This reverts commit ccf97a4143.
reason for revert: this change may break tests, which expect that ServesMetrics.GetMetric() fails
when the given metric doesn't exist in the output.
It is better to add 'TryGetMetric() (float64, bool)' function, which would return '(0, false)'
when the given metric doesn't exist, so the caller could decide what to do next.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9773
Add ad-hoc filters to query stats and operator dashboards.
These filters are useful for exploring non-uniform metrics sets
without distinct job/instance filters.
The previous text didn't contain links to vmagent's capabilities.
Instead, it contained misleading multitenancy-mode link that doesn't
seem to be related to the subject.
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Previously, `GetMetric` do `t.Fatalf` immediately when the target metric
not exist in `/metrics` page.
However, some metrics may start to appear after the process has been
running for a while. `t.Fatalf` invalidates the retry mechanism of
assertions, if the metric is not found the first time, the test case
will terminate.
This commit request changes `t.Fatalf` to `t.Logf` (instead of `t.Errorf`,
because error output may be considered a test case failure in some
scenarios).
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9773
Follow-up for cea9505bab
fastcache.Cache allocates off-heap memory, which must be explicitly
returned back to the pool with Reset method call.
After changed made at commit above, during cache transit from whole to
split mode, it's possible that current cache is referenced by Cache.Get
or Cache.Call atomic pointers. It leads to potential memory leaks, since
we don't have any memory synchronization for atomic.Pointer.Store calls.
This commit adds `Finalizer` to the `fastcache.Cache` instances.
It properly releases memory, when cache is no reachable.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9769
Previously, cache state transition from split into whole could left
cache into broken state, if Reset cache method was called in switching
mode.
Also, cache Reset didn't start background workers and didn't change
cache size.
This commit properly check mode during cache transition. In addition,
it no longer stops background workers after whole mode transition and
always start workers during start-up.
Access to the prev, curr and mode Cache fields are properly locked
in order to mitigate possible race conditions.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9769
It seems like go compilator skipped computations and allocations for samples
as they weren't used afterwards. Sinking results into global variable removes
this optimizations and benchmark starts showing allocations within `pushSamples` fn.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
- Make SearchTSIDs look similar to SearchMetricNames, i.e. search for metricIDs within the method
- Make the corresponding corrupted index test look similar to one for metric names search
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
### Describe Your Changes
Consistently use the `v0.0.0-YYYYMMDDHHMMSS-commit_hash` reference for
the internal deps such as `github.com/VictoriaMetrics/VictoriaMetrics`
dependency, since it allows referring any commit without waiting for the
release tag.
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
### Describe Your Changes
This pull request consists of the following:
1. Markdown fixes
following https://www.markdownguide.org/basic-syntax/
and https://github.com/markdownlint/markdownlint/blob/main/docs/RULES.md
- Add empty lines after headers or lists
- Remove extra lines between paragraphs
- Remove extra spaces at the end of a line
- Add language to code quote
- Consistent list (dont mix astrixes and dashes on same file, choose one
and be consistent in the same file)
- Proper URL links
- Use meaningful context to URLs instead of "here".
2. Concise language
3. Grammar fixes
- removing extra spaces between words
- there are multiple ones but i picked the basic ones that triggered my
eye :)
4. Spelling fixes
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
---------
Co-authored-by: hagen1778 <roman@victoriametrics.com>
add ability to limit available in datePicker dates using `minDate` and
`maxDate` parameters. all dates before `minDate` and after `maxDate`
cannot be picked. lower and upper bounds can be set independently.
This `minDate` and `maxDate` parameters aren't set by default in vmui.
The datepicker component with these params is re-used elsewhere.
The change also explciitly mentions `out-of-order` phrase, as it is commonly
used in Prometheus ecosystem.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Searching metricName by metricID happens many times during a single API
call. This requires getting the current set of idbs before those calls
happen. Which is fine but requires propagating idbs across the code
base. This is also fine in case of OSS version as it is used in Search
only.
Propagating idbs across the code base becomes a problem in Enterprise
version as it is used in at least 3 places. As a result it becomes very
difficult to merge things from OSS to Ent.
Localizing the all the dependencies in one searchMetricName type and
reusing this type everywhere should make things simpler.
Related enterprise changes:
https://github.com/VictoriaMetrics/VictoriaMetrics-enterprise/compare/search-metric-name-ent?expand=1
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9756
A small refactoring that reduces Search dependency on Storage:
- Move searchTSIDs() from Search to Storage because this method does not
depend on anything Search-specific but does depend on Storage.
- Use metricsTracker instead of storage.metricTracker.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9754
Benchmarking storage search api requires taking into account many
parameters, such as:
- data configuration: how many series, deleted series, search time range
- where the index data recides: prev and or indexDB
- which search operation to measure
While adding a new benchmark use case involves a lot boilerplate code.
This pr implements a framework for testing storage search ops that can
be relatively easily extended. This come in expecially handy when adding
new cases for parition index.
The current set of params will result of a lot of benchmarks to be run
which most probably does not make sense because:
- it will take a lot of time and
- the output data is hard to compare manually.
However, these benchmarks are very useful when only small set of params
is of interest. For example, if I want to compare the search of 100k
metric names when the index data resides in prevOnly, currOnly or
prevAndCurr indexDBs. This would translate in the following cmd:
```shell
go test ./lib/storage --loggerLevel=ERROR -run=^$ -bench=^BenchmarkSearch/MetricNames/.*/VariableSeries/100000$
```
Why this change:
- I often need to run benchmarks with configs that I did not have
before, requires either modifying the existing one or writing a new one.
It is easy to get lost and make benchmark non-comparable
- I need some way to make legacy and pt index benchmarks comparable
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
- state that it is unsafe to use lifecycle rules and describe the reason
- update formatting according latest changes in docs
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Co-authored-by: Max Kotliar <mkotlyar@victoriametrics.com>
### Describe Your Changes
This pull request consists of the following:
1. Markdown fixes
following https://www.markdownguide.org/basic-syntax/
and https://github.com/markdownlint/markdownlint/blob/main/docs/RULES.md
- Add empty lines after headers or lists
- Remove extra lines between paragraphs
- Remove extra spaces at the end of a line
- Add language to code quote
- Consistent list (dont mix astrixes and dashes on same file, choose one
and be consistent in the same file)
- Proper URL links
- Use meaningful context to URLs instead of "here".
2. Concise language
3. Grammar fixes
- removing extra spaces between words
- there are multiple ones but i picked the basic ones that triggered my
eye :)
4. Spelling fixes
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
- Rename copyStream to copyStreamToClient in order to make it more clear
that the stream must be copied from backend to client.
- Make sure that the client implements net/http.Flusher interface.
It is a programming error (BUG) if the client passed to copyStreamToClient
doesn't implement net/http.Flusher interface.
