Alert `TooHighChurnRate24h` suppose to cover cases when churn rate
is low but results in multiple times higher number than total
number of active series.
* dashboard: update single node dashboard
* add number of new series created over last 24h;
* bump version requirements.
* dashboard: update vmagent dashboard
* add panel for open file descriptors;
* add panel for disk I/O;
* add panel for `vmagent_remotewrite_packets_dropped_total` metric;
* bump version requirements.
* adds blocks drop at 400 BadRequest status code
recieved from remote storage,
not expected that remote storage will be able to handle it on retry
* removes error logging for dropped blocks,
its expected error
3s evaluation interval is too small for practical setups. It can result in increased load on datasource.
So it is better to remove it from example config args, which are usually copy-pasted by novice users.
This adds the following new metrics for each VictoriaMetrics app:
* process_resident_memory_anonymous_bytes - the RSS share for memory allocated by the process itself.
This share cannot be freed by the OS, so it must be taken into account by OOM killer.
* process_resident_memory_pagecache_bytes - the RSS share for page cache memory (aka memory-mapped files).
This share can be freed by the OS at any time, so it must be ignored by OOM killer.
There was a signficant refactoring in the code responsible for time series search,
so it can result in both speed ups and slow downs depending on used queries.
The archive contains the following executables for Windows:
* vmagent
* vmalert
* vmauth
* vmctl
Other components - vmbackup, vmrestore, victoria-metrics - aren't supported for Windows yet
Do not pass filter metric ids to getMetricIDsForTagFilter, since it has been appeared that this slows down
the function by multiple times when it finds big number of metricIDs (tens of millions).
* dashboard: update single node dashboard
* add panel `Open FDs` for file descriptors metrics;
* add panel `Disk writes/reads` to show the real read/write
load on storage layer;
* add `process_resident_memory_bytes` metric to memory usage panel;
* add stats panel to show available CPUs, memory and disk space;
* rm flags panel since it didn't prove its usefulness.
* alerts: add alert for reaching FDs limit
Do not cache too big byte buffers and too big writeRequestCtx objects,
since it is cheaper to re-create them instead of wasting RAM for their caching.
This reverts 7f6f350ee1
Previously multiple scrape jobs could create multiple watchers for the same apiURL. Now only a single watcher is used.
This should reduce load on Kubernetes API server when many scrape job configs use Kubernetes service discovery.
See the description for `sample_limit` option from Prometheus docs:
Per-scrape limit on number of scraped samples that will be accepted.
If more than this number of samples are present after metric relabeling
the entire scrape will be treated as failed. 0 means no limit.
https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config
Serialize reloading per-role objects, so they don't occupy too much memory when objects for many scrape jobs are simultaneously refreshed.
Do not reload per-role objects if they were already refreshed by concurrent goroutines. This should reduce load on Kubernetes API server
when big number of scrape jobs are configured for the same Kubernetes role.
This is a follow-up for 17b87725ed
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1113
Previously vmagent was creating a separate Kubernetes object cache per each scrape job.
This could result in increased memory usage when monitoring a Kubernetes cluster with big number of objects (pods / nodes / services, etc.)
as seen at https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1113
Now it uses a shared map of scrape objects across multiple scrape jobs.
* fixes windows compilation,
adds signal impl for windows,
adds free space usage for windows,
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/70https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1036
NOTE victoria metrics database still CANNOT work under windows system,
only vmagent is supported.
To completly port victoria metrics, you have to fix issues with separators,
parsing and posix file removall
* rollback separator
* Adds windows setInformation api,
it must behave like unix, need to test it.
changes procutil
* check for invlaid param
* Fixes posix delete semantic
* refactored a bit
* fixes openbsd build
* removed windows api call
* Fixes code after windows add
* Update lib/procutil/signal_windows.go
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
The first and the last buckets are usually `[0 ... leMin]` and `(leMax ... +Inf)`. If they are merged with adjancent buckets,
then the resulting accuracy can suffer.
Main points:
* Revert changes outside lib/promscrape/discovery/kuberntes . These changes can be applied later in a separate commit
* Minimize changes in lib/promscrape/discovery/kubernetes compared to a93e644001
* Corner case fixes.
* started work on sd for k8s
* continue work on watch sd
* fixes
* continue work
* continue work on sd k8s
* disable gzip
* fixes typos
* log errror
* minor fix
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
The v1.15.0 exports the following additional metrics:
process_io_read_bytes_total - the number of bytes read via io syscalls such as read and pread
process_io_written_bytes_total - the number of bytes written via io syscalls such as write and pwrite
process_io_read_syscalls_total - the number of read syscalls such as read and pread
process_io_write_syscalls_total - the number of write syscalls such as write and pwrite
process_io_storage_read_bytes_total - the number of bytes read from storage layer
process_io_storage_written_bytes_total - the number of bytes written to storage layer
These metrics can be used for monitoring process io
While this may increase CPU and disk IO usage needed for background merge,
this also recudes CPU usage during queries in production. This is because
such queries tend to read recently added data and it is better to have lower number
of parts for such data in order to reduce CPU usage.
This partially reverts ebf8da3730
The filter arg has been removed in the commit c7ee2fabb8
because it was preventing from caching the number of matching time series per each tf.
Now the cache contains duration for tf execution, so the filter shouldn't break such caching.
For example `{label=~"foo\.bar"}` should be converted to `{label="foo.bar"}`. Previously it has was mistakenly conveted to `{label="foo\.bar"}` .
This could result in missing time series for such tag filters.
These metrics may result in big number of time series when vmagent scrapes thousands of targets and these targets constantly changes.
* It is better using `up == 0` query for determining failing targets.
* It is better using the following query for determining targets with exceeded limit on the number of metrics:
scrape_samples_scraped > 0 if up == 0
The `filter` arg breaks the logic for sorting tag filters by the matching metrics,
which may result in non-optimal performance during time series search.
Production workloads show that indexdb blocks must be cached unconditionally for reducing CPU usage.
This shouldn't increase memory usage too much, since unused blocks are removed from the cache every two minutes.
* init implementation for graphite alerts
* adds graphite support for vmalert
* small fix
* changes vmalert graphite api with type
* updates tests
* small fix
* fixes graphite parse
* Fixes graphite from time
It is better developing vmctl tool in VictoriaMetrics repository, so it could be released
together with the rest of vmutils tools such as vmalert, vmagent, vmbackup, vmrestore and vmauth.
These metrics could be useful for determining imporperly working scrape targets.
Note that these metrics are exported only for failing scrape targets. They aren't exposed for normally working targets.
* Include individual binary checksums for vmutils
* Consistent archive/binary artefacts between arm64/amd64 for vmutils
* architecture in arhcive, checksums
* not in binaries
* adds extra_label to all import apis,
changes priority for extra_label - now it has priority over original labels
* Update README.md
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
* Update README.md
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
* adds extra labels to vmagent import api
changes order for adding labels, now its added after user values
* adds tests for extra_label
* import fix
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
List contains examples for the alerting rules which might be executed
via `vmalert` to track the health state of VM components. It is assumed
that list will be revised and calibrated for each system individually.
On templates validation stage vmalert does not acutally send queries, so for complex
chained expression validation may fail. To avoid this, we add a blank sample in response
so validation can pass successfully. Later, during the rule execution, stub will be replaced
with real `query` function.
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/989
Currently, alertmanager spams logs with `Notify attempt failed, will retry later` message
because default receiver is unreachable. The change updates default configuration with
blackhole receiver which means alertmanager will continue to accept alerts but won't make
attempts to send them anywhere.
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/995
* adds snap docs,
adds release information for snap package,
adds docs notes about configuration management with snap package.
* adds release page mention
* version fix for snap, its awful
* revert version
* Apply suggestions from code review
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
Previously it could be modified in order to improve response cache hit ratio.
This is unneeded, since cache hit ratio should remain good because the query time range
should be already aligned to multiple of `step` values.
See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/976
* Adds query stat handler,
for query and query_range api, victoriametrics tracks query execution time,
stats are expored at /api/v1/status/queries endpoint with topN param
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/907
* fixed query stats bugs
* improves queryStats tracker
* improves query stat
* small fix
* fix tests
* added more tests
* fixes 386 tests
* naming fixes
* adds drop for outdated records
* adds proxy_url support,
adds proxy_url to the dockerswarm, eureka, kubernetes and consul service discovery,
adds proxy_url to the scrape_config for targets scrapping,
http based proxy is supported atm,
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/503
* fixes imports
Previously such parts could remain undeleted for long durations until they are merged with other parts.
This should help for `-retentionPeriod` values smaller than one month.
It is unclear why `go install` doesn't work in Github Actions. Needs additional investigation.
The following error is returned now:
cannot find package "golang.org/x/lint/golint" in any of:
/opt/hostedtoolcache/go/1.15.5/x64/src/golang.org/x/lint/golint (from $GOROOT)
/home/runner/go/src/golang.org/x/lint/golint (from $GOPATH)
* adds ArrayDuration and ArrayBool flags,
makes sendTimeout and tlsInsecure configurable per remoteWrite url
* added backward compatibility testcases for ArrayDuration and ArrayBool
* fixes bool flag
* fixes test cases
The commit adds a support for template function `query`,
`first` and `value`. The function `query` executes
a MetricsQL query for active alerts. In vmalert we
update templates on every evaluation for active alerts
to keep them up to date. With `query` func it may become
a perf issue since it will fire a query on every execution.
We should keep it in mind for now.
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/539
This is a preparation for Go 1.16, which deprecates `go get` for installing binaries.
See https://tip.golang.org/doc/go1.16#go-command :
go install, with or without a version suffix (as described above), is now the recommended way
to build and install packages in module mode. go get should be used with the -d flag to adjust
the current module's dependencies without building packages, and use of go get to build and install
packages is deprecated. In a future release, the -d flag will always be enabled.
This diff is just to suggest wording to let people know there is no future-compatible guaranteed way to make their own native format files for import yet.
This should prevent from possible 'memory leaks' when a pointer to ScrapeWork item stored in the slice
could prevent from releasing memory occupied by all the ScrapeWork items stored in the slice when they
are no longer used.
See the related commit e205975716 and the related issue https://github.com/VictoriaMetrics/VictoriaMetrics/issues/825
* adds consul watch api,
it must reduce load on consul service with blocking wait requests,
changed discoveryClient api with fetchResponseMeta callback.
* small fix
* fix after master merge
* adds watch client at discovery utils
* fixes consul watcher,
changes namings,
fixes data race
* small typo fix
* sanity fix
* fix naming and service node update
OpenMetrics timestamps are floating-point numbers, that represent Unix timestamp in seconds.
This differs from Prometheus exposition format, where timestamps are integer numbers representing Unix timestamp in milliseconds.
All the callers for fs.OpenReaderAt expect that the file will be opened.
So it is better to log fatal error inside fs.MustOpenReaderAt instead of leaving this to the caller.
It is recommended to have at least of 50% of free RAM on vmstorage nodes in order handle possible
RAM usage spikes during rolling upgrade for vmstorage nodes when time series
are re-routed from temporarily unavailable node to the remaining active nodes.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/891
This commit also adds `"isPartial":{true|false}` field to `/api/v1/*` responses. `"isPartial":true` is set when the response
is based on a partial data because some of vmstorage nodes weren't available during query processing.
* Add omitempty for DisableCompression and DisableKeepAlive fields in ScrapeConfig
* Add omitempty annotation to all the default/optional values
* Fix annotations after review
This should prevent from holding previously discovered []ScrapeWork slices when a part of discovered targets changes over time.
This should reduce memory usage for the case when big number of discovered scrape targets changes over time.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/825
The previous implementation treated extra labels (global and rule labels) as
separate label set to returned time series labels. Hence, time series always contained
only original labels and alert ID was generated from sorted labels key-values.
Extra labels didn't affect the generated ID and were applied on the following actions:
- templating for Summary and Annotations;
- persisting state via remote write;
- restoring state via remote read.
Such behaviour caused difficulties on restore procedure because extra labels had to be dropped
before checking the alert ID, but that not always worked. Consider the case when expression
returns the following time series `up{job="foo"}` and rule has extra label `job=bar`.
This would mean that restored alert ID will be always different to the real time series because
of collision.
