10.1.0",
     "eslint": "^9.39.2",
     "eslint-plugin-react": "^7.37.5",
     "eslint-plugin-unused-imports": "^4.4.1",
-    "globals": "^17.4.0",
+    "globals": "^17.5.0",
     "http-proxy-middleware": "^3.0.5",
-    "jsdom": "^29.0.1",
-    "postcss": "^8.5.8",
-    "sass-embedded": "^1.98.0",
-    "typescript": "^5.9.3",
-    "vitest": "^4.1.1"
+    "jsdom": "^29.0.2",
+    "postcss": "^8.5.10",
+    "sass-embedded": "^1.99.0",
+    "typescript": "^6.0.2",
+    "vitest": "^4.1.4"
   },
   "browserslist": {
     "production": [

app/vmui/packages/vmui/src/api/types.ts

@@ -3,6 +3,7 @@ export interface MetricBase {
   metric: {
     [key: string]: string;
   };
+  nullTimestamps?: number[];
 }
 
 export interface MetricResult extends MetricBase {

app/vmui/packages/vmui/src/components/Chart/ChartTooltip/ChartTooltip.tsx

@@ -16,6 +16,7 @@ export interface ChartTooltipProps {
   point: { top: number, left: number };
   unit?: string;
   statsFormatted?: SeriesItemStatsFormatted;
+  description?: ReactNode;
   isSticky?: boolean;
   info?: ReactNode;
   marker?: string;
@@ -34,6 +35,7 @@ const ChartTooltip: FC<ChartTooltipProps> = ({
   unit = "",
   info,
   statsFormatted,
+  description,
   isSticky,
   marker,
   duplicateCount = 0,
@@ -173,6 +175,7 @@ const ChartTooltip: FC<ChartTooltipProps> = ({
           ))}
         </table>
       )}
+      {description && <p className="vm-chart-tooltip__description">{description}</p>}
       {info && <p className="vm-chart-tooltip__info">{info}</p>}
     </div>
   ), u.root);

app/vmui/packages/vmui/src/components/Chart/ChartTooltip/style.scss

@@ -143,4 +143,10 @@ $chart-tooltip-y: -1 * ($padding-global + $chart-tooltip-half-icon);
     word-break: break-all;
     white-space: pre-wrap;
   }
+
+  &__description {
+    word-break: break-word;
+    white-space: pre-wrap;
+    opacity: 0.85;
+  }
 }

app/vmui/packages/vmui/src/components/ExploreAlerts/BaseAlert/index.tsx

@@ -1,7 +1,7 @@
 import { useMemo } from "preact/compat";
 import "./style.scss";
-import { Alert as APIAlert } from "../../../types";
-import { createSearchParams } from "react-router-dom";
+import { Alert as APIAlert, Group } from "../../../types";
+import { Link } from "react-router-dom";
 import Button from "../../Main/Button/Button";
 import Badges, { BadgeColor } from "../Badges";
 import { formatEventTime } from "../helpers";
@@ -9,12 +9,14 @@ import {
   SearchIcon,
 } from "../../Main/Icons";
 import CodeExample from "../../Main/CodeExample/CodeExample";
+import router from "../../../router";
 
 interface BaseAlertProps {
   item: APIAlert;
+  group?: Group;
 }
 
-const BaseAlert = ({ item }: BaseAlertProps) => {
+const BaseAlert = ({ item, group }: BaseAlertProps) => {
   const query = item?.expression;
   const alertLabels = item?.labels || {};
   const alertLabelsItems = useMemo(() => {
@@ -24,13 +26,19 @@ const BaseAlert = ({ item }: BaseAlertProps) => {
     }]));
   }, [alertLabels]);
 
-  const openQueryLink = () => {
-    const params = {
+  const queryLink = useMemo(() => {
+    if (!group?.interval) return;
+
+    const params = new URLSearchParams({
       "g0.expr": query,
-      "g0.end_time": ""
-    };
-    window.open(`#/?${createSearchParams(params).toString()}`, "_blank", "noopener noreferrer");
-  };
+      "g0.end_time": item.activeAt,
+      // Interval is the Group's evaluation interval in float seconds as present in the file. See: /app/vmalert/rule/web.go
+      "g0.step_input": `${group.interval}s`,
+      "g0.relative_time": "none",
+    });
+
+    return `${router.home}?${params.toString()}`;
+  }, [query, item.activeAt, group?.interval]);
 