- Do not write zero-length data to the backend.
Updates https://github.com/VictoriaMetrics/VictoriaLogs/issues/667
1beb629b removed logic which was used in order to keep full backup
location path in the restore mark file. Because of this, backups created
with a shortname (e.g. `vmbackupmanager restore create
daily/2025-09-12`) will fail as backup location is not prepended.
Fix that by properly constructing full backup name from parsed canonical
values.
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Previously, vmagent always set enable.auto.commit to false and manually
commited messages. It adds additional pressure to the kafka brokers and could slow down
data consumption.
This commit allows vmagent to skip manual commit and use auto-commit
based on provided configuration. Which may improve message read throughput.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics-enterprise/pull/931
### Describe Your Changes
This PR introduces a `make docs-update-flags` command that updates flags
in the documentation using the actual binaries compiled from the latest
`enterprise-single-node` and `enterprise-cluster` branches (hardcoded
for now). The command also normalizes the output format.
It can be run from any branch. All work happens inside temporary
directories under /tmp. The script checks out the required branch,
builds the binaries, and updates the documentation. The current Git
repository is not touched.
The command adjusts default values to more meaningful ones, such as
changing `-maxConcurrentInserts` (default 20) to (default
2*cgroup.AvailableCPUs()).
Currently the logic is implemented only for vminsert, vmstorage,
vmselect, vmagent, vmalert, and victoria-metrics (aka single).
The goal is to make it easy to keep documentation synchronized with real
binaries
_**Note:** Please ignore xxx_flags.md files for now. Review flags in
`README.md` and `Cluster-VictoriaMetrics.md`, and `vmagent.md`,
`vmalert.md` only. Once we agree on the changes in those files, I'll
replace the flags with the `{{% content "xxx_flags.md" %}}`._
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
* stress on requirement to have empty destination folder for copying;
* remove extra verbosity from docs;
* remove list vmctl migration options as they became unsynced. Instead of syncing,
refer to the vmctl docs;
* fix typos.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
The application version can be then displayed in the vmui. Showing the
application version in vmui should make it easier to determine currently
used VM version (at least vmselect version).
------------
@Loori-R it would be could to add the app version in vmui in a follow-up
PR or by pushing a commit to this branch.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
`Table` component:
- add `format` property for table column, which allows to apply custom
formatting depending on column type
- add `rowClasses` table property, that allows to pass function that
allows to customize row css class depending on row value
- add `rowAction` table property, that allows to execute action while
clicking on table row
`Popper` component:
- add `classes` to specify additional CSS classes for popper to
differentiate from other poppers, since it's mounted to a DOM root
`Switch` component:
- use gap instead of left-margin
`DateTimeInput` component:
- add `dateOnly` property to allow accepting only date in the input
additional fixes:
- fix TopQuery header fields alignment
<img width="1279" height="125" alt="image"
src="https://github.com/user-attachments/assets/08ad4dbc-19e5-47f5-9ccd-a9fb222335a4"
/>
### Describe Your Changes
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/).
### Describe Your Changes
Set rateEnabled to false for probe_success in VMUI
Fix issue:
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9655
Problem:
probe_success is incorrectly initialized with rateEnabled = true because
the regex detecting counters (/_sum?|_total?|_count?/) matches partial
strings like _su. This causes probe_success (a gauge) to be treated as a
counter, producing slightly misleading graphs. For example, when
rateEnabled is set to true, probe_success often shows as 0 in VMUI when
the probe is actually succeding.
It is not intuative for users to have to disable rateEnabled manually
just to get the correct value for probe_success in VMUI.
Solution:
Update the regex to strictly match suffixes:
`/_sum$|_total$|_count$/`
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
Signed-off-by: William Wren <william.wren@ericsson.com>
### Describe Your Changes
This pull request consists of the following:
1. Markdown fixes
following https://www.markdownguide.org/basic-syntax/
and https://github.com/markdownlint/markdownlint/blob/main/docs/RULES.md
- Add empty lines after headers or lists
- Remove extra lines between paragraphs
- Remove extra spaces at the end of a line
- Add language to code quote
- Consistent list (dont mix astrixes and dashes on same file, choose one
and be consistent in the same file)
- Proper URL links
- Use meaningful context to URLs instead of "here".
2. Concise language
3. Grammar fixes
- removing extra spaces between words
- there are multiple ones but i picked the basic ones that triggered my
eye :)
4. Spelling fixes
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
This is needed for VictoriaLogs, which allows limiting query results with the given set of extra filters
specified via extra_filters query arg. The request url can contain multiple extra_filters query args -
they are all applied with AND logic to the query. See https://docs.victoriametrics.com/victorialogs/querying/#extra-filters
The merge_query_args option at vmauth allows merging the extra_filters provided by the client
(such as Grafana plugin for VictoriaLogs or built-in web UI) with the extra_filters specified in the backend
url at vmauth config.
This is needed for https://github.com/VictoriaMetrics/VictoriaLogs/issues/106
`workingsetcache` is built on top of two
[fastcache](https://github.com/VictoriaMetrics/fastcache) instances
(curr and prev) that are rotated periodically (configurable via
`-cacheExpireDuration` flag). During the rotation curr becomes prev and
prev is discarded, new curr is an empty. If an entry is not found in
curr then the prev cache is checked, and if the entry is found there it
is copied to curr.
`workingsetcache` also exports metrics, such as `EntriesCount`,
`GetCalls`, `SetCalls`, and `Misses` counts. These metrics are currently
implemented as the sum of the same metrics in prev and curr `fastcache`
instances. Given to rotation logic, these counts can be incorrect:
1. `EntriesCount`. It is the sum of prev and curr entry counts. If an
entry is not found in curr and found in prev (and therefore is copied
from prev to curr) the resulting entry count will be incorrect, i.e. it
will count copied entries two times.
2. `GetCalls`. It is the sum of prev and curr get calls. If an entry is
not found in curr the logic will attempt to retrieve it from prev, which
will result in double counting. While it is actually one get call to
`workingsetcache`.
3. `SetCalls`. It is the sum of prev and curr get calls. If an entry is
not found in curr but found in prev it will be copied to curr resulting
in a set call to curr. While from the `workingsetcache` perspective
there hasn't been any set operation at all.
4. `Misses`. It is the sum of prev and curr misses. If an etry is not
found in curr, it is recorded as a miss. If it is then found in prev,
the entry is returned to the caller, but that cache miss remains. If it
is not found in prev, then there will be 2 misses for 1
`worksingsetcache` get call.
This PR introduces `GetCalls`, `SetCalls`, and `Misses` counts at the
`workingsetcache` level in order to count the calls correctly. It also
excludes duplicates from `EntriesCount`.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9553
### Describe Your Changes
Implemented the script that generates graphs using `gnuplot`.