To solve the situation extra labels are now always applied beforehand and `vmalert` doesn't
store original labels anymore. However, this could result into a new error situation.
Consider the case when expression returns two time series `up{job="foo"}` and `up{job="baz"}`,
while rule has extra label `job=bar`. In such case, applying extra labels will result into
two identical time series and `vmalert` will return error:
`result contains metrics with the same labelset after applying rule labels`
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/870
* adds leading forward slash check for scrapeURL path
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/835
* adds ready probe for scrape config initialization,
it should prevent metrics loss during vmagent rolling update,
/ready api will return 425 http code, if some scrape config still waits for initialization.
* updates docs
* Update app/vmagent/README.md
* renames var
* Update app/vmagent/README.md
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
Previously such blocks were cleaned after they weren't accessed during 10 minutes.
Now they are cleaned after one minute of missing access. This should reduce memory usage in general case.
* add short_version label to vm_app_version metric
use case: Version panel of Grafana dashboard should use a live query, but currently it uses a template query which becomes stale. Grafana is not able to preform regex substitution on labels.
* Update metrics.go
* fix compile
* dashboard: add `Storage full ETA` panel
The new panel suppose to help to estimate the time needed to run out of free
disk space.
Thx to @belm0 @hekmon
* disable legend for `Storage full ETA` panel
Label `alertgroup` was introduced in #611 and automatically added to generated
time series. By mistake, this new label wasn't correctly purged on restore event
and affected alert's ID uniqueness. This commit removes `alertgroup` label
in restore function.
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/870
The new release of github.com/VictoriaMetrics/metricsql adds more optimizations for `foo{filters1} op bar{filters2}`:
* rollup_func(foo[d]) op bar{filters}
* transform_func(foo) op bar{filters}
* num_or_scalar op bar op baz{filters}
This should prevent from double counting for time series at the time when it changes label.
The most common case is in K8S, which changes pod uid label with each new deployment.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/748
This reverts commit ac45082216.
Reason for revert: the previous behavior for VictoriaMetrics is easier to understand and use by users -
functions, which don't change the meaning of the time series shouldn't drop metric name.
Now the following functions do not drop metric names:
* ceil
* floor
* round
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/674
This reverts commit bb61a4769b.
Reason for revert: the previous behavior for VictoriaMetrics is easier to understand and use by users -
functions, which don't change the meaning of the time series shouldn't drop metric name.
Now the following functions do not drop metric name:
* clamp_min
* clamp_max
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/674
This reverts commit e5202a4eae.
Reason for revert: the previous behavior for VictoriaMetrics is easier to understand and use by users -
functions, which don't change the meaning of the time series shouldn't drop metric name.
Now the following functions do not drop metric name:
* max_over_time
* min_over_time
* avg_over_time
* quantile_over_time
* geomean_over_time
* mode_over_time
* holt_winters
* predict_linear
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/674
Show also original labels for duplicate targets in error message in order to simplify debugging the issue.
Now `/targets` endpoint accepts optional `show_original_labels=1` query arg, which shows original labels for each target.
This may simplify debugging for target relabeling.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/651
Remove the code that uses metricIDs caches for the current and the previous hour during metricIDs search,
since this code became unused after implementing per-day inverted index almost a year ago.
While at it, fix a bug, which could prevent from finding time series with names containing dots (aka Graphite-like names
such as `foo.bar.baz`).
Previously the maximum cache lifetime has been limited by 10 seconds. Now it is extended up to a day.
This should reduce CPU usage in the following cases:
* when querying recently added data with small churn rate for time series
* when querying historical data
This reverts commit bab6a15ae0.
Reason for revert: the `fetchData` arg is used in cluster branch.
Leaving this arg in master branch makes smaller the diff with cluster branch.
* Add improvements to ec2 discovery
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/771
role_arn support with aws sts
instance iam_role support
refreshing temporary tokens
* Apply suggestions from code review
Co-authored-by: Roman Khavronenko <hagen1778@gmail.com>
* changed implementation, removed tests, clean up code
* moved endpoint builder into getEC2APIResponse
Co-authored-by: Roman Khavronenko <hagen1778@gmail.com>
The strategy is:
- Periodical flushing of inmemory blocks to files, so they aren't lost on unclean shutdown.
- Periodical syncing of metadata for persisted queues, so the metadata remains in sync with the persisted data.
- Automatic adjusting of too big chunk size when opening the queue. The chunk size may be bigger than the writer offset after unclean shutdown.
- Skipping of broken chunk file if it cannot be read.
- Fsyncing finalized chunk files.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/687
Previously certain errors in timestamps and/or values could be silently skipped,
which could lead to samples with zero values stored in the database.
Updates https://github.com/VictoriaMetrics/vmctl/issues/25
On config reload event `vmalert` reloads configuration for every group. While
it works for simple configurations, the more complex and heavy installations may
suffer from frequent config reloads.
The change introduces the `checksum` field for every group and is set to md5 hash
of yaml configuration. The checksum will change if on any change to group
definition like rules order or annotation change. Comparing the `checksum` field
on config reload event helps to detect if group should be updated.
The groups update is now done concurrently, so reload duration will be limited by
the slowest group now.
Partially solves #691 by improving config reload speed.
This is similar to https://github.com/prometheus/prometheus/pull/6838 , which will be added in Prometheus v2.21.
See https://github.com/prometheus/prometheus/releases/tag/v2.21.0-rc.1
* Added endpointslices discovery to k8s api
Started from 1.17 k8s version endpointslices is beta,
it allows to query k8s api for endpoints more efficient.
It presents at scrape_config.yaml as separate role for kubernetes_sd_config.
kubernetes_sd_config:
- role: endpointslices
* fixed typos, changed EndpointConditions signature - with values instead of pointers
This should reduce disk space usage when scraping targets containing metrics with identical names
such as `node_cpu_seconds_total`, histograms, quantiles, etc.
Expose `vm_timestamps_blocks_merged_total` and `vm_timestamps_bytes_saved_total` metrics for monitoring
the effectiveness of timestamp blocks merging.
* revise /api/v1/series docs
Further clarification for #735
* clarify how default range differers from Prometheus API
* avoid `start=0` suggestion when confirming delete, because it will cause a timeout in most deployments
* Update README.md
aws sdk has complicated logic for chosing profile name and we shouldn't set
it to `default` value. It leads to bugs and improper configuration.
Set it to empty value by default is safe. It will be automatically set to `default` by sdk.
These functions returns the number of raw samples that don't exceed `le` or are bigger than `gt`.
These functions are complement to already existing `share_le_over_time(m[d], le)` and `share_gt_over_time(m[d], gt)`.
* VMAlert start with empty rules dir
There are some applications (operator for instance), that generates alerts configuration at runtime
and vmalert must start correctly without rules to support this behaviour.
Later application will add rules files and send SIGHUP to vmalert,
which will trigger reading rules files and start rules exectuion.
Removing rules files with SIGHUP signal must stop rules execution and
vmalert will wait for new rules.
* imports sorted
* added test cases for empty rules, removed blank line
* fixed imports conflict
* updated tests
* feat: spread load of rule evaluation by group when starting new groups
* review: reduce the resulting diff.
* Update app/vmalert/group.go
Co-authored-by: Roman Khavronenko <hagen1778@gmail.com>
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
Co-authored-by: Roman Khavronenko <hagen1778@gmail.com>
The previous basis on `cap(sw.labels)` doesn't work anymore after 7785869ccc ,
because `sw.labels` may be reset multiple times when processing big number of rows.
* dashboard: rename var `datasource` to `ds` for consistency reason
Dasbhoards for cluster version or vmagent operate with datasource variable
named `ds`. For consistency sake we rename this variable in single node version
as well.
* dashboard: add instance variable picker
See dashboard reviews here https://grafana.com/grafana/dashboards/10229/reviews
* dashboard: limit number of buckets in histogram to 12 for vmagent dashboard
* dashboard: bump version requirement in description for single version
* dashboard: drop extra series override for single version
* dashboard: set Y-min to zero for most of panels in vmagent dashboard
Description for `-rule` flag uses as example specific chars like asterisks
which could be interpreted wrong by different shells. To avoid this, description
now contains quoted flag values.
See also #708
Arch 386 is a 32-bit architecture and interprets int type for numbers as an explicit int32,
whereas on most modern CPUs int is implicitly an int64. This makes tests to fail with
`int overflow` error.
Previously targets with big number of metrics and/or labels could generated too big buffers,
which then could be re-used when scraping targets with small number of metrics.
This resulted in memory waste.
Now big buffers are used only for targets with big number of metrics / labels,
while small buffers are used for targets with small number of metrics / labels.
The previous notion was inconsistent with what `decimal.Round` does.
According to [wiki](https://en.wikipedia.org/wiki/Significant_figures) rounding
applied to all significant figures, not just decimal ones.
* app/vmalert: extend metrics set exported by `vmalert` #573
New metrics were added to improve observability:
+ vmalert_alerts_pending{alertname, group} - number of pending alerts per group
per alert;
+ vmalert_alerts_acitve{alertname, group} - number of active alerts per group
per alert;
+ vmalert_alerts_error{alertname, group} - is 1 if alertname ended up with error
during prev execution, is 0 if no errors happened;
+ vmalert_recording_rules_error{recording, group} - is 1 if recording rule
ended up with error during prev execution, is 0 if no errors happened;
* vmalert_iteration_total{group, file} - now contains group and file name labels.
This should improve control over specific groups;
* vmalert_iteration_duration_seconds{group, file} - now contains group and file name labels. This should improve control over specific groups;
Some collisions for alerts and recording rules are possible, because neither
group name nor alert/recording rule name are unique for compatibility reasons.
Commit contains list of TODOs for Unregistering metrics since groups and rules
are ephemeral and could be removed without application restart. In order to
unlock Unregistering feature corresponding PR was filed - https://github.com/VictoriaMetrics/metrics/pull/13
* app/vmalert: extend metrics set exported by `vmalert` #573
The changes are following:
* add an ID label to rules metrics, since `name` collisions within one group is
a common case - see the k8s example alerts;
* supports metrics unregistering on rule updates. Consider the case when one rule
was added or removed from the group, or the whole group was added or removed.
The change depends on https://github.com/VictoriaMetrics/metrics/pull/16
where race condition for Unregister method was fixed.
Previously the limit has been raised to GOMAXPROCS, but it has been appeared that this
increases query latencies since more CPUs are busy with merges.
While at it, substitute `*MergeConcurrencyLimitCh` channels with simple integer limits.
Previously the `vm_slow_row_inserts_total` metric may be incremented multiple times for different data points per a single time series,
while only a single increment is needed when inserting the first data point for this time series.
`external.label` flag supposed to help to distinguish alert or recording rules
source in situations when more than one `vmalert` runs for the same datasource
or AlertManager.
due to memory align the metaindexRow structure use 64-byte pre object.
this commit changes the order of field, make metaindexRow use 56-byte pre
object.
Signed-off-by: Sasasu <su@sasasu.me>
This function works with both Prometheus-style and VictoriaMetrics-style buckets.
The function removes buckets with the lowest values in order to reserve the highest precision.
The function is useful for building heatmaps in Grafana from too big number of buckets.
Previously the time spent on inverted index search could exceed the configured `-search.maxQueryDuration`.
This commit stops searching in inverted index on query timeout.
Previously the enabled relabeling with `-relabelConfig` command-line flag could result in missing labels
if a single Influx line protocol message contains multiple field values.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/638
This condition may occur after the following sequence of events:
1) A goroutine enters the loop body when len(addRowsConcurrencyCh) == cap(addRowsConcurrencyCh) inside Storage.searchTSIDs.
2) All the goroutines return from Storage.AddRows.
3) The goroutine from step 1 blocks on searchTSIDsCond.Wait() inside the loop body.
The goroutine remains blocked until the next call to Storage.AddRows, which calls searchTSIDsCond.Signal().
This may take indefinite time.
This should prvent from inflated `increase()` results for time series that start from big initial values.
Such cases may occur when a label value changes in a metric without counter reset.
`vmagent` Grafana dashboard suppose to provide basic observability over multiple
`vmagent` instances. Dashboard is saved in Grafana export format so it can be easily
imported. It was also integrated into docker-compose environment.