   return (
     <div className="vm-explore-alerts-alert-item">
@@ -45,15 +53,22 @@ const BaseAlert = ({ item }: BaseAlertProps) => {
               style={{ "text-align": "end" }}
               colSpan={2}
             >
-              <Button
-                size="small"
-                variant="outlined"
-                color="gray"
-                startIcon={<SearchIcon />}
-                onClick={openQueryLink}
-              >
-                <span className="vm-button-text">Run query</span>
-              </Button>
+              {queryLink && (
+                <Link
+                  to={queryLink}
+                  target={"_blank"}
+                  rel="noreferrer"
+                >
+                  <Button
+                    size="small"
+                    variant="outlined"
+                    color="gray"
+                    startIcon={<SearchIcon />}
+                  >
+                    <span className="vm-button-text">Run query</span>
+                  </Button>
+                </Link>
+              )}
             </td>
           </tr>
           <tr>

app/vmui/packages/vmui/src/components/ExploreAlerts/BaseRule/index.tsx

@@ -1,19 +1,21 @@
 import { useMemo } from "preact/compat";
 import "./style.scss";
-import { Rule as APIRule } from "../../../types";
-import { useNavigate, createSearchParams } from "react-router-dom";
+import { Group, Rule as APIRule } from "../../../types";
+import { useNavigate, Link } from "react-router-dom";
 import { SearchIcon, DetailsIcon } from "../../Main/Icons";
 import Button from "../../Main/Button/Button";
 import Alert from "../../Main/Alert/Alert";
 import Badges, { BadgeColor } from "../Badges";
 import { formatDuration, formatEventTime } from "../helpers";
 import CodeExample from "../../Main/CodeExample/CodeExample";
+import router from "../../../router";
 
 interface BaseRuleProps {
   item: APIRule;
+  group?: Group;
 }
 
-const BaseRule = ({ item }: BaseRuleProps) => {
+const BaseRule = ({ item, group }: BaseRuleProps) => {
   const query = item?.query;
   const navigate = useNavigate();
   const openAlertLink = (id: string) => {
@@ -33,13 +35,19 @@ const BaseRule = ({ item }: BaseRuleProps) => {
     }]));
   }, [ruleLabels]);
 
-  const openQueryLink = () => {
-    const params = {
+  const queryLink = useMemo(() => {
+    if (!group?.interval) return;
+
+    const params = new URLSearchParams({
       "g0.expr": query,
-      "g0.end_time": ""
-    };
-    window.open(`#/?${createSearchParams(params).toString()}`, "_blank", "noopener noreferrer");
-  };
+      "g0.end_time": item.lastEvaluation,
+      // Interval is the Group's evaluation interval in float seconds as present in the file. See: /app/vmalert/rule/web.go
+      "g0.step_input": `${group.interval}s`,
+      "g0.relative_time": "none",
+    });
+
+    return `${router.home}?${params.toString()}`;
+  }, [query, item.lastEvaluation, group?.interval]);
 
   return (
     <div className="vm-explore-alerts-rule-item">
@@ -54,15 +62,22 @@ const BaseRule = ({ item }: BaseRuleProps) => {
               style={{ "text-align": "end" }}
               colSpan={2}
             >
-              <Button
-                size="small"
-                variant="outlined"
-                color="gray"
-                startIcon={<SearchIcon />}
-                onClick={openQueryLink}
-              >
-                <span className="vm-button-text">Run query</span>
-              </Button>
+              {queryLink && (
+                <Link
+                  to={queryLink}
+                  target={"_blank"}
+                  rel="noreferrer"
+                >
+                  <Button
+                    size="small"
+                    variant="outlined"
+                    color="gray"
+                    startIcon={<SearchIcon />}
+                  >
+                    <span className="vm-button-text">Run query</span>
+                  </Button>
+                </Link>
+              )}
             </td>
           </tr>
           <tr>

app/vmui/packages/vmui/src/components/ExploreAlerts/Rule/index.tsx

@@ -2,15 +2,16 @@ import { FC } from "preact/compat";
 import ItemHeader from "../ItemHeader";
 import Accordion from "../../Main/Accordion/Accordion";
 import "./style.scss";
-import { Rule as APIRule } from "../../../types";
+import { Group, Rule as APIRule } from "../../../types";
 import BaseRule from "../BaseRule";
 
 interface RuleProps {
   states: Record<string, number>;
   rule: APIRule;
+  group: Group;
 }
 