Those graphs show the write speed to the db.
How to use it:
1. From the root run `make tsbs`;
2. The file will be generated automatically
`/tmp/tsbs-load-100000-2025-07-22T00:00:00Z-2025-07-23T00:00:00Z-80s.csv`
4. From the root run `make tsbs-plot-load` and observe the result
5. If you have two files with the `tsbs_load_victoriametrics` output,
just define the second in the
`TSBS_LOAD_RESULT_CSV_FILE_COMPARE=/tmp/tsbs-load-10
0000-2025-07-22T01:00:00Z-2025-07-23T01:00:00Z-80s.csv
`
To plot the measurements from some other benchmark, run
`make tsbs-plot-load TSBS_LOAD_RESULT_CSV_FILE=/path/to/file.csv`
To plot the measurements from two benchmarks, run
`make tsbs-plot-load TSBS_LOAD_RESULT_CSV_FILE=/path/to/file1.csv
TSBS_LOAD_RESULT_CSV_FILE_COMPARE=/path/to/file2.csv`
This command should generate a graph like described in the picture
<img width="638" height="578" alt="Screenshot 2025-07-25 at 15 35 42"
src="https://github.com/user-attachments/assets/900b05ab-0b98-4f7f-8f2c-18d28ad2eab6"
/>
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
---------
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
Co-authored-by: Artem Fetishev <rtm@victoriametrics.com>
Co-authored-by: Artem Fetishev <149964189+rtm0@users.noreply.github.com>
This helps to improve readability of changes, so users
can see more important changes first, and see changes related
to the same component one after another.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
### Describe Your Changes
Previously mock storage `net.Listen("tcp", …)` could succeed even if
another process was bound to the same port, due to dual-stack behavior
(`[::]:port` vs `0.0.0.0:port`). That lead to strange test results that
hard to bound to port misuse. Tests queried not mock server but whatever
was running on that port.
Switched to `"tcp4"` to ensure conflicts are detected correctly.
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
### Describe Your Changes
As there are quite a few files, and each file might have multiple
changes and to make it easily to review, i limited the PR to 5 files at
a time.
I suggest you take a look at markdownlint and add it as part of your CI,
similar to
https://github.com/MicrosoftDocs/PowerShell-Docs/blob/main/.markdownlint.yaml
And while at it, take a look at cspell and how its used in thier repo
and replace the python one you have in your current implementation -
might open a PR with it after all the fixes PRs).
This pull request consists of the following:
1. Markdown fixes
following https://www.markdownguide.org/basic-syntax/
and https://github.com/markdownlint/markdownlint/blob/main/docs/RULES.md
- Add empty lines after headers or lists
- Remove extra lines between paragraphs
- Remove extra spaces at the end of a line
- Add language to code quote
- Consistent list (dont mix astrixes and dashes on same file, choose one
and be consistent in the same file)
- Proper URL links
- Use meaningful context to URLs instead of "here".
2. Concise language
3. Grammar fixes
- removing extra spaces between words
- there are multiple ones but i picked the basic ones that triggered my
eye :)
4. Spelling fixes
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
This allows performing a single MustFsyncPath() for the parent directory after multiple calls to these functions.
This clarifies code paths, which call these functions, and makes them more maintainable.
This also removes a redundant fsync() call for the parent directory when creating a file-based part.
Previously the first fsync() was indirectly called when the directory was created via MustMkdirFailIfExist()
and the second fsync() was called via MustSyncPathAndParentDir() after all the data is written to the part.
The fs.MustWriteSync() already fsyncs the created file, so there is no need in additional fsync() call.
While at it, add missing fsync for the parent directory after creating a directory for persistent queue.
The source file contents should be already fsynced to disk before creating a hard link,
so there is no sense in calling fsync() on the created hard link.
This commit ensures that the -search.maxQueryLen flag applies to Graphite
queries, matching the behavior already present for Prometheus queries.
Previously, Graphite queries could bypass this limit, creating an
inconsistency and a potential vector for resource exhaustion.
Key changes:
Added getMaxQueryLen() to access the global query length limit.
Enforced query length validation in execExpr() for Graphite queries.
Added comprehensive tests for the new validation logic and edge cases.
Error messages are consistent with Prometheus query validation.
The default limit is 16KB (configurable via -search.maxQueryLen).
Setting the limit to 0 disables validation.
This change closes the gap where Graphite queries could exceed
configured length limits, providing consistent protection against
excessively long queries across both query APIs.
Follow-up for https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9534
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9600
The vm_deleted_metrics_total metric value represents the number of
metricIDs stored in deletedMetricIDs cache. This cache lives at the
storage level and stores the deleted metrics from both prev and curr
idbs. However, the metric is populated at the idb level. Since there are
always 2 idbs (prev and curr), the value is populated twice. Hence the
doubled value of the metric.
The fix is to populate the metric value at the storage level.
Related issue https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9602
- load and parse static`/vmui/config.json`, modify it according to
runtime values and use it as a replacement for static config.json
- remove using `/flags` endpoint for checking features, that should be
enabled on VMUI
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9635
`router.home` represents `/` path, which is the same for all UI apps,
but content and title for root path differs depending on application
type. added `getDefaultOptions` function, which returns proper home
route configuration depending on application type, which allows to
remove renamings in respective layouts
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9641
The commit
25cd5637bc
introduced the `-enableMetadata` flag and the
`promscrape.IsMetadataEnabled()` function, which is now used in multiple
places, including the `app/vminsert/prometheusimport` [request
handler](b24b76ff08/app/vminsert/prometheusimport/request_handler.go (L36)).
Because of the use of `promscrape` package vminsert registered all
`-promscrape.*` service discovery flags, which were not relevant for
`vminsert`.
This change moves the metadata flag logic into a dedicated package,
preventing vminsert from unintentionally loading unrelated promscrape
flags.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9631
This is an attempt to adjust image styles to GitHub themes, because
existing images with transparent backround become unreadable on dark theme.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
This is an attempt to adjust image styles to GitHub themes, because
existing images with transparent backround become unreadable on dark theme.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
When having a `match` of `__name__` key alone for labels api, it's going
to hit max series limit in case of high cardinality metric name.
Instead, we can skip looking by `metricIDs` and fallback to inverted
index scan with a `composite key` since we only have some `__name__` and
a label name.
Common requests for optimisations are:
1) /api/v1/labels?match=up or /api/v1/labels?extra_filters=up
2) /api/v1/label/job/values?match=up or /api/v1/labels?extra_filters =up
It's widely used by grafana variables.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9489
a.Subtract(b) perfomance degrades as b becomes bigger than a. For
example if len(b2) == 10xlen(b1) then time(a.Subtract(b2)) == 10x
time(a.Subtract(b1)).