This is a follow-up commit after 12b16077c4 ,
which didn't reset the `tsidCache` in all the required places.
This could result in indefinite errors like:
missing metricName by metricID ...; this could be the case after unclean shutdown; deleting the metricID, so it could be re-created next time
Fix this by resetting the cache inside deleteMetricIDs function.
* add vm_protoparser_rows_read_total metrics to promscrape
move vm_protoparser_rows_read_total for promscrape to better place
move vm_protoparser_rows_read_total for promscrape to better place
* remove possibility of infinity loop at prometheus parser
Array type flag is now defined as `value` type in flag description when printed.
This change adds additional description to every Array type flag so it would be
clear what exact type is used:
```
-remoteWrite.urlRelabelConfig array
Optional path to relabel config for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
```
Metric `vm_persistentqueue_bytes_pending` is a gauge that shows current amount
of bytes in persistentqueue flushed on disk as a difference between write and read
offsets. This metric is very similar to `vmagent_remotewrite_pending_data_bytes`
except of accounting for bytes in-memory.
The change to `vm_promscrape_targets` metric suppose to improve observability
for `vmagent` so it will be possible to track how many targets are up or down
for every specific scrape group:
```
vm_promscrape_targets{type="static_configs", status="down"} 1
vm_promscrape_targets{type="static_configs", status="up"} 2
```
Previously multiple goroutines could access remoteWriteCtx.tss concurrently, which could lead to data race
and improper relabeling. Now each goroutine has its own copy of tss during relabeling.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/599
Previously the concurrency has been limited to GOMAXPROCS*2. This had little sense,
since every call to Storage.AddRows is bound to CPU, so the maximum ingestion bandwidth
is achieved when the number of concurrent calls to Storage.AddRows is limited to the number of CPUs,
i.e. to GOMAXPROCS.
Heavy queries could result in the lack of CPU resources for processing the current data ingestion stream.
Prevent this by delaying queries' execution until free resources are available for data ingestion.
Expose `vm_search_delays_total` metric, which may be used in for alerting when there is no enough CPU resources
for data ingestion and/or for executing heavy queries.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/291
* app/vmalert: add retries to remotewrite
Remotewrite pkg now does limited number of retries if write request failed.
This suppose to make vmalert state persisting more reliable.
New metrics were added to remotewrite in order to track rows/bytes sent/dropped.
defaultFlushInterval was increased from 1s to 5s for sanity reasons.
* fix
* wip
* wip
* wip
* fix bits alignment bug for 32-bit systems
* fix mistakenly dropped field
The maximum deadline miss duration is reduced to 2x scrape_interval in the worst case.
By default it is limited to scrape_interval configured for the given scrape target.
* Fix Auto metrics relabeled errors
* Finalize auto-genenated Labels
* Fix Test Errors
* fix error logs when queue is full
Co-authored-by: xinyulong <xinyulong@kuaishou.com>
VictoriaMetrics returns `503 Service Unavailable` http error for requests with time ranges outside the configured retention
if `-denyQueriesOutsideRetention` command-line flag is set.
* app/vmalert: support multiple notifier urls (#584)
User now can set multiple notifier URLs in the same fashion
as for other vmutils (e.g. vmagent). The same is correct for
TLS setting for every configured URL. Alerts sending is done
in sequential way for respecting the specified URLs order.
* app/vmalert: add basicAuth support for notifier client (#585)
The change adds possibility to set basicAuth creds for notifier
client in the same fasion as for remote write/read and datasource.
vmagent replaces Prometheus to perform scrapes and writes
into VictoriaMetrics installation. Prometheus datasource was
dropped, but its config was reused to feed vmagent.
Change also contains simplification in dashboard propagation
to Grafana container by removing excessive json manipulation
steps.
The change adds no new functionality and aims to move flags definitions
to subpackages that are using them. This should improve readability
of the main function.
As per documentation on `label_replace` function: "If the regular
expression doesn't match then the timeseries is returned unchanged".
Currently this behavior is not enforced, if a regexp on an existing
tag doesn't match then the tag value is copied as-is in the destination
tag. This fix first checks that the regular expression matches the
source tag before applying anything.
Given the current implementation, this fix also changes the behavior
of the **MetricsQL** `label_transform` function which does not
document this behavior at the moment.
Previously VictoriaMetrics was processing up to 32 time series in a single goroutine.
This could be slow if each time series contains big number of data points (10M+ or more), since only a single CPU core could be loaded with work,
while other CPU cores were idle. Fix this by launching GOMAXPROCS workers for time series processing.
This should help with https://github.com/VictoriaMetrics/VictoriaMetrics/issues/572
app/vmalert: add support for TLS configuration
Add support for TLS optional configuration in a similar fashion to what
is currently supported in other vmutils such as vmagent. TLS
configuration options are distinct for datasource, remoteRead,
remoteWrite as well as notifier.
These actions may be useful for filtering out unneeded targets and/or metrics if they contain equal label values.
For example, the following rule would leave the target only if __meta_kubernetes_annotation_prometheus_io_port
equals __meta_kubernetes_pod_container_port_number:
- action: keep_if_equal
source_labels: [__meta_kubernetes_annotation_prometheus_io_port, __meta_kubernetes_pod_container_port_number]
app/vmalert: Support custom URL for alerts source
Add flag `external.alert.source` for configuring custom URL
for alert's source. This may be handy to re-point default source
URL to other systems like Grafana.
Updates #517
Uniqueness of rule is now defined by combination of its name, expression and
labels. The hash of the combination is now used as rule ID and identifies rule within the group.
Set of rules from coreos/kube-prometheus was added for testing purposes to
verify compatibility. The check also showed that `vmalert` doesn't support
`query` template function that was mentioned as limitation in README.
The feature allows to speed up group rules execution by
executing them concurrently.
Change also contains README changes to reflect configuration
details.
* vmalert: Add recording rules support.
Recording rules support required additional service refactoring since
it wasn't planned to support them from the very beginning. The list
of changes is following:
* new entity RecordingRule was added for writing results of MetricsQL
expressions into remote storage;
* interface Rule now unites both recording and alerting rules;
* configuration parser was moved to separate package and now performs
more strict validation;
* new endpoint for listing all groups and rules in json format was added;
* evaluation interval may be set to every particular group;
* vmalert: uncomment tests
* vmalert: rm outdated TODO
* vmalert: fix typos in README
v1.36.0 always returns empty responses for Graphite wildcards like the following
{__name__=~"foo\\.[^.]*\\.bar\\.baz"}
Temporary workaround for v1.36.0 is to add `[^.]*` to the end of the regexp.
Add index for reverse Graphite-like metric names with dots. Use this index during search for filters
like `__name__=~"foo\\.[^.]*\\.bar\\.baz"` which end with non-empty suffix with dots, i.e. `.bar.baz` in this case.
This change may "hide" historical time series during queries. The workaround is to add `[.]*` to the end of regexp label filter,
i.e. "foo\\.[^.]*\\.bar\\.baz" should be substituted with "foo\\.[^.]*\\.bar\\.baz[.]*".
* docs/Single-server-VictoriaMetrics.md: add link to Wiki page so it may get more attention
* docs/Single-server-VictoriaMetrics.md: mention case for changing `-retentionPeriod` setting
* Slow metrics load panel was removed since it is hard to interpret without
additional metrics and stats;
* Slow inserts panel was updated to display percentage of slow inserts comparing
to total number of inserts to show the real impact.
The new update introduces new row "Troubleshooting" that
contains panels for churn rate and slow-queries/inserts/loads metrics. This row supposed to be reveal the cause of low performance or other issues.
Panels for storage were updated with "bytes-per-datapoint" and "remaining disk size" panels.
Before the change we were sending notifications to notifier
if following conditions are met:
* alert is in Fire state
* alert is in Inactive state
We were sending Inactive notifications to resolve alert ASAP.
Unfortunately, we were sending resolves for Pending alerts that become
Inactive, which is wrong.
In this change we delete alert from the active list if
it was Pending and become Inactive. In this way we now
have Inactive alerts only if they were in state Fire before.
See test change for example.
During group's update rules deletion was causing slice
mutations while slice index was assumed to be unchanged.
This caused "slice bounds out of range" errors when multiple
rules were deleted sequentially.
The check for non-nil remoteRead was mistakenly dropped
during refactoring which caused panics when `vmalert`
wasn't configured with `remoteRead` flag.
* feat: add remoteWrite.maxQueueSize to reduce queue full
* rename remote(write|read) flags to remote(Write|Read) for the sake of consistency
Co-authored-by: xiaobeibei <xiaobeibei@bigo.sg>
The change introduces new entity `manager` which replaces
`watchdog`, decouples requestHandler and groups. Manager
supposed to control life cycle of groups, rules and
config reloads.
Groups export an ID method which returns a hash
from filename and group name. ID supposed to be unique
identifier across all loaded groups.
Some tests were added to improve coverage.
Bug with wrong annotation value if $value is used in
templates after metrics being restored fixed.
Notifier interface was extended to accept context.
New set of metrics was introduced for config reload.
The http server returns 503 non-OK error at `/health` page during grace period,
so load balancers in front of the http server could re-route incoming requests
to other servers.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/463
Previously the duration for graceful shutdown for http server could take more than a minute
because of imporperly set timeouts in setNetworkTimeout.
Now typical duration for graceful shutdown should be reduced to less than 5 seconds.
* app/vmalert: restore alerts state from datasource metrics
Vmalert will restore alerts state for rules that have `rule.For` > 0 from previously written timeseries via `remotewrite.url` flag.
* app/vmalert: mention remotewerite and remoteread configuration in README
The upstream fasthttp may contain issues like 996610f021 ,
plus a code that isn't used by VictoriaMetrics. So let's use a private copy under our control instead.
Newly added index entries can be missing after unclean shutdown, since they didn't flush to persistent storage yet.
Log about this and delete the corresponding metricID, so it could be re-created next time.
* app/vmalert: initial remote-write support for alerts state persistence.
If `remotewrite.url` flag is set, vmalert will send alerts state via remote-write protocol to remote storage. The sending is asynchronous to avoid blocking calls in rules evaluation loop.
* app/vmalert: merge with master
* app/vmalert: write both `instant` and `for` alerts timeseries states in remote storage.
This eliminates the need for storing block data into temporary files on a single-node VictoriaMetrics
during heavy queries, which touch big number of time series over long time ranges.
This improves single-node VM performance on heavy queries by up to 2x.
Now it leaves only the first data point on each `-dedup.minScrapeInterval` interval.
Previously it may leave two data points on the interval. This could lead to unexpected results
for `histogram_quantile(phi, sum(rate(buckets)) by (le))` query.
This should reduce the frequency of the following errors:
cannot find tag filter matching less than N time series; either increase -search.maxUniqueTimeseries or use more specific tag filters
more than N time series found on the time range [...]; either increase -search.maxUniqueTimeseries or shrink the time range
This should fix the following error when sending data to remote storage:
couldn't send a block with size XX bytes to "YYY": the server closed connection before returning the first response byte. Make sure the server returns 'Connection: close' response header before closing the connection
This should fix the following error:
the server closed connection before returning the first response byte. Make sure the server returns 'Connection: close' response header before closing the connection
This case is possible when the corresponding metricID->metricName entry didn't propagate to inverted index yet.
This should fix the following error:
error when searching tsids for tfss [...]: cannot find metricName by metricID 1582417212213420669: EOF
This is a common error whith improperly configured target autodiscovery and/or relabeling.
This error leads to duplicate scraping of the same targets with the same set of labels, which leads
to duplicate samples in time series.
Previously targets with duplicate scrape urls were merged into a single line on the page.
Now each target with duplicate scrape url is displayed on a separate line.
This should reduce CPU load during scraping when target discovery generates
big number of `__meta_*` labels (for instance, k8s discovery).
See https://www.robustperception.io/life-of-a-label for details.
vmalert: Extend web responses for alerts
* populate apiAlert object with additional fields
* return all active alerts, not only firing
* sort list of API alerts for deterministic output
* add helper for available path list
This should prevent from cache stats counters going down after cache rotation,
which may corrupt `cache hit ratio` graph on the official Grafan dasbhoards
when using the following query:
1 - (sum(rate(vm_cache_misses_total[5m])) by (type) / sum(rate(vm_cache_requests_total[5m])) by (type))
* Initial rules evaluation support.