-const Rule: FC<RuleProps> = ({ states, rule }) => {
+const Rule: FC<RuleProps> = ({ states, rule, group }) => {
   const state = Object.keys(states).length > 0 ? Object.keys(states)[0] : "ok";
   return (
     <div className={`vm-explore-alerts-rule vm-badge-item ${state.replace(" ", "-")}`}>
@@ -25,7 +26,10 @@ const Rule: FC<RuleProps> = ({ states, rule }) => {
           name={rule.name}
         />}
       >
-        <BaseRule item={rule} />
+        <BaseRule
+          item={rule}
+          group={group}
+        />
       </Accordion>
     </div>
   );

app/vmui/packages/vmui/src/components/ExploreAlerts/RulesHeader/index.tsx

@@ -50,7 +50,6 @@ const RulesHeader = ({
             label="Rule type"
             placeholder="Please select rule type"
             onChange={onChangeRuleType}
-            autofocus={!!types.length && !isMobile}
             includeAll
             searchable
           />

app/vmui/packages/vmui/src/components/ExploreAlerts/helpers.ts

@@ -17,7 +17,7 @@ export const formatDuration = (raw: number) => {
 
 export const formatEventTime = (raw: string) => {
   const t = dayjs(raw);
-  return t.year() <= 1 ? "Never" : t.format("DD MMM YYYY HH:mm:ss");
+  return t.year() <= 1 ? "Never" : t.tz().format("DD MMM YYYY HH:mm:ss");
 };
 
 export const getStates = (rule: Rule) => {

app/vmui/packages/vmui/src/hooks/uplot/useLineTooltip.ts

@@ -4,7 +4,7 @@ import { ChartTooltipProps } from "../../components/Chart/ChartTooltip/ChartTool
 import { SeriesItem } from "../../types";
 import dayjs from "dayjs";
 import { DATE_FULL_TIMEZONE_FORMAT } from "../../constants/date";
-import { formatPrettyNumber, getMetricName } from "../../utils/uplot";
+import { getMetricName } from "../../utils/uplot";
 import { MetricResult } from "../../api/types";
 import useEventListener from "../useEventListener";
 
@@ -15,19 +15,65 @@ interface LineTooltipHook {
   unit?: string;
 }
 
+// Pixel proximity for detecting hover over null-timestamp X markers drawn at chart bottom.
+const NULL_HOVER_PROX = 8;
+// Half the visual marker height in CSS px (BASE_POINT_SIZE * 1.4 / 2 from scatter.ts).
+// scatter.ts lifts the marker center by this amount above yMin so the icon sits inside
+// the plot area; the hover y-anchor must match that offset.
+const NULL_MARKER_HALF_CSS = 2.8;
+
+interface NullHover {
+  seriesIdx: number;
+  timestamp: number;
+}
+
+const findNullHover = (u: uPlot): NullHover | null => {
+  const cursorLeft = u.cursor.left ?? -1;
+  const cursorTop = u.cursor.top ?? -1;
+  if (cursorLeft < 0 || cursorTop < 0) return null;
+
+  const scaleY = u.scales["1"];
+  if (!scaleY || scaleY.min == null) return null;
+  const yPos = u.valToPos(scaleY.min, "1") - NULL_MARKER_HALF_CSS;
+  if (Math.abs(cursorTop - yPos) > NULL_HOVER_PROX) return null;
+
+  let best: { seriesIdx: number; timestamp: number; dist: number } | null = null;
+  for (let s = 1; s < u.series.length; s++) {
+    const seriesItem = u.series[s] as SeriesItem;
+    if (!seriesItem.show) continue;
+    const nullTs = seriesItem.nullTimestamps;
+    if (!nullTs || !nullTs.length) continue;
+
+    for (let i = 0; i < nullTs.length; i++) {
+      const t = nullTs[i];
+      const xPos = u.valToPos(t, "x");
+      const dist = Math.abs(cursorLeft - xPos);
+      if (dist < NULL_HOVER_PROX && (best === null || dist < best.dist)) {
+        best = { seriesIdx: s, timestamp: t, dist };
+      }
+    }
+  }
+  return best ? { seriesIdx: best.seriesIdx, timestamp: best.timestamp } : null;
+};
+
+const NULL_DESCRIPTION = "\"null\" can be a staleness marker or an actual NaN/null value produced by exporter.";
+
 const useLineTooltip = ({ u, metrics, series, unit }: LineTooltipHook) => {
   const [showTooltip, setShowTooltip] = useState(false);
   const [tooltipIdx, setTooltipIdx] = useState({ seriesIdx: -1, dataIdx: -1 });
+  const [nullTooltip, setNullTooltip] = useState<NullHover | null>(null);
   const [stickyTooltips, setStickyToolTips] = useState<ChartTooltipProps[]>([]);
 