A quick fix is to iterate over a elements in len(b) > len(a). Iterating
over a's elements and at the same time deleting should be safe since no
elements are actually deleted (i.e. memory freed, etc). Deletion here
means setting a corresponding bit from 1 to 0.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9602
### Describe Your Changes
Add the support of all standard TSDB query types that can be executed
against VictoriaMetrics. `double-groupby-all` is commented out as it
attempts to retrieve all 1B samples and fails. While this can be fixed
by setting the `-search.maxSamplesPerQuery` this query is left disabled
anyway because it will consume way too much memory and cpu time.
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
### Describe Your Changes
New benchmarks for storage search (data and index):
- Use the same dataset that accounts for prev and curr indexDBs and
deleted series
- The code is more structured
- Account for various numbers of series in response including higher
numbers (>10k) as this appears to be a quite common use case.
These bechmarks were used for investigating #9602 performance issue and
helped discover that prefetching metric names needed to be restored
#9619.
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
### Describe Your Changes
Some messages were written to `stdout` using `fmt.Printf` and
`fmt.Println`, while the other messages like import statistics were
written to `stderr` through the `log` package.
This led to ordering problems where the `Import finished!` +
`VictoriaMetrics importer stats` messages, which expected to be the last
messages, appeared before `Continue import process with filter`
messages, creating confusing output for users.
```
2025/08/20 13:07:26 Import finished!
2025/08/20 13:07:26 VictoriaMetrics importer stats:
time spent while importing: 20h49m10.8497184s;
total bytes: 277.1 GB;
bytes/s: 3.7 MB;
requests: 7978614;
requests retries: 0;
2025/08/20 13:07:26 Total time: 20h49m10.851006088s
Continue import process with filter
filter: match[]={__name__!=""}
start: 2025-08-08T00:00:00Z
end: 2025-08-15T00:00:00Z:
Continue import process with filter
filter: match[]={__name__!=""}
start: 2025-08-15T00:00:00Z
end: 2025-08-19T16:18:15Z:
```
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
### Describe Your Changes
It seems db39f045e1 accidentally reverted
#9419 changes.
```patch
--- a/app/vmagent/remotewrite/client.go
+++ b/app/vmagent/remotewrite/client.go
@@ -448,7 +448,8 @@ again:
}
metrics.GetOrCreateCounter(fmt.Sprintf(`vmagent_remotewrite_requests_total{url=%q, status_code="%d"}`, c.sanitizedURL, statusCode)).Inc()
- if statusCode == 409 {
+ switch statusCode {
+ case 409:
logBlockRejected(block, c.sanitizedURL, resp)
// Just drop block on 409 status code like Prometheus does.
@@ -461,7 +462,13 @@ again:
// - 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
- } else if statusCode == 415 || statusCode == 400 {
+ case 415, 400:
+ if c.canDowngradeVMProto.Swap(false) {
+ 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", c.sanitizedURL)
+ c.useVMProto.Store(false)
+ }
+
if encoding.IsZstd(block) {
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", c.sanitizedURL)
```
cc @makasim
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
Previously, if pushBlockPubSub function returned error, vmagent stopped
remote write worker thread assigned for it. Expected behavior for this
scenario is to retry error inside pushBlockPubSub function. It must
return only on vmagent shutdown.
This commit properly handles this error and prevents from ingestion
stop.
Add build tag `disable_grpc_modules` for vmbackup, vmrestore and
vmbackupmanager. Binary size increases only for 3MB with it. It's
acceptable trade-off for security and feature updates.
Related issue:
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8008
* check if built binary is present for `make tsbs-build`. Before, if
build fails, the command stopped working.
* make ENV variables configurable from command line, so `TSBS_STEP=15s
make tsbs-generate-data` would respect the configured step.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Another batch of documentation improvements
Fix Spelling in:
- Comments in code
- Displayed strings
One change was in a json file used for the anomaly dashboard in docker,
else no other code was changed.
Some Markdown changes, related to standards:
- URLs
- List numbering
- Empty spaces at the end of a line
Previously, vmagent treated differently the following configuration:
1) ./bin/vmagent --remoteWrite.url=url-0 --remoteWrite.url=url-1 --remoteWrite.disableOndiskQueue
and
2)./bin/vmagent --remoteWrite.url=url-0 --remoteWrite.url=url-1 --remoteWrite.disableOndiskQueue=true,true
In first case, it could produce duplicates and blocks ingestion requests if one of remote write targets were not accessible.
In second case, it implicitly added --remoteWrite.dropSamplesOnOverload as true and silently dropped samples for inaccessible target.
This commit treat this configuration as the same and silently drop samples on both cases to mitigate possible duplicates.
It's expected, that vmagent provides delivery guarantees, only if it has a single remote write target, when flag remoteWrite.disableOndiskQueue=true is set.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9565
While working on #9431 there has been introduced 2 bugs related to
indexDB.searchMetricName():
1. During the search the index records are unconditionally placed in
sparse index
2. If search touches index records in both prev and curr indexDBs, there
will be possible cases that metricIDs can be unintentionally removed
using `wasMetricIDMissingBefore()` logic
Additionally, the PR moves the searchMetricName from indexDB and Search
to Storage which simplifies the code and makes it spossible to reuse the
function as-is in enterprise code.
Follow up for #9431.
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
`-eula` was deprecated and made no-op in v1.123.0, so examples with
`-eula` will no longer work.
Replace those with proper license configuration.
While at it, remove license flags from vmbackupmanager CLI commands as
it is not required when using CLI.
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
### Describe Your Changes
Do not check vmauth_config_last_reload_success_timestamp_seconds since
it may contain the timestamp < time.Now() due to how lib/fasttime works.
Instead, compare the number of config reloads.
follow up on
https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9369 and
https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9572
Also, split the config update and reload into two separate functions.
master:
```
$gotest -race ./apptest/tests/ -run=TestSingleVMAuthRouterWithInternalAddr -count=40
ok github.com/VictoriaMetrics/VictoriaMetrics/apptest/tests 90.176s
```
pr:
```
$gotest -race ./apptest/tests/ -run=TestSingleVMAuthRouterWithInternalAddr -count=40
ok github.com/VictoriaMetrics/VictoriaMetrics/apptest/tests 46.130s
```
### 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/).
Removing extDB from indexDB makes prev, curr, and next indexDBs independent.
I.e. the search is performed independently in prev and curr, the results are
then merged.
Additionally, since no search is now performed in extDB:
- all indexDB search methods now return the original maps used for populating
the result, without invermediate conversion to slices.
- `NoExtDB` suffix has been removed from method names
This has been extracted from #8134.