Rules are now store alerts state in private field `alerts`. Every evaluation updates
the alerts and state. Every unique metric received from datastore represents a unique alert,
uniqueness is guaranteed by hashing ordered labelset.
* merge with master
* cleanup
* support endAt parameter as 3*evaluationInterval for active alerts
* make golint happy
- Sort tag filters in the ascending number of matching time series
in order to apply the most specific filters first.
- Fall back to metricName search for filters matching big number of time series
(usually this are negative filters or regexp filters).
If docker app is upgraded from root to non-root, then the data pointed by `-storageDataPath` or similar flags
becomes denied to non-root user after the upgrade. This breaks upgrade path. So revert back to default root user
for docker apps.
Users may explicitly execute `docker run --user <non_root_user>` for running docker apps under non-root user.
The way how regex for column style in Table panel should be applied has changed in 6.7 Grafana version. The change supposed to fix Flags panel column styles accordingly.
* Remember the last accessed bucket on Has() call.
* Inline fast paths inside Add() and Has() calls.
* Remove fragile code with maxUnsortedBuckets inside bucket32.
This guarantees that the snapshot contains all the recently added data
from inmemory buffers when multiple concurrent calls to Storage.CreateSnapshot are performed.
The following corner cases now supported:
* label_map(q, "label", "", "foo") - adds `label="foo"` to series with missing `label`
* label_map(q, "label", "foo", "") - removes `label="foo"` from series
All the unmatched labels are kept unchanged.
Prometheus docs at https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config say:
> In communication with external systems, they are always applied only
> when a time series does not have a given label yet and are ignored otherwise.
Though this may result in consistency chaos when scrape targets override `external_labels`,
let's stick with Prometheus behavior for the sake of backwards compatibility.
There is last resort in vmagent with `-remoteWrite.label`, which consistently
sets the configured labels to all the metrics before sending them to remote storage.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/366
This should prevent from the following unexpected side-effects of idempotent request retries:
- increased actual timeout when scraping the target comparing to the configured scrape_timeout
- increased load on the target
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/357
This flag can be used for removing gaps on graphs if the difference between the current time
and the timestamps from the ingested data exceeds 5 minutes.
This is the case when the time between data sources and VictoriaMetrics is improperly synchronized.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/312
This should reduce CPU usage when adding new entries to inverted index.
This should alos prevent from creating stalled cleaner goroutines for the created temporary parts,
since they were never closed.
This should fix the following issue: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/316 .
A single heavy request can saturate all the available CPUs, so let's limit the number of concurrent requests to lower value.
This will give more chances for executing insert path.
This makes the behaviour for these functions similar to Prometheus when processing broken time series with irregular data points
like `gitlab_runner_jobs`. See https://gitlab.com/gitlab-org/gitlab-exporter/issues/50 for details.
This function can be used for simultaneous calculating of multiple `aggr_func*` functions
that accept range vector. For example, `aggr_over_time(("min_over_time", "max_over_time"), m[d])`
would calculate `min_over_time` and `max_over_time` for `m[d]`.
This should prevent from restoring from incomplete backups.
Add `-skipBackupCompleteCheck` command-line flag to `vmrestore` in order to be able restoring from old backups without `backup complete` file.
This should prevent from the following panic at app/vmselect/promql/binary_op.go:255:
BUG: len(leftVaues) must match len(rightValues) and len(dstValues)
The full list of functions added:
- `topk_min(k, q)` - returns top K time series with the max minimums on the given time range
- `topk_max(k, q)` - returns top K time series with the max maximums on the given time range
- `topk_avg(k, q)` - returns top K time series with the max averages on the given time range
- `topk_median(k, q)` - returns top K time series with the max medians on the given time range
- `bottomk_min(k, q)` - returns bottom K time series with the min minimums on the given time range
- `bottomk_max(k, q)` - returns bottom K time series with the min maximums on the given time range
- `bottomk_avg(k, q)` - returns bottom K time series with the min averages on the given time range
- `bottomk_median(k, q)` - returns bottom K time series with the min medians on the given time range
`runTransactions` call issues async deletions for transaction files. The previously issued transaction deletions
can race with the next call to `runTransactions`. Prevent this by waiting until all the pending transaction
deletions are funished in the beginning of `runTransactions`. Also make sure that all the pending transaction
deletions are finished before returning from `runTransactions`.
This should fix the issue on NFS when incompletely removed dirs may be left
after unclean shutdown (OOM, kill -9, hard reset, etc.), while the corresponding transaction
files are already removed.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/162
The metricID->metricName entry can be missing in the indexdb after unclean shutdown
when only a part of entries for new time series is written into indexdb.
Recover from such a situation by removing the broken metricID. New metricID
will be automatically created for time series with the given metricName
when new data point will arive to it.
This fixes an issue when VictoriaMetrics doesn't see the restored data after the following operations:
1. Stop VictoriaMetrics.
2. Delete `<-storageDataPath>` dir.
3. Start VictoriaMetrics, then stop it.
4. Restore data from backup with `vmrestore`.
5. Start VictoriaMetrics.
`vmrestore` didn't delete properly empty dirs in `<-storageDataPath>/indexdb` because of the remaining `flock.lock` files in these dirs.
`start` arg has higher chances to be set properly comparing to `end` arg,
so it is expected that the `end` arg could be adjusted if it was set incorrectly.
See the corresponding benchmark in Prometheus - 23c0299d85/tsdb/head_bench_test.go (L52)
The benchmark allows performing apples-to-apples comparison of time series search
in Prometheus and VictoriaMetrics. The following article - https://www.robustperception.io/evaluating-performance-and-correctness -
contains incorrect numbers for VictoriaMetrics, since there wasn't this benchmark yet. Fix this.
Benchmarks can be repeated with the following commands from Prometheus and VictoriaMetrics source code roots:
- Prometheus: GOMAXPROCS=1 go test ./tsdb/ -run=111 -bench=BenchmarkHeadPostingForMatchers
- VictoriaMetrics: GOMAXPROCS=1 go test ./lib/storage/ -run=111 -bench=BenchmarkHeadPostingForMatchers
Benchmark results:
benchmark old ns/op new ns/op delta
BenchmarkHeadPostingForMatchers/n="1" 272756688 364977 -99.87%
BenchmarkHeadPostingForMatchers/n="1",j="foo" 138132923 1181636 -99.14%
BenchmarkHeadPostingForMatchers/j="foo",n="1" 134723762 1141578 -99.15%
BenchmarkHeadPostingForMatchers/n="1",j!="foo" 195823953 1148056 -99.41%
BenchmarkHeadPostingForMatchers/i=~".*" 7962582919 8716755 -99.89%
BenchmarkHeadPostingForMatchers/i=~".+" 7589543864 12096587 -99.84%
BenchmarkHeadPostingForMatchers/i=~"" 1142371741 16164560 -98.59%
BenchmarkHeadPostingForMatchers/i!="" 9964150263 12230021 -99.88%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",j="foo" 216995884 1173476 -99.46%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",i!="2",j="foo" 202541348 1299743 -99.36%
BenchmarkHeadPostingForMatchers/n="1",i!="" 486285711 11555193 -97.62%
BenchmarkHeadPostingForMatchers/n="1",i!="",j="foo" 350776931 5607506 -98.40%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",j="foo" 380888565 6380335 -98.32%
BenchmarkHeadPostingForMatchers/n="1",i=~"1.+",j="foo" 89500296 2078970 -97.68%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!="2",j="foo" 379529654 6561368 -98.27%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!~"2.*",j="foo" 424563825 6757132 -98.41%
The first column (old) is for Prometheus, the second column (new) is for VictoriaMetrics.
As you can see, VictoriaMetrics outperforms Prometheus by more than 100x in almost all the test cases of this benchmark.
Prometheus was using 3.5GB of RAM during the benchmark, while VictoriaMetrics was using 400MB of RAM.
The current day could miss entries for already stopped time series before
enabling per-day index.
This fixes the issue when queries return empty results during the first hour after
upgrading to v1.29.*
Production workload shows that the index requires ~4Kb of RAM per active time series.
This is too much for high number of active time series, so let's delete this index.
Now the queries should fall back to the index for the current day instead of the index
for the recent hour. The query performance for the current day index should be good enough
given the 100M rows/sec scan speed per CPU core.
Issues fixed:
- Slow startup times. Now the index is loaded from cache during start.
- High memory usage related to superflouos index copies every 10 seconds.
Production load with >10M active time series showed it could
slow down VictoriaMetrics startup times and could eat
all the memory leading to OOM.
Remove inmemory inverted index for recent hours until thorough
testing on production data shows it works OK.
Continue trying to remove NFS directory on temporary errors for up to a minute.
The previous async removal process breaks in the following case during VictoriaMetrics start
- VictoriaMetrics opens index, finds incomplete merge transactions and starts replaying them.
- The transaction instructs removing old directories for parts, which were already merged into bigger part.
- VictoriaMetrics removes these directories, but their removal is delayed due to NFS errors.
- VictoriaMetrics scans partition directory after all the incomplete merge transactions are finished
and finds directories, which should be removed, but weren't still removed due to NFS errors.
- VictoriaMetrics panics when it finds unexpected empty directory.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/162
Incremental aggregate functions don't keep all the selected time series in memory -
they keep only up to GOMAXPROCS time series for incremental aggregations.
Take into account that the number of time series in RAM can be higher if they are split
into many groups with `by (...)` or `without (...)` modifiers.
This should reduce the number of `not enough memory for processing ... data points` false
positive errors.
The origin of the error has been detected and documented in the code,
so it is enough to export a counter for such errors at `vm_index_blocks_with_metric_ids_incorrect_order_total`,
so it could be monitored and alerted on high error rates.
Export also the counter for processed index blocks with metricIDs - `vm_index_blocks_with_metric_ids_processed_total`,
so its' rate could be compared to `rate(vm_index_blocks_with_metric_ids_incorrect_order_total)`.
Slow loops could require seeks and expensive regexp matching, while fast loops just scans
all the metricIDs for the given `tag=value` prefix. So these operations must have separate
max loops multiplier.
The fastest tag filters are non-negative non-regexp, since they are the most specific.
The slowest tag filters are negative regexp, since they require scanning
all the entries for the given label.
This flag is similar to `-search.lookback-delta` if set. The max lookback interval is determined dynamically
from interval between datapoints for each input time series if the flag isn't set.
The interval can be overriden on per-query basis by passing `max_lookback=<duration>` query arg to `/api/v1/query` and `/api/v1/query_range`.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/209
`increases_over_time(q[d])` returns the number of `q` increases during the given duration `d`.
`decreases_over_time(q[d])` returns the number of `q` decreases during the given duration `d`.
This should improve inverted index search performance for filters matching big number of time series,
since `lib/uint64set.Set` is faster than `map[uint64]struct{}` for both `Add` and `Has` calls.
See the corresponding benchmarks in `lib/uint64set`.
This should improve lookup performance if the same `label=value` pair exists
in big number of time series.
This should also reduce memory usage for mergeset data cache, since `tag->metricIDs` rows
occupy less space than the original `tag->metricID` rows.
The number of MetricID->TSID entries is smaller than the number of tag->MetricID entries
and MetricID->TSID entries are usually shorter than tag->MetricID entries.
This should improve performance when selecting all the metricIDs.
The follosing issues were fixed:
- VictoriaMetrics could leave superflouos labels when using `on` or `ignoring` modifiers
- VictoriaMetrics could return `duplicate timeseries` error when using `group_left` or `group_right` with non-empty label list
Do not give up after 11 attempts of directory removal on laggy NFS.
Add `vm_nfs_dir_remove_failed_attempts_total` metric for counting the number of failed attempts
on directory removal.
Log failed attempts on directory removal after long sleep times.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/162
The maximum inmemory file size now depends on `-memory.allowedPercent`.
This should improve performance and reduce the number of filesystem calls
on machines with big amounts of RAM when performing heavy queries
over big number of samples and time series.
The commend became outdated after the commit ed6ac1a5df027f0dfc22448e3b27c26b6f77c67a,
which stops merging of small parts on graceful shutdown instead of waiting
for their completion.
This function is useful for determining time series lifetime.