   const resetTooltips = () => {
     setStickyToolTips([]);
     setTooltipIdx({ seriesIdx: -1, dataIdx: -1 });
+    setNullTooltip(null);
   };
 
   const setCursor = (u: uPlot) => {
     const dataIdx = u.cursor.idx ?? -1;
     setTooltipIdx(prev => ({ ...prev, dataIdx }));
+    setNullTooltip(findNullHover(u));
   };
 
   const seriesFocus = (u: uPlot, sidx: (number | null)) => {
@@ -35,7 +81,36 @@ const useLineTooltip = ({ u, metrics, series, unit }: LineTooltipHook) => {
     setTooltipIdx(prev => ({ ...prev, seriesIdx }));
   };
 
+  const getNullTooltipProps = (hit: NullHover): ChartTooltipProps => {
+    const { seriesIdx, timestamp } = hit;
+    const metricItem = metrics[seriesIdx - 1];
+    const seriesItem = series[seriesIdx] as SeriesItem;
+
+    const groups = new Set(metrics.map(m => m.group));
+    const group = metricItem?.group || 0;
+
+    const yMin = u?.scales?.[1]?.min ?? 0;
+    const point = {
+      top: u ? u.valToPos(yMin, seriesItem?.scale || "1") - NULL_MARKER_HALF_CSS : 0,
+      left: u ? u.valToPos(timestamp, "x") : 0,
+    };
+
+    return {
+      u,
+      id: `null_${seriesIdx}_${timestamp}`,
+      title: groups.size > 1 ? `Query ${group}` : "",
+      dates: [dayjs(timestamp * 1000).tz().format(DATE_FULL_TIMEZONE_FORMAT)],
+      value: "null",
+      info: getMetricName(metricItem, seriesItem),
+      description: NULL_DESCRIPTION,
+      marker: `${seriesItem?.stroke}`,
+      point,
+    };
+  };
+
   const getTooltipProps = useCallback((): ChartTooltipProps => {
+    if (nullTooltip) return getNullTooltipProps(nullTooltip);
+
     const { seriesIdx, dataIdx } = tooltipIdx;
     const metricItem = metrics[seriesIdx - 1];
     const seriesItem = series[seriesIdx] as SeriesItem;
@@ -44,8 +119,6 @@ const useLineTooltip = ({ u, metrics, series, unit }: LineTooltipHook) => {
     const group = metricItem?.group || 0;
 
     const value = u?.data?.[seriesIdx]?.[dataIdx] || 0;
-    const min = u?.scales?.[1]?.min || 0;
-    const max = u?.scales?.[1]?.max || 1;
     const date = u?.data?.[0]?.[dataIdx] || 0;
 
     let duplicateCount = 1;
@@ -80,13 +153,13 @@ const useLineTooltip = ({ u, metrics, series, unit }: LineTooltipHook) => {
       id: `${seriesIdx}_${dataIdx}`,
       title: groups.size > 1 ? `Query ${group}` : "",
       dates: [date ? dayjs(date * 1000).tz().format(DATE_FULL_TIMEZONE_FORMAT) : "-"],
-      value: formatPrettyNumber(value, min, max),
+      value: value.toLocaleString("en-US", { maximumFractionDigits: 20 }),
       info: getMetricName(metricItem, seriesItem),
       statsFormatted: seriesItem?.statsFormatted,
       marker: `${seriesItem?.stroke}`,
       duplicateCount,
     };
-  }, [u, tooltipIdx, metrics, series, unit]);
+  }, [u, tooltipIdx, metrics, series, unit, nullTooltip]);
 
   const handleClick = useCallback(() => {
     if (!showTooltip) return;
@@ -101,8 +174,9 @@ const useLineTooltip = ({ u, metrics, series, unit }: LineTooltipHook) => {
   };
 