Signed-off-by: Andrei Baidarov <baidarov@nebius.com>
Co-authored-by: Artem Fetishev <rtm@victoriametrics.com>
Before, we showed summarized 99th percentile for query complexity across
all available instance. This doesn't make much sense, as it doesn't
answer on the following questions:
1. What complexity limits to set per vmselect
2. What are the most expensive queries
The change is to use `max` instead of `sum`, to show only outliers, the
heaviest served queries. The update should help answering on questions
above.
### Describe Your Changes
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/).
Signed-off-by: hagen1778 <roman@victoriametrics.com>
fixes case, when `histogram_quantile` result contains gaps, that occur
in same time range, where NaNs are present in a first bucket of a
histogram
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
Fix flaky integration test `TestSingleVMAuthRouterWithAuth`.
The flakiness is caused by the
`vmauth_config_last_reload_success_timestamp_seconds` metric, which
reports time with second-level precision.
Update the test to account for this when verifying that the config
reloads correctly.
Previously, if limit was reached for cardinality limiter, vmstorage
started to perform index lookups for any series exceed limit. Since
storage must skip index creation for such series, it's not possible to
cache it. It resulted into opposite effect of cardinality limiter -
instead of reducing resource usage, it increased it instead.
This commit changes cardinality limit calculation from metricID to the
hash from raw metricName. It could slightly increase CPU usage if
cardinality limiter is configured, since hash must be calculated for
each metricName row. But it mitigates excessive CPU and memory usage on
limit hit
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9554
The current one-second timeout for individual read or write operations
during the handshake phase has proven to be insufficient in some
scenarios
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9345. For
example, short-lived CPU spikes lasting a few seconds can cause
handshake failures due to the low timeout threshold.
While a small timeout may work well in environments with fast and
reliable networking, such as within a single datacenter, it becomes
problematic in more complex setups—particularly in a [multi-level
cluster
setup](https://docs.victoriametrics.com/victoriametrics/cluster-victoriametrics/#multi-level-cluster-setup)
where the top-level vmselect may reside in a different availability zone
and work on a less reliable network.
Another issue with the per-operation timeout approach is that it allows
the total time for a handshake to accumulate significantly in the
worst-case scenario. If each operation experiences a delay just under
the timeout threshold, the entire handshake process could take up to 6s.
Which accounts for 60% of `-search.maxQueueDuration` and leaves only 4s
for the actual query.
Introducing a single timeout for the entire handshake process would
provide more predictable behavior and improve usability from a
configuration standpoint. The timeout for the whole handshake op is also
easier to understand from the operator's point of view. Increasing the
timeout value and providing a configuration option for it would make the
system more resilient to transient conditions like CPU contention and
better suited for use cases involving cross-AZ communication.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9345
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
### Describe Your Changes
The commit changes CI behavior:
- Run build in parallel for different os\arch
- Run unit\integration\lint in parallel
- Remove the custom Go cache step in favor of the logic provided in
`actions/setup-go`. The custom cache was used to build key based on
go.sum and makefiles. This logic is preserved.
- Introduce cache for golangci-lint.
### 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/).
vmselect is experiencing memory exhaustion and OOM kills
when processing complex Graphite queries with nested functions and large
numbers of label selectors (30k+ values).
The root cause was unbounded growth of the pathExpression field.
This commit adds configurable truncation for Graphite pathExpression fields to
prevent memory exhaustion while preserving query functionality:
New flag: -search.maxGraphitePathExpressionLen=1024 (default 1024
characters)
Safe truncation: Long expressions are truncated with "..." suffix
Zero disables: Set to 0 to disable truncation entirely
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9534/
Previously, tcp listener perform synchronous proxy protocol header
read during connection accept. It could significantly reduce vmauth
performance and lead to timeout at serving http requests.
This commit changes this logic and performs proxy protocol header
parsing during first Read request from connection or RemoteAddr method
call. It significantly improves performance and reduce possible
bottleneck at connections accept method.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9546/
- Rename WriteRequestUnmarshaller to WriteRequestUnmarshaler
- Add a description to WriteRequestUnmarshaler struct
Review comments
b98e592752 (r163365472)
Follow up on
b98e592752
* remove duplicated content between single and cluster versions
* mention recommendation to group component types by jobs in scrape
config
* link the example of scrape configs
* update wording
Signed-off-by: hagen1778 <roman@victoriametrics.com>
This metric can be used for building alerts and graphs for free disk space usage percentage by using the following MetricsQL query:
100 * (vm_free_disk_space_bytes / vm_total_disk_space_bytes)
### Describe Your Changes
By default `zstd.Reader` creates multiple goroutines to process a single
connection:
- It doesn't match cgo behavior, which works synchronously, and creates
a lot more concurrent goroutines (0.5k -> 5k on my workload)
- It results in non-zero `vm_tcpdialer_errors_total{type="read"}` errors
on vmselect because an underlying connection is closed while a goroutine
is still reading from it. The goroutine created by
`zstd.NewReader`/`zstd.Reset`
abb348e4db/lib/handshake/buffered_conn.go (L113-L120)
- vmselect (and vmagent) doesn't benefit from async mode since it has
multiple readers in-use at the same time, which usually exceeds the
number of cpu cores
Partly related to #9218
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/victoriametrics/contributing/#pull-request-checklist).
- [x] My change adheres to [VictoriaMetrics development
goals](https://docs.victoriametrics.com/victoriametrics/goals/).
This commit introduces new storage API: `/internal/log_new_series`.
It helps to dynamically debug newly created series. Changing `-logNewSeries` value requires storage restart,
which may introduce downtime and is not recommended for production deployments.
In addition, this commit adds flags: `-logNewSeriesAuthKey`, which protects newly added API.
Fixes: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8879
This change will expose any IPv6 addresses assigned to an instance under
the meta labels:
* `__meta_gce_public_ipv6` -native IPv6 address, globally routed
* `__meta_gce_internal_ipv6` - unique local address (ULA).
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9370
Previously, the tenant selector was hidden when only one tenant
was returned, making it impossible to run queries in multi-tenant mode.
Now, the selector is always shown as long as at least one tenant exists.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9396
The prompb and prompbmarshal share exactly the same models and provide
marshal and unmarshale capabilities for them. This creates duplication
(changes in one model has to be made in another, case with metadata) and
confusion where for example you compare same looking models but golang
says they are not the same (because of the type).
This commit merge prompbmarshal logic into prompb so the rest of the
code is aligned on prompb models.
Moves samplesPool and labelsPool to WriteRequestUnmarshaller.
Make WriteRequest struct clean from unmarshal logic.