`d` must exceed the expected lifetime of the time series, otherwise
the function would return values close to `d`.
This should reduce the amount of RAM required for processing time series
with non-zero churn rate.
The previous cache behavior can be restored with `-cache.oldBehavior` command-line flag.
Track also the number of dropped rows due to the exceeded timeout
on concurrency limit for Storage.AddRows. This number is tracked in `vm_concurrent_addrows_dropped_rows_total`
Real-world production data shows higher jitter than 1/8 of scrape interval.
This may results in gaps on the graph. So increase the allowed jitter to 1/4
of scrape interval in order to reduce the probability of gaps on the graphs
over time series with high jitter for scrape_interval.
Default rollup function is `last_over_time`. It must support adjusting
the provided window in order to prevent from gaps on the graph
for window values smaller than scrape interval.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/134
- Use `-prod` binaries instead of development binaries for both deb and rpm packages.
- Fix binary directory from /usr/sbin to /usr/local/bin as outlined in package/victoria-metrics.service
- Fix binary name from `victoriametrics` to `victoria-metrics-prod` in package/victoria-metrics.service
* initial commit of deb packaging
* Incorporated feedback from @valyala:
- Put data directory under /var/lib
- More beef in systemd file
- Packaging for arm64
- Package all target which builds and packages both amd64 and arm64
* Remove PIDFile from systemd unit, useless
* per PR feedback, move debian specific files into deb subdirectory
Updates #107 .
- Series with negative values are always gauges
- Counters may only have increasing values with possible counter resets
This should improve compression ratio for gauge series which
were previously mistakenly detected as counters.
Calculate incremental aggregates for `aggr(metric_selector)` function instead of
keeping all the time series matching the given `metric_selector` in memory.
- Do not overwrite last points by the previous NaNs, since this may result in empty time series.
- Overwrite the last 2 points instead of 3. This should be enough in most cases.
Data in ExtDB cannot be changed, so it is OK to use unversioned keys for tag cache.
This should improve performance for index lookups over big amount of time series.
`deriv_fast` calculates derivative based on the first and the last point on the interval
instead of calculating linear regression based on all the data points on the interval.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/73
Prometheus establishes a connection per shard in remote_write config.
By default it establishes up to 1000 connections to remote storage (max_shards: 1000).
This is quite big, so set `max_shards: 100` in the recommmended Prometheus config.
This should improve performance after restart when the db contains a lot of time series
with high time series churn (i.e. metrics from Kubernetes with many pods and frequent deployments)
Unremoved directories may lead to inconsistent data directory,
so VictoriaMetrics will fail to start next time.
So panic on the first error when trying to remove directory in order
to simplify recover process.
This should improve speed for searching metrics among high number of time series
with high churn rate like in big Kubernetes clusters with frequent deployments.
ReadLinesBlock may accept dstBuf with non-zero length. In this case the last line without trailing newline isn't read.
Fix this by comparing len(dstBuf) to 0 instead of its original length.
Add sigint as stopsignal to docker file. You can find more here: https://docs.docker.com/engine/reference/builder/#usage
With this change, the main process inside the container will receive SIGINT, and after a grace period, SIGKILL.
Previously arbitrary time series could be returned from `quantile`
depending on sort order for the last data point in the selected range.
Fix this by returning the calculated time series.
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/55
- Increase update iterval from 1s to 10s. This should reduce CPU usage
for large amounts of metric ids with constant churn.
- Reduce pendingTodayMetricIDsLock lock duration during the update.
A clear and concise description of what the bug is.
It would be great [upgrading](https://victoriametrics.github.io/#how-to-upgrade) to [the latest avaialble release](https://github.com/VictoriaMetrics/VictoriaMetrics/releases)
and verifying whether the bug is reproducible there.
It is also recommended reading [troubleshooting docs](https://victoriametrics.github.io/#troubleshooting).
**To Reproduce**
Steps to reproduce the behavior.
**Expected behavior**
A clear and concise description of what you expected to happen.
**Screenshots**
If applicable, add screenshots to help explain your problem.
**Version**
The line returned when passing `--version` command line flag to binary. For example:
Before filling a bug report it would be great to [upgrade](https://docs.victoriametrics.com/victoriametrics/single-server-victoriametrics/#how-to-upgrade)
to [the latest available release](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/latest)
and verify whether the bug is reproducible there.
It's also recommended to read the [troubleshooting docs](https://docs.victoriametrics.com/victoriametrics/troubleshooting/) first.
- type:textarea
id:describe-the-bug
attributes:
label:Describe the bug
description:|
A clear and concise description of what the bug is.
placeholder:|
When I do `A` VictoriaMetrics does `B`. I expect it to do `C`.
validations:
required:true
- type:textarea
id:to-reproduce
attributes:
label:To Reproduce
description:|
Steps to reproduce the behavior.
If reproducing an issue requires some specific configuration file, please paste it here.
placeholder:|
Steps to reproduce the behavior.
validations:
required:true
- type:textarea
id:version
attributes:
label:Version
description:|
The line returned when passing `--version` command line flag to the binary. For example:
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.
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/).
httpListenAddr=flag.String("httpListenAddr",":8428","TCP address to listen for http connections")
minScrapeInterval=flag.Duration("dedup.minScrapeInterval",0,"Remove superflouos samples from time series if they are located closer to each other than this duration. "+
"This may be useful for reducing overhead when multiple identically configured Prometheus instances write data to the same VictoriaMetrics. "+
"Deduplication is disabled if the -dedup.minScrapeInterval is 0")
dryRun=flag.Bool("dryRun",false,"Whether to check only -promscrape.config and then exit. "+
"Unknown config entries are allowed in -promscrape.config by default. This can be changed with -promscrape.config.strictParse")
)
funcmain(){
// Write flags and help message to stdout, since it is easier to grep or pipe.
flag.CommandLine.SetOutput(os.Stdout)
flag.Usage=usage
envflag.Parse()
buildinfo.Init()
logger.Init()
ifpromscrape.IsDryRun(){
*dryRun=true
}
if*dryRun{
iferr:=promscrape.CheckConfig();err!=nil{
logger.Fatalf("error when checking -promscrape.config: %s",err)
}
logger.Infof("-promscrape.config is ok; exitting with 0 status code")
return
}
logger.Infof("starting VictoriaMetrics at %q...",*httpListenAddr)
See vmagent docs [here](https://docs.victoriametrics.com/victoriametrics/vmagent/).
## vmagent
vmagent docs can be edited at [docs/vmagent.md](https://github.com/VictoriaMetrics/VictoriaMetrics/blob/master/docs/victoriametrics/vmagent.md).
`vmagent` is a tiny but mighty agent which helps you collect metrics from various sources
and store them in [VictoriaMetrics](https://github.com/VictoriaMetrics/VictoriaMetrics)
or any other Prometheus-compatible storage systems that support the `remote_write` protocol.
<img alt="vmagent" src="vmagent.png">
## Motivation
While VictoriaMetrics provides an efficient solution to store and observe metrics, our users needed something fast
and RAM friendly to scrape metrics from Prometheus-compatible exporters into VictoriaMetrics.
Also, we found that our user's infrastructure are like snowflakes in that no two are alike. Therefore we decided to add more flexibility
to `vmagent` such as the ability to push metrics instead of pulling them. We did our best and will continue to improve vmagent.
## Features
* Can be used as a drop-in replacement for Prometheus for scraping targets such as [node_exporter](https://github.com/prometheus/node_exporter).
See [Quick Start](#quick-start) for details.
* Can add, remove and modify labels (aka tags) via Prometheus relabeling. Can filter data before sending it to remote storage. See [these docs](#relabeling) for details.
* Accepts data via all ingestion protocols supported by VictoriaMetrics:
* Influx line protocol via `http://<vmagent>:8429/write`. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-send-data-from-influxdb-compatible-agents-such-as-telegraf).
* Graphite plaintext protocol if `-graphiteListenAddr` command-line flag is set. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-send-data-from-graphite-compatible-agents-such-as-statsd).
* OpenTSDB telnet and http protocols if `-opentsdbListenAddr` command-line flag is set. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-send-data-from-opentsdb-compatible-agents).
* Prometheus remote write protocol via `http://<vmagent>:8429/api/v1/write`.
* JSON lines import protocol via `http://<vmagent>:8429/api/v1/import`. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-import-data-in-json-line-format).
* Native data import protocol via `http://<vmagent>:8429/api/v1/import/native`. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-import-data-in-native-format).
* Data in Prometheus exposition format. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-import-data-in-prometheus-exposition-format) for details.
* Arbitrary CSV data via `http://<vmagent>:8429/api/v1/import/csv`. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-import-csv-data).
* Can replicate collected metrics simultaneously to multiple remote storage systems.
* Works smoothly in environments with unstable connections to remote storage. If the remote storage is unavailable, the collected metrics
are buffered at `-remoteWrite.tmpDataPath`. The buffered metrics are sent to remote storage as soon as the connection
to the remote storage is repaired. The maximum disk usage for the buffer can be limited with `-remoteWrite.maxDiskUsagePerURL`.
* Uses lower amounts of RAM, CPU, disk IO and network bandwidth compared with Prometheus.
* Scrape targets can be spread among multiple `vmagent` instances when big number of targets must be scraped. See [these docs](#scraping-big-number-of-targets) for details.
## Quick Start
Please download `vmutils-*` archive from [releases page](https://github.com/VictoriaMetrics/VictoriaMetrics/releases), unpack it
and configure the following flags to the `vmagent` binary in order to start scraping Prometheus targets:
*`-promscrape.config` with the path to Prometheus config file (usually located at `/etc/prometheus/prometheus.yml`)
*`-remoteWrite.url` with the remote storage endpoint such as VictoriaMetrics, the `-remoteWrite.url` argument can be specified multiple times to replicate data concurrently to an arbitrary number of remote storage systems.
Then send Influx data to `http://vmagent-host:8429`. See [these docs](https://victoriametrics.github.io/Single-server-VictoriaMetrics.html#how-to-send-data-from-influxdb-compatible-agents-such-as-telegraf) for more details.
`vmagent` is also available in [docker images](https://hub.docker.com/r/victoriametrics/vmagent/tags).
Pass `-help` to `vmagent` in order to see [the full list of supported command-line flags with their descriptions](#advanced-usage).
## Configuration update
`vmagent` should be restarted in order to update config options set via command-line args.
`vmagent` supports multiple approaches for reloading configs from updated config files such as `-promscrape.config`, `-remoteWrite.relabelConfig` and `-remoteWrite.urlRelabelConfig`:
* Sending `SUGHUP` signal to `vmagent` process:
```bash
kill -SIGHUP `pidof vmagent`
```
* Sending HTTP request to `http://vmagent:8429/-/reload` endpoint.
There is also `-promscrape.configCheckInterval` command-line option, which can be used for automatic reloading configs from updated `-promscrape.config` file.
## Use cases
### IoT and Edge monitoring
`vmagent` can run and collect metrics in IoT and industrial networks with unreliable or scheduled connections to their remote storage.
It buffers the collected data in local files until the connection to remote storage becomes available and then sends the buffered
data to the remote storage. It re-tries sending the data to remote storage until any errors are resolved.
The maximum buffer size can be limited with `-remoteWrite.maxDiskUsagePerURL`.
`vmagent` works on various architectures from the IoT world - 32-bit arm, 64-bit arm, ppc64, 386, amd64.
See [the corresponding Makefile rules](https://github.com/VictoriaMetrics/VictoriaMetrics/blob/master/app/vmagent/Makefile) for details.
### Drop-in replacement for Prometheus
If you use Prometheus only for scraping metrics from various targets and forwarding those metrics to remote storage
then `vmagent` can replace Prometheus. Typically, `vmagent` requires lower amounts of RAM, CPU and network bandwidth compared with Prometheus.
See [these docs](#how-to-collect-metrics-in-prometheus-format) for details.
### Replication and high availability
`vmagent` replicates the collected metrics among multiple remote storage instances configured via `-remoteWrite.url` args.
If a single remote storage instance temporarily is out of service, then the collected data remains available in another remote storage instance.
`vmagent` buffers the collected data in files at `-remoteWrite.tmpDataPath` until the remote storage becomes available again and then it sends the buffered data to the remote storage in order to prevent data gaps.