   useEffect(() => {
-    setShowTooltip(tooltipIdx.dataIdx !== -1 && tooltipIdx.seriesIdx !== -1);
-  }, [tooltipIdx]);
+    const normalHit = tooltipIdx.dataIdx !== -1 && tooltipIdx.seriesIdx !== -1;
+    setShowTooltip(normalHit || nullTooltip !== null);
+  }, [tooltipIdx, nullTooltip]);
 
   useEventListener("click", handleClick);
 

app/vmui/packages/vmui/src/pages/CustomPanel/utils.test.ts

@@ -20,7 +20,7 @@ describe("convertMetricsDataToCSV", () => {
       },
     ];
     const result = convertMetricsDataToCSV(data);
-    expect(result).toBe("header1,header2\n123,value2");
+    expect(result).toBe("header1,header2,__timestamp__,__value__\n123,value2,1623945600,123");
   });
 
   it("should return a valid CSV string for multiple metric entries with values", () => {
@@ -43,7 +43,7 @@ describe("convertMetricsDataToCSV", () => {
       },
     ];
     const result = convertMetricsDataToCSV(data);
-    expect(result).toBe("header1,header2\n123,value2\n456,value4");
+    expect(result).toBe("header1,header2,__timestamp__,__value__\n123,value2,1623945600,123\n456,value4,1623949200,456");
   });
 
   it("should handle metric entries with multiple values field", () => {
@@ -58,7 +58,7 @@ describe("convertMetricsDataToCSV", () => {
       },
     ];
     const result = convertMetricsDataToCSV(data);
-    expect(result).toBe("header1,header2\n123-456,values");
+    expect(result).toBe("header1,header2,__timestamp__,__value__\n123-456,values,-,-");
   });
 
   it("should handle a combination of metric entries with value and values", () => {
@@ -81,6 +81,19 @@ describe("convertMetricsDataToCSV", () => {
       },
     ];
     const result = convertMetricsDataToCSV(data);
-    expect(result).toBe("header1,header2\n123,first\n456-789,second");
+    expect(result).toBe("header1,header2,__timestamp__,__value__\n123,first,1623945600,123\n456-789,second,-,-");
   });
+
+  it("should return value and timestamp if metric field is empty", () => {
+    const data: InstantMetricResult[] = [
+      {
+        value: [1623945600, "123"],
+        group: 0,
+        metric: {}
+      },
+    ];
+    const result = convertMetricsDataToCSV(data);
+    expect(result).toBe("__timestamp__,__value__\n1623945600,123");
+  });
+
 });

app/vmui/packages/vmui/src/pages/CustomPanel/utils.ts

@@ -3,16 +3,22 @@ import { getColumns, MetricCategory } from "../../hooks/useSortedCategories";
 import { formatValueToCSV } from "../../utils/csv";
 
 const getHeaders = (data: InstantMetricResult[]): string => {
-  return getColumns(data).map(({ key }) => key).join(",");
+  const metricHeaders = getColumns(data).map(({ key }) => key);
+  return [...metricHeaders, "__timestamp__", "__value__"].join(",");
 };
 
 const getRows = (data: InstantMetricResult[], headers: MetricCategory[]) => {
-  return data?.map(d => headers.map(c => formatValueToCSV(d.metric[c.key] || "-")).join(","));
+  return data?.map(d => {
+    const metricPart = headers.map(c => formatValueToCSV(d.metric[c.key] || "-"));
+    const timestamp = d.value ? formatValueToCSV(String(d.value[0])) : "-";
+    const value = d.value ? formatValueToCSV(d.value[1]) : "-";
+    return [...metricPart, timestamp, value].join(",");
+  });
 };
 
 export const convertMetricsDataToCSV = (data: InstantMetricResult[]): string => {
+  if (!data.length) return "";
   const headers = getHeaders(data);
-  if (!headers.length) return "";
   const rows = getRows(data, getColumns(data));
   return [headers, ...rows].join("\n");
 };

app/vmui/packages/vmui/src/pages/ExploreAlerts/ExploreAlert.tsx

@@ -2,10 +2,11 @@ import Spinner from "../../components/Main/Spinner/Spinner";
 import Alert from "../../components/Main/Alert/Alert";
 import { useFetchItem } from "./hooks/useFetchItem";
 import "./style.scss";
-import { Alert as APIAlert } from "../../types";
+import { Alert as APIAlert, Group as APIGroup } from "../../types";
 import ItemHeader from "../../components/ExploreAlerts/ItemHeader";
 import BaseAlert from "../../components/ExploreAlerts/BaseAlert";
 import Modal from "../../components/Main/Modal/Modal";
+import { useFetchGroup } from "./hooks/useFetchGroup";
 