The benchmark shows no significant changes:
$benchstat prompbmarshal.bench prompb2.bench
goos: darwin
goarch: arm64
pkg: github.com/VictoriaMetrics/VictoriaMetrics/lib/prompb
cpu: Apple M1 Pro
│ prompbmarshal.bench │ prompb2.bench │
│ sec/op │ sec/op vs base │
WriteRequestUnmarshalProtobuf-10 189.2µ ± 5% 190.8µ ± 8% ~ (p=0.579 n=10)
WriteRequestMarshalProtobuf-10 145.3µ ± 7% 143.6µ ± 2% ~ (p=0.143 n=10)
geomean 165.8µ 165.5µ -0.14%
│ prompbmarshal.bench │ prompb2.bench │
│ B/s │ B/s vs base │
WriteRequestUnmarshalProtobuf-10 50.42Mi ± 5% 49.99Mi ± 8% ~ (p=0.593 n=10)
WriteRequestMarshalProtobuf-10 65.64Mi ± 7% 66.39Mi ± 2% ~ (p=0.143 n=10)
geomean 57.53Mi 57.61Mi +0.14%
│ prompbmarshal.bench │ prompb2.bench │
│ B/op │ B/op vs base │
WriteRequestUnmarshalProtobuf-10 27.70Ki ± 4% 26.90Ki ± 7% ~ (p=0.190 n=10)
WriteRequestMarshalProtobuf-10 3.267Ki ± 12% 3.273Ki ± 12% ~ (p=0.971 n=10)
geomean 9.514Ki 9.383Ki -1.38%
│ prompbmarshal.bench │ prompb2.bench │
│ allocs/op │ allocs/op vs base │
WriteRequestUnmarshalProtobuf-10 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
WriteRequestMarshalProtobuf-10 0.000 ± 0% 0.000 ± 0% ~ (p=1.000 n=10) ¹
geomean ² +0.00% ²
¹ all samples are equal
² summaries must be >0 to compute geomean
This commit introduces new flag: `-storage.idbPrefillStart` with
default value of `1h`. It allows to adjust start time of the prefill
process for the data written into the next indexDB.
By default, VictoriaMetrics starts prefill indexDB at 3 A.M UTC, while
indexDB rotates at 4 A.M UTC. It could be useful to change start time
from 3 A.M. to 1 A.M or 00:00 A.M. It should smooth overall resource
usage.
However, changing value to the number bigger than 4 hours for the
default
installations doesn't make much sense ( like 11 P.M. for the day before
rotation). Since, VictoriaMetrics maintenances daily-indexes and it have
to repopulate it twice.
But it could be useful in conjunction with `-retentionTimezoneOffset`,
which could delay index rotation for the current day and give more time
for the prefill process.
As an example, `-retentionTimezoneOffset=4h` adds an additional 4 hours
to the rotation time and `-storage.idbPrefillStart` could be changed
accordingly.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9393
### Describe Your Changes
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/).
---------
Signed-off-by: f41gh7 <nik@victoriametrics.com>
Co-authored-by: Roman Khavronenko <roman@victoriametrics.com>
Previously, vmctl tried to create a tmp directory in a directory with
Prometheus snapshot. This is not always possible as snapshot can be
mounted in read-only mode.
Use a system default temporary location and allow user to customize the
tmp path in order to avoid issues with custom tmp locations.
Closes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9505
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
The `groupWatchersCleaner` iterates through all watchers, attempting to
lock them sequentially while holding the global `groupWatchersLock`.
Therefore, a large number of group watchers can cause the
`groupWatchersLock.Lock()` to block for a noticeable period.
Proposing to use `TryLock` as an optimistic case because if the watcher
lock is held, it's more likely that the watcher is still in-use so no
further cleanup is required.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9354#issuecomment-3089108889
The lib/fs.MustCloseParallel() accepts a slice of MustWriter items, which must implement only
a single method - MustWrite(). The previous lib/filestream.MustCloseWritersParallel() was
accepting CloseWriter items, which must implement Write() and Path() methods additionally
to MustClose() method. This was adding artificial restrictions on the applicability
of the MustCloseWritersParallel() method. Remove these restrictions.
This should reduce the time needed for the deletion of VictoriaLogs parts
with big number of files, which are created when wide events are ingested into VictoriaLogs
(e.g. logs with big number of log fields).
This may help improving scalability of VictoriaLogs at systems with big number of CPU cores,
which store data into high-latency storage such as Ceph or NFS.
See https://github.com/VictoriaMetrics/VictoriaLogs/issues/517
The import aliases may complicate maintenance of the code in the long term
if they aren't used consistently, e.g. if one file imports the apptest under the default name
while the other file imports the apptest under the "at" name.
The aliases also complicate grepping the code by apptest.* or prompbmarshal.* .
- Drop the code needed for asynchronous removal of the directory on NFS shares.
This code was needed when VictoriaMetrics could keep open files after their deletion
or renaming. This is no longer the case after the commit 43b24164ef .
Now files are deleted only after all the readers close them.
This updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/61
- Unify MustRemoveAll() and MustRemoveDirAtomic() into MustRemoveDir() and MustRemovePath()
functions:
- The MustRemoveDir() deletes the given directory with all its contents, in an "atomic" way:
it creates a special `.delete-this-dir` file in the directory, then removes all its contents
except of this file, and later removes the `.delete-this-dir` file together with the directory
itself. This makes possible easily determining whether the given directory needs to be deleted
after unclean shutdown - if it contains the `.delete-this-dir` file or if it is empty, it must be deleted.
Add IsPartiallyRemovedDir() function, which can be used for detecting whether the given directory must be removed
at starup.
Previously the MustRemoveDirAtomic() was using a "trick" for atomic directory removal: it was "atomically" renaming
the directory to a temporary directory with '.must-remove.' marker in the directory name, and after that it
was removing the renamed directory. On startup all the directories with the `.must-remove.` marker were deleted
if they are left after unclean shutdown. This "trick" doesn't work for NFS and object storage such as S3,
since these storage systems do not support atomic renaming of directories with multiple entries inside.
The new MustRemoveDir() function doesn't use this "trick", so it can be safely used in NFS and S3-like storage systems.
This is based on the pull request from @func25 - https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9486/files .
- The MustRemovePath() deletes the given file or an empty directory.
- Delete the existing parts and partitions at startup if they were partially deleted.
- Consistently use fs.MustRemoveDir() and fs.MustRemovePath() instead of os.RemoveAll() across the codebase.
This reduces the amounts of bolierplate code related to error handling.
- Consistently use fs.MustWriteSync() instead of os.WriteFile() across the codebase.
- Consistently use fs.IsPathExist() for checking whether the given path exists on the filesystem.
- Consistently use fs.MustWriteAtomic() for atomic store of the serialized state into file.
- Consistently panic with 'BUG:' prefix on unexpected errors.
- Read and write the state file contents in one go. This simplifies the code for loading and storing the state.