### Relabeling and filtering
`vmagent` can add, remove or update labels on the collected data before sending it to the remote storage. Additionally,
it can remove unwanted samples via Prometheus-like relabeling before sending the collected data to remote storage.
Please see [these docs](#relabeling) for details.
### Splitting data streams among multiple systems
`vmagent` supports splitting the collected data between muliple destinations with the help of `-remoteWrite.urlRelabelConfig`,
which is applied independently for each configured `-remoteWrite.url` destination. For example, it is possible to replicate or split
data among long-term remote storage, short-term remote storage and a real-time analytical system [built on top of Kafka](https://github.com/Telefonica/prometheus-kafka-adapter).
Note that each destination can receive it's own subset of the collected data due to per-destination relabeling via `-remoteWrite.urlRelabelConfig`.
### Prometheus remote_write proxy
`vmagent` can be used as a proxy for Prometheus data sent via Prometheus `remote_write` protocol. It can accept data via the `remote_write` API
at the`/api/v1/write` endpoint. Then apply relabeling and filtering and proxy it to another `remote_write` system .
The `vmagent` can be configured to encrypt the incoming `remote_write` requests with `-tls*` command-line flags.
Also, Basic Auth can be enabled for the incoming `remote_write` requests with `-httpAuth.*` command-line flags.
### remote_write for clustered version
While `vmagent` can accept data in several supported protocols (OpenTSDB, Influx, Prometheus, Graphite) and scrape data from various targets, writes are always peformed in Promethes remote_write protocol. Therefore for the [clustered version](https://victoriametrics.github.io/Cluster-VictoriaMetrics.html), `-remoteWrite.url` the command-line flag should be configured as `<schema>://<vminsert-host>:8480/insert/<customer-id>/prometheus/api/v1/write`
## How to collect metrics in Prometheus format
Specify the path to `prometheus.yml` file via `-promscrape.config` command-line flag. `vmagent` takes into account the following
sections from [Prometheus config file](https://prometheus.io/docs/prometheus/latest/configuration/configuration/):
* `global`
* `scrape_configs`
All other sections are ignored, including the [remote_write](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#remote_write) section.
Use `-remoteWrite.*` command-line flag instead for configuring remote write settings.
The following scrape types in [scrape_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config) section are supported:
* `static_configs` - is for scraping statically defined targets. See [these docs](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#static_config) for details.
* `file_sd_configs` - is for scraping targets defined in external files (aka file-based service discover).
See [these docs](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#file_sd_config) for details
* `kubernetes_sd_configs` - for scraping targets in Kubernetes (k8s).
See [kubernetes_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#kubernetes_sd_config) for details.
* `ec2_sd_configs` - is for scraping targets in Amazon EC2.
See [ec2_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#ec2_sd_config) for details.
`vmagent` doesn't support the `profile` config param yet.
* `gce_sd_configs` - is for scraping targets in Google Compute Engine (GCE).
See [gce_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#gce_sd_config) for details.
`vmagent` provides the following additional functionality for `gce_sd_config`:
* if `project` arg is missing then `vmagent` uses the project for the instance where it runs;
* if `zone` arg is missing then `vmagent` uses the zone for the instance where it runs;
* if `zone` arg is equal to `"*"`, then `vmagent` discovers all the zones for the given project;
* `zone` may contain an arbitrary number of zones, i.e. `zone: [us-east1-a, us-east1-b]`.
* `consul_sd_configs` - is for scraping the targets registered in Consul.
See [consul_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#consul_sd_config) for details.
* `dns_sd_configs` - is for scraping targets discovered from DNS records (SRV, A and AAAA).
See [dns_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#dns_sd_config) for details.
* `openstack_sd_configs` - is for scraping OpenStack targets.
See [openstack_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#openstack_sd_config) for details.
[OpenStack identity API v3](https://docs.openstack.org/api-ref/identity/v3/) is supported only.
* `dockerswarm_sd_configs` - is for scraping Docker Swarm targets.
See [dockerswarm_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#dockerswarm_sd_config) for details.
* `eureka_sd_configs` - is for scraping targets registered in [Netflix Eureka](https://github.com/Netflix/eureka).
See [eureka_sd_config](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#eureka_sd_config) for details.
Please file feature requests to [our issue tracker](https://github.com/VictoriaMetrics/VictoriaMetrics/issues) if you need other service discovery mechanisms to be supported by `vmagent`.
`vmagent` also support the following additional options in `scrape_config` section:
* `disable_compression: true` - to disable response compression on a per-job basis. By default `vmagent` requests compressed responses from scrape targets
to save network bandwidth.
* `disable_keepalive: true` - to disable [HTTP keep-alive connections](https://en.wikipedia.org/wiki/HTTP_persistent_connection) on a per-job basis.
By default, `vmagent` uses keep-alive connections to scrape targets to reduce overhead on connection re-establishing.
* `stream_parse: true` - for scraping targets in a streaming manner. This may be useful for targets exporting big number of metrics.
Note that `vmagent` doesn't support `refresh_interval` option for these scrape configs. Use the corresponding `-promscrape.*CheckInterval`
command-line flag instead. For example, `-promscrape.consulSDCheckInterval=60s` sets `refresh_interval` for all the `consul_sd_configs`
entries to 60s. Run `vmagent -help` in order to see default values for the `-promscrape.*CheckInterval` flags.
The file pointed by `-promscrape.config` may contain `%{ENV_VAR}` placeholders which are substituted by the corresponding `ENV_VAR` environment variable values.
## Adding labels to metrics
Labels can be added to metrics by the following mechanisms:
* The `global -> external_labels` section in `-promscrape.config` file. These labels are added only to metrics scraped from targets configured in the `-promscrape.config` file. They aren't added to metrics collected via other [data ingestion protocols](https://victoriametrics.github.io/#how-to-import-time-series-data).
* The `-remoteWrite.label` command-line flag. These labels are added to all the collected metrics before sending them to `-remoteWrite.url`. For example, the following command will start `vmagent`, which will add `{datacenter="foobar"}` label to all the metrics pushed to all the configured remote storage systems (all the `-remoteWrite.url` flag values):
* `replace_all`: replaces all of the occurences of `regex` in the values of `source_labels` with the `replacement` and stores the results in the `target_label`.
* `labelmap_all`: replaces all of the occurences of `regex` in all the label names with the `replacement`.
* `keep_if_equal`: keeps the entry if all the label values from `source_labels` are equal.
* `drop_if_equal`: drops the entry if all the label values from `source_labels` are equal.
The relabeling can be defined in the following places:
* At the `scrape_config -> relabel_configs` section in `-promscrape.config` file. This relabeling is applied to target labels.
* At the `scrape_config -> metric_relabel_configs` section in `-promscrape.config` file. This relabeling is applied to all the scraped metrics in the given `scrape_config`.
* At the `-remoteWrite.relabelConfig` file. This relabeling is aplied to all the collected metrics before sending them to remote storage.
* At the `-remoteWrite.urlRelabelConfig` files. This relabeling is applied to metrics before sending them to the corresponding `-remoteWrite.url`.
You can read more about relabeling in the following articles:
* [How to use Relabeling in Prometheus and VictoriaMetrics](https://valyala.medium.com/how-to-use-relabeling-in-prometheus-and-victoriametrics-8b90fc22c4b2)
* [Life of a label](https://www.robustperception.io/life-of-a-label)
* [Discarding targets and timeseries with relabeling](https://www.robustperception.io/relabelling-can-discard-targets-timeseries-and-alerts)
* [Dropping labels at scrape time](https://www.robustperception.io/dropping-metrics-at-scrape-time-with-prometheus)
* [Extracting labels from legacy metric names](https://www.robustperception.io/extracting-labels-from-legacy-metric-names)
* [relabel_configs vs metric_relabel_configs](https://www.robustperception.io/relabel_configs-vs-metric_relabel_configs)
## Scraping big number of targets
A single `vmagent` instance can scrape tens of thousands of scrape targets. Sometimes this isn't enough due to limitations on CPU, network, RAM, etc.
In this case scrape targets can be split among multiple `vmagent` instances (aka `vmagent` horizontal scaling and clustering).
Each `vmagent` instance in the cluster must use identical `-promscrape.config` files with distinct `-promscrape.cluster.memberNum` values.
The flag value must be in the range `0 ... N-1`, where `N` is the number of `vmagent` instances in the cluster.
The number of `vmagent` instances in the cluster must be passed to `-promscrape.cluster.membersCount` command-line flag. For example, the following commands
spread scrape targets among a cluster of two `vmagent` instances:
By default each scrape target is scraped only by a single `vmagent` instance in the cluster. If there is a need for replicating scrape targets among multiple `vmagent` instances,
then `-promscrape.cluster.replicationFactor` command-line flag must be set to the desired number of replicas. For example, the following commands
start a cluster of three `vmagent` instances, where each target is scraped by two `vmagent` instances:
If each target is scraped by multiple `vmagent` instances, then data deduplication must be enabled at remote storage pointed by `-remoteWrite.url`.
See [these docs](https://victoriametrics.github.io/#deduplication) for details.
## Scraping targets via a proxy
`vmagent` supports scraping targets via http and https proxies. Proxy address must be specified in `proxy_url` option. For example, the following scrape config instructs
target scraping via https proxy at `https://proxy-addr:1234`:
```yml
scrape_configs:
- job_name: foo
proxy_url: https://proxy-addr:1234
```
Proxy can be configured with the following optional settings:
* `proxy_bearer_token` and `proxy_bearer_token_file` for Bearer token authorization
* `proxy_basic_auth` for Basic authorization. See [these docs](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#scrape_config).
* `proxy_tls_config` for TLS config. See [these docs](https://prometheus.io/docs/prometheus/latest/configuration/configuration/#tls_config).
For example:
```yml
scrape_configs:
- job_name: foo
proxy_url: https://proxy-addr:1234
proxy_basic_auth:
username: foobar
password: secret
proxy_tls_config:
insecure_skip_verify: true
cert_file: /path/to/cert
key_file: /path/to/key
ca_file: /path/to/ca
server_name: real-server-name
```
## Monitoring
`vmagent` exports various metrics in Prometheus exposition format at `http://vmagent-host:8429/metrics` page. We recommend setting up regular scraping of this page
either through `vmagent` itself or by Prometheus so that the exported metrics may be analyzed later.
Use official [Grafana dashboard](https://grafana.com/grafana/dashboards/12683) for `vmagent` state overview.
If you have suggestions for improvements or have found a bug - please open an issue on github or add a review to the dashboard.
`vmagent` also exports the status for various targets at the following handlers:
* `http://vmagent-host:8429/targets`. This handler returns human-readable status for every active target.
This page is easy to query from the command line with `wget`, `curl` or similar tools.
It accepts optional `show_original_labels=1` query arg which shows the original labels per each target before applying the relabeling.
This information may be useful for debugging target relabeling.
* `http://vmagent-host:8429/api/v1/targets`. This handler returns data compatible with [the corresponding page from Prometheus API](https://prometheus.io/docs/prometheus/latest/querying/api/#targets).
* `http://vmagent-host:8429/ready`. This handler returns http 200 status code when `vmagent` finishes it's initialization for all service_discovery configs.
It may be useful to perform `vmagent` rolling update without any scrape loss.
## Troubleshooting
* We recommend you [set up the official Grafana dashboard](#monitoring) in order to monitor the state of `vmagent'.
* We recommend you increase the maximum number of open files in the system (`ulimit -n`) when scraping a big number of targets,
as `vmagent` establishes at least a single TCP connection per target.
* When `vmagent` scrapes many unreliable targets, it can flood the error log with scrape errors. These errors can be suppressed
by passing `-promscrape.suppressScrapeErrors` command-line flag to `vmagent`. The most recent scrape error per each target can be observed at `http://vmagent-host:8429/targets`
and `http://vmagent-host:8429/api/v1/targets`.
* The `/api/v1/targets` page could be useful for debugging relabeling process for scrape targets.
This page contains original labels for targets dropped during relabeling (see "droppedTargets" section in the page output). By default the `-promscrape.maxDroppedTargets` targets are shown here. If your setup drops more targets during relabeling, then increase `-promscrape.maxDroppedTargets` command-line flag value to see all the dropped targets. Note that tracking each dropped target requires up to 10Kb of RAM. Therefore big values for `-promscrape.maxDroppedTargets` may result in increased memory usage if a big number of scrape targets are dropped during relabeling.