 interface ExploreAlertProps {
   groupId: string;
@@ -17,10 +18,19 @@ interface ExploreAlertProps {
 const ExploreAlert = ({ groupId, id, mode, onClose }: ExploreAlertProps) => {
   const {
     item,
-    isLoading,
-    error,
+    isLoading: isLoadingItem,
+    error: errorItem,
   } = useFetchItem<APIAlert>({ groupId, id, mode });
 
+  const {
+    group,
+    isLoading: isLoadingGroup,
+    error: errorGroup,
+  } = useFetchGroup<APIGroup>({ id: groupId });
+
+  const error =  errorItem || errorGroup;
+  const isLoading = isLoadingItem || isLoadingGroup;
+
   if (isLoading) return (
     <Spinner />
   );
@@ -51,7 +61,12 @@ const ExploreAlert = ({ groupId, id, mode, onClose }: ExploreAlertProps) => {
       onClose={onClose}
     >
       <div className="vm-explore-alerts">
-        {item && (<BaseAlert item={item} />) || (
+        {item ? (
+          <BaseAlert
+            item={item}
+            group={group}
+          />
+        ) : (
           <Alert variant="info">{noItemFound}</Alert>
         )}
       </div>

app/vmui/packages/vmui/src/pages/ExploreAlerts/ExploreRule.tsx

@@ -2,11 +2,12 @@ import Spinner from "../../components/Main/Spinner/Spinner";
 import Alert from "../../components/Main/Alert/Alert";
 import { useFetchItem } from "./hooks/useFetchItem";
 import "./style.scss";
-import { Rule as APIRule } from "../../types";
+import { Group as APIGroup, Rule as APIRule } from "../../types";
 import ItemHeader from "../../components/ExploreAlerts/ItemHeader";
 import BaseRule from "../../components/ExploreAlerts/BaseRule";
 import Modal from "../../components/Main/Modal/Modal";
 import { getStates } from "../../components/ExploreAlerts/helpers";
+import { useFetchGroup } from "./hooks/useFetchGroup";
 
 interface ExploreRuleProps {
   groupId: string;
@@ -18,10 +19,19 @@ interface ExploreRuleProps {
 const ExploreRule = ({ groupId, id, mode, onClose }: ExploreRuleProps) => {
   const {
     item,
-    isLoading,
-    error,
+    isLoading: isLoadingItem,
+    error: errorItem,
   } = useFetchItem<APIRule>({ groupId, id, mode });
 
+  const {
+    group,
+    isLoading: isLoadingGroup,
+    error: errorGroup,
+  } = useFetchGroup<APIGroup>({ id: groupId });
+
+  const error =  errorItem || errorGroup;
+  const isLoading = isLoadingItem || isLoadingGroup;
+
   if (isLoading) return (
     <Spinner />
   );
@@ -49,7 +59,12 @@ const ExploreRule = ({ groupId, id, mode, onClose }: ExploreRuleProps) => {
       onClose={onClose}
     >
       <div className="vm-explore-alerts">
-        {item && (<BaseRule item={item} />) || (
+        {item ? (
+          <BaseRule
+            item={item}
+            group={group}
+          />
+        ) : (
           <Alert variant="info">{noItemFound}</Alert>
         )}
       </div>

app/vmui/packages/vmui/src/pages/ExploreAlerts/ExploreRules.tsx

@@ -132,7 +132,7 @@ const ExploreRules: FC = () => {
     newParams.set("page_num", "1");
     setSearchParams(newParams);
     const changes = getChanges(title, states);
-    setStates(changes.length == allStates.length ? [] : changes);
+    setStates(changes.length === allStates.length ? [] : changes);
   }, [states, searchParams]);
 
   const handleChangeRuleType = useCallback((title: string) => {
@@ -186,6 +186,7 @@ const ExploreRules: FC = () => {
                         <Rule
                           key={`rule-${rule.id}`}
                           rule={rule}
+                          group={group}
                           states={getStates(rule)}
                         />
                       ))}