This shouldn't increase memory usage too much, since the parsed state is already stored in RAM.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9074
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9102
The following renamings were requested in #8134 to be done in a separate
PR:
- Rename putTSIDToCache(s) to storeTSIDToCache(s). This is because
get/put prefixes are normally reserved for pools and ref counters. This
unclear though whether name `getTSIDFromCache` is okay in this context.
- Rename `addPartitionNolock` to `addPartitionLocked`, because the
`Locked` suffix is used everywhere else
- Rename deleteMetricIDs to saveDeletedMetricIDs, because no deletion is
actually happening. The deleted metric ids are still present in the
index. One needs to filter the deleted metric ids out after the
retrieval.
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
Previously, it was not possible to compile netBSD binary due to missing OS constrains at lib/fs and lib/filestream packages.
This commit fixes it by:
* apply proper constrains at lib/filestream
* Introduce statfs_t and statfs() to abstract unix internals: NetBSD
needs to use unix.Statvfs_t and unix.Statvfs() unlike other Unix-es.
* apply proper constrain for vmctl terminal package
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9473
The newly created data part could become missing after unclean shutdown (such as hardware power off),
since the contents of the parent directory wasn't synced to disk before storing the newly created data part in the parts.json file.
Fix this by syncing the parent directory contents before storing the newly created part in the parts.json file.
This commit is based on https://github.com/VictoriaMetrics/VictoriaLogs/pull/507
This commit adds tmp inmemory and data blocks buffers for
index search requests. It allows to reduce memory allocations on block
cache misses. Since block cache puts block into cache only on after
configured number of cache misses.
Related PR https://github.com/VictoriaMetrics/VictoriaMetrics/pull/9324
This commit respects online CPU count for process_cpu_cores_available metric. Since CPUQuota may exceed it.
Detailed description:
There might be a case when `getCPUQuota()` returns a value bigger than
available logic CPU cores. In this case the lower value should be used.
These changes also reflect upcoming go1.25 changes
https://tip.golang.org/doc/go1.25#container-aware-gomaxprocs
> If the CPU bandwidth limit is lower than the number of logical CPUs
available, GOMAXPROCS will default to the lower limit
In practice that happens with [CPU
Manager](https://kubernetes.io/blog/2022/12/27/cpumanager-ga/) enabled.
It requires setting `/sys/devices/system/cpu/online` which shows the
total number of cores on a node. The container doesn't have any cgroup
limits, but it's pinned to a limited subset of cores.
```
root@vmselect:/# cat /sys/fs/cgroup/cpu/cpu.cfs_quota_us
-1
root@vmselect:/# cat /sys/fs/cgroup/cpu/cpu.cfs_period_us
100000
root@vmselect:/# cat /sys/devices/system/cpu/online
0-255
```
go1.25 and go.uber.org/automaxprocs don't look into
`/sys/devices/system/cpu/online`, VictoriaMetrics does
The chunkedbuffer is released twice, leading to concurrent use of the
buffer after it's acquired from a pool by two different goroutines.
Issue was introduced at v1.115.0 at the following commit 5b87aff830
### Describe Your Changes
Previously, if I run `make publish-vmstorage` it composed the tag name
like:
```
docker.io/victoriametrics/vmstorage:heads-tcpdialer-increase-idle-timeout-0-g1b20c2dbd3-dirty-84b03c0eEXTRA_DOCKER_TAG_SUFFIX
```
It would work okay if I explisitly provide empty
`EXTRA_DOCKER_TAG_SUFFIX=` env var.
### 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/).
This reverts commit 63e1bf5d97. The reason
for revert is that the change makes increase/rate behavior less
predictable for users. See
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9420.
Another negative impact of this change is that it negatively affects
recording rules that calculate increases or rates over time series on
big intervals. For example,
`increase(http_errors_total{instance="foo"}[30d])` would stop producing
results if `http_errors_total{instance="foo"}` was marked as stale. But
this is logically incorrect, as `http_errors_total{instance="foo"}` over
last 30d should still return results.
-------------
This change doesn't revert
63e1bf5d97
completely. It keeps changes made to `removeCounterResets` in order to
preserve original `staleNaN` values. Before, `staleNaN` were compared
with float values and producing `NaN` instead. Which could confuse us in
future. So keeping the fix and tests. It shouldn't have any negative
effect.
### Describe Your Changes
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/).
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Previously, snapshots were never deleted automatically. This lead to
disk space waste in case snapshots were left behind and never deleted,
which happened in case of backups failure. This required manual
investigation and snapshots cleanup.
This change enables removal of snapshots older than 3 days by default.
This should give enough time to upload backup in time and also make sure
old snapshots are not wasting disk space.
Closes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/9344
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
### Describe Your Changes
Related issue: #9388
- removed all the code related to VictoriaLogs
- updated dependencies to the latest versions
- removed unnecessary `React` import from components
- removed deprecated dependencies such as `lodash.get`,
`@babel/plugin-proposal-nullish-coalescing-operator`,
`@babel/plugin-proposal-private-property-in-object`
- removed unused packages
- fixed proxy for local playground development (after refresh the page
app cannot properly parse the `/#/` in the path)
### 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/).
Add unit tests that verify filesystem hierarchy of the indexdb after
opening the storage and after rotation. The test have been originally
added in #8134 but are still applicable to the current version and will
also reduce the diff.
Co-authored-by: Artem Fetishev <rtm@victoriametrics.com>
- Move `prefetchMetricNames`, `SearchMetricNames`, and
`SearchLabelValues` from `Storage` to `indexDB`
- Rename `searchLabelValuesWithFiltersOnTimeRange` to
`searchLabelValuesOnTimeRange`
- Rename `searchLabelValuesWithFiltersOnDate` to
`searchLabelValuesOnDate`
Extracted from #8134.
Co-authored-by: Artem Fetishev <rtm@victoriametrics.com>
### Describe Your Changes
- Split the release process into two steps.
- Add some links
- Described in more details what should be done to check branches in
sync.
- Added some small checks, like running tests, and testing final release
on sandbox.
- Changed the order of some actions (like build final image before
publish release).]
### 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/).
2025-07-10 12:14:17 +03:00
2199 changed files with 455417 additions and 70359 deletions
VictoriaMetrics is a fast, cost-saving, and scalable solution for monitoring and managing time series data. It delivers high performance and reliability, making it an ideal choice for businesses of all sizes.
## Folder Structure
-`/app`: Contains the compilable binaries.
-`/lib`: Contains the golang reusable libraries
-`/docs/victoriametrics`: Contains documentation for the project.
-`/apptest/tests`: Contains integration tests.
## Libraries and Frameworks
- Backend: Golang, no framework. Use third-party libraries sparingly.
- Frontend: React.
## Code review guidelines
Ensure the feature or bugfix includes a changelog entry in /docs/victoriametrics/changelog/CHANGELOG.md.