* If `vmagent` scrapes a big number of targets then the `-promscrape.dropOriginalLabels` command-line option may be passed to `vmagent` in order to reduce memory usage.
This option drops `"discoveredLabels"` and `"droppedTargets"` lists at `/api/v1/targets` page, which may result in reduced debuggability for improperly configured per-target relabeling.
* If `vmagent` scrapes targets with millions of metrics per target (for example, when scraping [federation endpoints](https://prometheus.io/docs/prometheus/latest/federation/)),
we recommend enabling `stream parsing mode` in order to reduce memory usage during scraping. This mode may be enabled either globally for all of the scrape targets
by passing `-promscrape.streamParse` command-line flag or on a per-scrape target basis with `stream_parse: true` option. For example:
```yml
scrape_configs:
- job_name: 'big-federate'
stream_parse: true
static_configs:
- targets:
- big-prometeus1
- big-prometeus2
honor_labels: true
metrics_path: /federate
params:
'match[]': ['{__name__!=""}']
```
Note that `sample_limit` option doesn't work if stream parsing is enabled because the parsed data is pushed to remote storage as soon as it is parsed. Therefore the `sample_limit` option
doesn't make sense during stream parsing.
* We recommend you increase `-remoteWrite.queues` if `vmagent_remotewrite_pending_data_bytes` metric exported at `http://vmagent-host:8429/metrics` page grows constantly.
* If you see gaps in the data pushed by `vmagent` to remote storage when `-remoteWrite.maxDiskUsagePerURL` is set, try increasing `-remoteWrite.queues`.
Such gaps may appear because `vmagent` cannot keep up with sending the collected data to remote storage. Therefore it starts dropping the buffered data
if the on-disk buffer size exceeds `-remoteWrite.maxDiskUsagePerURL`.
* `vmagent` drops data blocks if remote storage replies with `400 Bad Request` and `409 Conflict` HTTP responses. The number of dropped blocks can be monitored via `vmagent_remotewrite_packets_dropped_total` metric exported at [/metrics page](#monitoring).
* Use `-remoteWrite.queues=1` when `-remoteWrite.url` points to remote storage, which doesn't accept out-of-order samples (aka data backfilling). Such storage systems include Prometheus, Cortex and Thanos.
* `vmagent` buffers scraped data at the `-remoteWrite.tmpDataPath` directory until it is sent to `-remoteWrite.url`.
The directory can grow large when remote storage is unavailable for extended periods of time and if `-remoteWrite.maxDiskUsagePerURL` isn't set.
If you don't want to send all the data from the directory to remote storage then simply stop `vmagent` and delete the directory.
* By default `vmagent` masks `-remoteWrite.url` with `secret-url` values in logs and at `/metrics` page because
the url may contain sensitive information such as auth tokens or passwords.
Pass `-remoteWrite.showURL` command-line flag when starting `vmagent` in order to see all the valid urls.
* By default `vmagent` evenly spreads scrape load in time. If a particular scrape target must be scraped at the beginning of some interval,
then `scrape_align_interval` option must be used. For example, the following config aligns hourly scrapes to the beginning of hour:
```yml
scrape_configs:
- job_name: foo
scrape_interval: 1h
scrape_align_interval: 1h
```
* By default `vmagent` evenly spreads scrape load in time. If a particular scrape target must be scraped at specific offset, then `scrape_offset` option must be used.
For example, the following config instructs `vmagent` to scrape the target at 10 seconds of every minute:
```yml
scrape_configs:
- job_name: foo
scrape_interval: 1m
scrape_offset: 10s
```
* If you see `skipping duplicate scrape target with identical labels` errors when scraping Kubernetes pods, then it is likely these pods listen to multiple ports
or they use an init container. These errors can either be fixed or suppressed with the `-promscrape.suppressDuplicateScrapeTargetErrors` command-line flag.
See the available options below if you prefer fixing the root cause of the error:
The following relabeling rule may be added to `relabel_configs` section in order to filter out pods with unneeded ports:
2. Run `make vmagent-prod` from the root folder of [the repository](https://github.com/VictoriaMetrics/VictoriaMetrics).
It builds `vmagent-prod` binary and puts it into the `bin` folder.
### Building docker images
Run `make package-vmagent`. It builds `victoriametrics/vmagent:<PKG_TAG>` docker image locally.
`<PKG_TAG>` is an auto-generated image tag, which depends on source code in [the repository](https://github.com/VictoriaMetrics/VictoriaMetrics).
The `<PKG_TAG>` may be manually set via `PKG_TAG=foobar make package-vmagent`.
The base docker image is [alpine](https://hub.docker.com/_/alpine) but it is possible to use any other base image
by setting it via `<ROOT_IMAGE>` environment variable. For example, the following command builds the image on top of [scratch](https://hub.docker.com/_/scratch) image:
```bash
ROOT_IMAGE=scratch make package-vmagent
```
### ARM build
ARM build may run on Raspberry Pi or on [energy-efficient ARM servers](https://blog.cloudflare.com/arm-takes-wing/).
### Development ARM build
1. [Install Go](https://golang.org/doc/install). The minimum supported version is Go 1.14.
2. Run `make vmagent-arm` or `make vmagent-arm64` from the root folder of [the repository](https://github.com/VictoriaMetrics/VictoriaMetrics)
It builds `vmagent-arm` or `vmagent-arm64` binary respectively and puts it into the `bin` folder.
2. Run `make vmagent-arm-prod` or `make vmagent-arm64-prod` from the root folder of [the repository](https://github.com/VictoriaMetrics/VictoriaMetrics).
It builds `vmagent-arm-prod` or `vmagent-arm64-prod` binary respectively and puts it into the `bin` folder.
## Profiling
`vmagent` provides handlers for collecting the following [Go profiles](https://blog.golang.org/profiling-go-programs):
* Memory profile can be collected with the following command:
The command for collecting CPU profile waits for 30 seconds before returning.
The collected profiles may be analyzed with [go tool pprof](https://github.com/google/pprof).
## Advanced usage
`vmagent` can be fine-tuned with various command-line flags. Run `./vmagent -help` in order to see the full list of these flags with their desciptions and default values:
```
./vmagent -help
vmagent collects metrics data via popular data ingestion protocols and routes them to VictoriaMetrics.
See the docs at https://victoriametrics.github.io/vmagent.html .
-csvTrimTimestamp duration
Trim timestamps when importing csv data to this duration. Minimum practical duration is 1ms. Higher duration (i.e. 1s) may be used for reducing disk space usage for timestamp data (default 1ms)
-dryRun
Whether to check only config files without running vmagent. The following files are checked: -promscrape.config, -remoteWrite.relabelConfig, -remoteWrite.urlRelabelConfig . Unknown config entries are allowed in -promscrape.config by default. This can be changed with -promscrape.config.strictParse
-enableTCP6
Whether to enable IPv6 for listening and dialing. By default only IPv4 TCP and UDP is used
-envflag.enable
Whether to enable reading flags from environment variables additionally to command line. Command line flag values have priority over values from environment vars. Flags are read only from command line if this flag isn't set
-envflag.prefix string
Prefix for environment variables if -envflag.enable is set
-fs.disableMmap
Whether to use pread() instead of mmap() for reading data files. By default mmap() is used for 64-bit arches and pread() is used for 32-bit arches, since they cannot read data files bigger than 2^32 bytes in memory. mmap() is usually faster for reading small data chunks than pread()
-graphiteListenAddr string
TCP and UDP address to listen for Graphite plaintext data. Usually :2003 must be set. Doesn't work if empty
-graphiteTrimTimestamp duration
Trim timestamps for Graphite data to this duration. Minimum practical duration is 1s. Higher duration (i.e. 1m) may be used for reducing disk space usage for timestamp data (default 1s)
-http.connTimeout duration
Incoming http connections are closed after the configured timeout. This may help spreading incoming load among a cluster of services behind load balancer. Note that the real timeout may be bigger by up to 10% as a protection from Thundering herd problem (default 2m0s)
-http.disableResponseCompression
Disable compression of HTTP responses for saving CPU resources. By default compression is enabled to save network bandwidth
-http.idleConnTimeout duration
Timeout for incoming idle http connections (default 1m0s)
-http.maxGracefulShutdownDuration duration
The maximum duration for graceful shutdown of HTTP server. Highly loaded server may require increased value for graceful shutdown (default 7s)
-http.pathPrefix string
An optional prefix to add to all the paths handled by http server. For example, if '-http.pathPrefix=/foo/bar' is set, then all the http requests will be handled on '/foo/bar/*' paths. This may be useful for proxied requests. See https://www.robustperception.io/using-external-urls-and-proxies-with-prometheus
-http.shutdownDelay duration
Optional delay before http server shutdown. During this dealy the servier returns non-OK responses from /health page, so load balancers can route new requests to other servers
-httpAuth.password string
Password for HTTP Basic Auth. The authentication is disabled if -httpAuth.username is empty
-httpAuth.username string
Username for HTTP Basic Auth. The authentication is disabled if empty. See also -httpAuth.password
-httpListenAddr string
TCP address to listen for http connections. Set this flag to empty value in order to disable listening on any port. This mode may be useful for running multiple vmagent instances on the same server. Note that /targets and /metrics pages aren't available if -httpListenAddr='' (default ":8429")
-import.maxLineLen size
The maximum length in bytes of a single line accepted by /api/v1/import; the line length can be limited with 'max_rows_per_line' query arg passed to /api/v1/export
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 104857600)
-influx.databaseNames array
Comma-separated list of database names to return from /query and /influx/query API. This can be needed for accepting data from Telegraf plugins such as https://github.com/fangli/fluent-plugin-influxdb
Supports array of values separated by comma or specified via multiple flags.