Verify the entry is under the ## tip section and matches the structure and style of existing entries.
Chore-only changes may be omitted from the changelog.
// Just drop block on 409 status code like Prometheus does.
@@ -461,22 +462,28 @@ again:
// - 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
}elseifstatusCode==415||statusCode==400{
ifc.canDowngradeVMProto.Swap(false){
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",c.sanitizedURL)
c.useVMProto.Store(false)
}
case415,400:
ifencoding.IsZstd(block){
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",c.sanitizedURL)
block=mustRepackBlockFromZstdToSnappy(block)
zstdBlockLen:=len(block)
block,err=repackBlockFromZstdToSnappy(block)
iferr==nil{
ifc.canDowngradeVMProto.Swap(false){
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",c.sanitizedURL)
c.useVMProto.Store(false)
}
c.retriesCount.Inc()
_=resp.Body.Close()
gotoagain
c.retriesCount.Inc()
_=resp.Body.Close()
gotoagain
}
logger.Warnf("failed to repack zstd block (%s bytes) to snappy: %s; The block will be rejected. "+
"Possible cause: ungraceful shutdown leading to persisted queue corruption.",
zstdBlockLen,err)
}
// Just drop snappy blocks on 400 or 415 status codes like Prometheus does.
flushInterval=flag.Duration("remoteWrite.flushInterval",time.Second,"Interval for flushing the data to remote storage. "+
"This option takes effect only when less than 10K data points per second are pushed to -remoteWrite.url")
"This option takes effect only when less than -remoteWrite.maxRowsPerBlock data points per -remoteWrite.flushInterval are pushed to -remoteWrite.url")
maxUnpackedBlockSize=flagutil.NewBytes("remoteWrite.maxBlockSize",8*1024*1024,"The maximum block size to send to remote storage. Bigger blocks may improve performance at the cost of the increased memory usage. See also -remoteWrite.maxRowsPerBlock")
maxRowsPerBlock=flag.Int("remoteWrite.maxRowsPerBlock",10000,"The maximum number of samples to send in each block to remote storage. Higher number may improve performance at the cost of the increased memory usage. See also -remoteWrite.maxBlockSize")
maxMetadataPerBlock=flag.Int("remoteWrite.maxMetadataPerBlock",5000,"The maximum number of metadata to send in each block to remote storage. Higher number may improve performance at the cost of the increased memory usage. See also -remoteWrite.maxBlockSize")
vmProtoCompressLevel=flag.Int("remoteWrite.vmProtoCompressLevel",0,"The compression level for VictoriaMetrics remote write protocol. "+
"Higher values reduce network traffic at the cost of higher CPU usage. Negative values reduce CPU usage at the cost of increased network traffic. "+
"See https://docs.victoriametrics.com/victoriametrics/vmagent/#victoriametrics-remote-write-protocol")
// A single time series left. Recursively split its samples and metadata into smaller parts if possible.
samples:=wr.Timeseries[0].Samples
iflen(samples)==1{
logger.Warnf("dropping a sample for metric with too long labels exceeding -remoteWrite.maxBlockSize=%d bytes",maxUnpackedBlockSize.N)
metaData:=wr.Metadata
iflen(samples)==1&&len(metaData)<=1{
logger.Warnf("dropping a sample for metric and %d metadata which are exceeding -remoteWrite.maxBlockSize=%d bytes",len(metaData),maxUnpackedBlockSize.N)
returnfmt.Errorf("too many -remoteWrite.streamAggr.config args: %d; it mustn't exceed the number of -remoteWrite.url args: %d",len(*streamAggrConfig),len(*remoteWriteURLs))
"Defines how many retries to make before giving up on rule if request for it returns an error.")
disableProgressBar=flag.Bool("replay.disableProgressBar",false,"Whether to disable rendering progress bars during the replay. "+
"Progress bar rendering might be verbose or break the logs parsing, so it is recommended to be disabled when not used in interactive mode.")
ruleEvaluationConcurrency=flag.Int("replay.ruleEvaluationConcurrency",1,"The maximum number of concurrent `/query_range` requests for a single rule. "+
"Increasing this value when replaying for a long time and a single request range is limited by `-replay.maxDatapointsPerQuery`.")
ruleEvaluationConcurrency=flag.Int("replay.ruleEvaluationConcurrency",1,"The maximum number of concurrent '/query_range' requests when replay recording rule or alerting rule with for=0. "+
"Increasing this value when replaying for a long time, since each request is limited by -replay.maxDatapointsPerQuery.")
ruleResultsLimit=flag.Int("rule.resultsLimit",0,"Limits the number of alerts or recording results a single rule can produce. "+
"Can be overridden by the limit option under group if specified. "+
"If exceeded, the rule will be marked with an error and all its results will be discarded. "+
"0 means no limit.")
ruleUpdateEntriesLimit=flag.Int("rule.updateEntriesLimit",20,"Defines the max number of rule's state updates stored in-memory. "+
"Rule's updates are available on rule's Details page and are used for debugging purposes. The number of stored updates can be overridden per rule via update_entries_limit param.")
resendDelay=flag.Duration("rule.resendDelay",0,"MiniMum amount of time to wait before resending an alert to notifier.")
"See https://docs.victoriametrics.com/victoriametrics/vmauth/#load-balancing for details")
defaultLoadBalancingPolicy=flag.String("loadBalancingPolicy","least_loaded","The default load balancing policy to use for backend urls specified inside url_prefix section. "+
"Supported policies: least_loaded, first_available. See https://docs.victoriametrics.com/victoriametrics/vmauth/#load-balancing")
defaultMergeQueryArgs=flagutil.NewArrayString("mergeQueryArgs","An optional list of client query arg names, which must be merged with args at backend urls. "+
"The rest of client query args are replaced by the corresponding query args from backend urls for security reasons; "+
"see https://docs.victoriametrics.com/victoriametrics/vmauth/#query-args-handling")
discoverBackendIPsGlobal=flag.Bool("discoverBackendIPs",false,"Whether to discover backend IPs via periodic DNS queries to hostnames specified in url_prefix. "+
"This may be useful when url_prefix points to a hostname with dynamically scaled instances behind it. See https://docs.victoriametrics.com/victoriametrics/vmauth/#discovering-backend-ips")
discoverBackendIPsInterval=flag.Duration("discoverBackendIPsInterval",10*time.Second,"The interval for re-discovering backend IPs if -discoverBackendIPs command-line flag is set. "+
// check for any lingering errors on the query side
forotsdbErr:=rangeerrCh{
returnfmt.Errorf("Import process failed: \n%s",otsdbErr)
returnfmt.Errorf("import process failed: \n%s",otsdbErr)
}
bar.Finish()
log.Print(op.im.Stats())
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