-influx.maxLineSize size
The maximum size in bytes for a single Influx line during parsing
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 262144)
-influxListenAddr string
TCP and UDP address to listen for Influx line protocol data. Usually :8189 must be set. Doesn't work if empty. This flag isn't needed when ingesting data over HTTP - just send it to http://<vmagent>:8429/write
-influxMeasurementFieldSeparator string
Separator for '{measurement}{separator}{field_name}' metric name when inserted via Influx line protocol (default "_")
-influxSkipMeasurement
Uses '{field_name}' as a metric name while ignoring '{measurement}' and '-influxMeasurementFieldSeparator'
-influxSkipSingleField
Uses '{measurement}' instead of '{measurement}{separator}{field_name}' for metic name if Influx line contains only a single field
-influxTrimTimestamp duration
Trim timestamps for Influx line protocol data to this duration. Minimum practical duration is 1ms. Higher duration (i.e. 1s) may be used for reducing disk space usage for timestamp data (default 1ms)
-insert.maxQueueDuration duration
The maximum duration for waiting in the queue for insert requests due to -maxConcurrentInserts (default 1m0s)
-loggerDisableTimestamps
Whether to disable writing timestamps in logs
-loggerErrorsPerSecondLimit int
Per-second limit on the number of ERROR messages. If more than the given number of errors are emitted per second, then the remaining errors are suppressed. Zero value disables the rate limit
-loggerFormat string
Format for logs. Possible values: default, json (default "default")
-loggerLevel string
Minimum level of errors to log. Possible values: INFO, WARN, ERROR, FATAL, PANIC (default "INFO")
-loggerOutput string
Output for the logs. Supported values: stderr, stdout (default "stderr")
-loggerTimezone string
Timezone to use for timestamps in logs. Timezone must be a valid IANA Time Zone. For example: America/New_York, Europe/Berlin, Etc/GMT+3 or Local (default "UTC")
-loggerWarnsPerSecondLimit int
Per-second limit on the number of WARN messages. If more than the given number of warns are emitted per second, then the remaining warns are suppressed. Zero value disables the rate limit
-maxConcurrentInserts int
The maximum number of concurrent inserts. Default value should work for most cases, since it minimizes the overhead for concurrent inserts. This option is tigthly coupled with -insert.maxQueueDuration (default 16)
-maxInsertRequestSize size
The maximum size in bytes of a single Prometheus remote_write API request
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 33554432)
-memory.allowedBytes size
Allowed size of system memory VictoriaMetrics caches may occupy. This option overrides -memory.allowedPercent if set to non-zero value. Too low value may increase cache miss rate, which usually results in higher CPU and disk IO usage. Too high value may evict too much data from OS page cache, which will result in higher disk IO usage
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 0)
-memory.allowedPercent float
Allowed percent of system memory VictoriaMetrics caches may occupy. See also -memory.allowedBytes. Too low value may increase cache miss rate, which usually results in higher CPU and disk IO usage. Too high value may evict too much data from OS page cache, which will result in higher disk IO usage (default 60)
-metricsAuthKey string
Auth key for /metrics. It overrides httpAuth settings
-opentsdbHTTPListenAddr string
TCP address to listen for OpentTSDB HTTP put requests. Usually :4242 must be set. Doesn't work if empty
-opentsdbListenAddr string
TCP and UDP address to listen for OpentTSDB metrics. Telnet put messages and HTTP /api/put messages are simultaneously served on TCP port. Usually :4242 must be set. Doesn't work if empty
-opentsdbTrimTimestamp duration
Trim timestamps for OpenTSDB 'telnet put' data to this duration. Minimum practical duration is 1s. Higher duration (i.e. 1m) may be used for reducing disk space usage for timestamp data (default 1s)
-opentsdbhttp.maxInsertRequestSize size
The maximum size of OpenTSDB HTTP put request
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 33554432)
-opentsdbhttpTrimTimestamp duration
Trim timestamps for OpenTSDB HTTP data to this duration. Minimum practical duration is 1ms. Higher duration (i.e. 1s) may be used for reducing disk space usage for timestamp data (default 1ms)
-pprofAuthKey string
Auth key for /debug/pprof. It overrides httpAuth settings
-promscrape.cluster.memberNum int
The number of number in the cluster of scrapers. It must be an unique value in the range 0 ... promscrape.cluster.membersCount-1 across scrapers in the cluster
-promscrape.cluster.membersCount int
The number of members in a cluster of scrapers. Each member must have an unique -promscrape.cluster.memberNum in the range 0 ... promscrape.cluster.membersCount-1 . Each member then scrapes roughly 1/N of all the targets. By default cluster scraping is disabled, i.e. a single scraper scrapes all the targets
-promscrape.cluster.replicationFactor int
The number of members in the cluster, which scrape the same targets. If the replication factor is greater than 2, then the deduplication must be enabled at remote storage side. See https://victoriametrics.github.io/#deduplication (default 1)
-promscrape.config string
Optional path to Prometheus config file with 'scrape_configs' section containing targets to scrape. See https://victoriametrics.github.io/#how-to-scrape-prometheus-exporters-such-as-node-exporter for details
-promscrape.config.dryRun
Checks -promscrape.config file for errors and unsupported fields and then exits. Returns non-zero exit code on parsing errors and emits these errors to stderr. See also -promscrape.config.strictParse command-line flag. Pass -loggerLevel=ERROR if you don't need to see info messages in the output.
-promscrape.config.strictParse
Whether to allow only supported fields in -promscrape.config . By default unsupported fields are silently skipped
-promscrape.configCheckInterval duration
Interval for checking for changes in '-promscrape.config' file. By default the checking is disabled. Send SIGHUP signal in order to force config check for changes
-promscrape.consul.waitTime duration
Wait time used by Consul service discovery. Default value is used if not set
-promscrape.consulSDCheckInterval duration
Interval for checking for changes in Consul. This works only if consul_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#consul_sd_config for details (default 30s)
-promscrape.disableCompression
Whether to disable sending 'Accept-Encoding: gzip' request headers to all the scrape targets. This may reduce CPU usage on scrape targets at the cost of higher network bandwidth utilization. It is possible to set 'disable_compression: true' individually per each 'scrape_config' section in '-promscrape.config' for fine grained control
-promscrape.disableKeepAlive
Whether to disable HTTP keep-alive connections when scraping all the targets. This may be useful when targets has no support for HTTP keep-alive connection. It is possible to set 'disable_keepalive: true' individually per each 'scrape_config' section in '-promscrape.config' for fine grained control. Note that disabling HTTP keep-alive may increase load on both vmagent and scrape targets
-promscrape.discovery.concurrency int
The maximum number of concurrent requests to Prometheus autodiscovery API (Consul, Kubernetes, etc.) (default 100)
-promscrape.discovery.concurrentWaitTime duration
The maximum duration for waiting to perform API requests if more than -promscrape.discovery.concurrency requests are simultaneously performed (default 1m0s)
-promscrape.dnsSDCheckInterval duration
Interval for checking for changes in dns. This works only if dns_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#dns_sd_config for details (default 30s)
-promscrape.dockerswarmSDCheckInterval duration
Interval for checking for changes in dockerswarm. This works only if dockerswarm_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#dockerswarm_sd_config for details (default 30s)
-promscrape.dropOriginalLabels
Whether to drop original labels for scrape targets at /targets and /api/v1/targets pages. This may be needed for reducing memory usage when original labels for big number of scrape targets occupy big amounts of memory. Note that this reduces debuggability for improper per-target relabeling configs
-promscrape.ec2SDCheckInterval duration
Interval for checking for changes in ec2. This works only if ec2_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#ec2_sd_config for details (default 1m0s)
-promscrape.eurekaSDCheckInterval duration
Interval for checking for changes in eureka. This works only if eureka_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#eureka_sd_config for details (default 30s)
-promscrape.fileSDCheckInterval duration
Interval for checking for changes in 'file_sd_config'. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#file_sd_config for details (default 30s)
-promscrape.gceSDCheckInterval duration
Interval for checking for changes in gce. This works only if gce_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#gce_sd_config for details (default 1m0s)
-promscrape.kubernetes.apiServerTimeout duration
How frequently to reload the full state from Kuberntes API server (default 30m0s)
-promscrape.kubernetesSDCheckInterval duration
Interval for checking for changes in Kubernetes API server. This works only if kubernetes_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#kubernetes_sd_config for details (default 30s)
-promscrape.maxDroppedTargets int
The maximum number of droppedTargets to show at /api/v1/targets page. Increase this value if your setup drops more scrape targets during relabeling and you need investigating labels for all the dropped targets. Note that the increased number of tracked dropped targets may result in increased memory usage (default 1000)
-promscrape.maxScrapeSize size
The maximum size of scrape response in bytes to process from Prometheus targets. Bigger responses are rejected
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 16777216)
-promscrape.openstackSDCheckInterval duration
Interval for checking for changes in openstack API server. This works only if openstack_sd_configs is configured in '-promscrape.config' file. See https://prometheus.io/docs/prometheus/latest/configuration/configuration/#openstack_sd_config for details (default 30s)
-promscrape.streamParse
Whether to enable stream parsing for metrics obtained from scrape targets. This may be useful for reducing memory usage when millions of metrics are exposed per each scrape target. It is posible to set 'stream_parse: true' individually per each 'scrape_config' section in '-promscrape.config' for fine grained control
-promscrape.suppressDuplicateScrapeTargetErrors
Whether to suppress 'duplicate scrape target' errors; see https://victoriametrics.github.io/vmagent.html#troubleshooting for details
-promscrape.suppressScrapeErrors
Whether to suppress scrape errors logging. The last error for each target is always available at '/targets' page even if scrape errors logging is suppressed
-remoteWrite.basicAuth.password array
Optional basic auth password to use for -remoteWrite.url. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.basicAuth.username array
Optional basic auth username to use for -remoteWrite.url. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.bearerToken array
Optional bearer auth token to use for -remoteWrite.url. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.flushInterval duration
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 (default 1s)
-remoteWrite.label array
Optional label in the form 'name=value' to add to all the metrics before sending them to -remoteWrite.url. Pass multiple -remoteWrite.label flags in order to add multiple labels to metrics before sending them to remote storage
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.maxBlockSize size
The maximum size in bytes of unpacked request to send to remote storage. It shouldn't exceed -maxInsertRequestSize from VictoriaMetrics
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 8388608)
-remoteWrite.maxDiskUsagePerURL size
The maximum file-based buffer size in bytes at -remoteWrite.tmpDataPath for each -remoteWrite.url. When buffer size reaches the configured maximum, then old data is dropped when adding new data to the buffer. Buffered data is stored in ~500MB chunks, so the minimum practical value for this flag is 500000000. Disk usage is unlimited if the value is set to 0
Supports the following optional suffixes for size values: KB, MB, GB, KiB, MiB, GiB (default 0)
-remoteWrite.proxyURL array
Optional proxy URL for writing data to -remoteWrite.url. Supported proxies: http, https, socks5. Example: -remoteWrite.proxyURL=socks5://proxy:1234
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.queues int
The number of concurrent queues to each -remoteWrite.url. Set more queues if default number of queues isn't enough for sending high volume of collected data to remote storage (default 4)
-remoteWrite.rateLimit array
Optional rate limit in bytes per second for data sent to -remoteWrite.url. By default the rate limit is disabled. It can be useful for limiting load on remote storage when big amounts of buffered data is sent after temporary unavailability of the remote storage
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.relabelConfig string
Optional path to file with relabel_config entries. These entries are applied to all the metrics before sending them to -remoteWrite.url. See https://victoriametrics.github.io/vmagent.html#relabeling for details
-remoteWrite.roundDigits array
Round metric values to this number of decimal digits after the point before writing them to remote storage. Examples: -remoteWrite.roundDigits=2 would round 1.236 to 1.24, while -remoteWrite.roundDigits=-1 would round 126.78 to 130. By default digits rounding is disabled. Set it to 100 for disabling it for a particular remote storage. This option may be used for improving data compression for the stored metrics
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.sendTimeout array
Timeout for sending a single block of data to -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.showURL
Whether to show -remoteWrite.url in the exported metrics. It is hidden by default, since it can contain sensitive info such as auth key
-remoteWrite.significantFigures array
The number of significant figures to leave in metric values before writing them to remote storage. See https://en.wikipedia.org/wiki/Significant_figures . Zero value saves all the significant figures. This option may be used for improving data compression for the stored metrics. See also -remoteWrite.roundDigits
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.tlsCAFile array
Optional path to TLS CA file to use for verifying connections to -remoteWrite.url. By default system CA is used. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.tlsCertFile array
Optional path to client-side TLS certificate file to use when connecting to -remoteWrite.url. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.tlsInsecureSkipVerify array
Whether to skip tls verification when connecting to -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.tlsKeyFile array
Optional path to client-side TLS certificate key to use when connecting to -remoteWrite.url. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.tlsServerName array
Optional TLS server name to use for connections to -remoteWrite.url. By default the server name from -remoteWrite.url is used. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.tmpDataPath string
Path to directory where temporary data for remote write component is stored. See also -remoteWrite.maxDiskUsagePerURL (default "vmagent-remotewrite-data")
-remoteWrite.url array
Remote storage URL to write data to. It must support Prometheus remote_write API. It is recommended using VictoriaMetrics as remote storage. Example url: http://<victoriametrics-host>:8428/api/v1/write . Pass multiple -remoteWrite.url flags in order to write data concurrently to multiple remote storage systems
Supports array of values separated by comma or specified via multiple flags.
-remoteWrite.urlRelabelConfig array
Optional path to relabel config for the corresponding -remoteWrite.url
Supports array of values separated by comma or specified via multiple flags.
-tls
Whether to enable TLS (aka HTTPS) for incoming requests. -tlsCertFile and -tlsKeyFile must be set if -tls is set
-tlsCertFile string
Path to file with TLS certificate. Used only if -tls is set. Prefer ECDSA certs instead of RSA certs, since RSA certs are slow
-tlsKeyFile string
Path to file with TLS key. Used only if -tls is set
measurementFieldSeparator=flag.String("influxMeasurementFieldSeparator","_","Separator for '{measurement}{separator}{field_name}' metric name when inserted via InfluxDB line protocol")
skipSingleField=flag.Bool("influxSkipSingleField",false,"Uses '{measurement}' instead of '{measurement}{separator}{field_name}' for metric name if InfluxDB line contains only a single field")
measurementFieldSeparator=flag.String("influxMeasurementFieldSeparator","_","Separator for '{measurement}{separator}{field_name}' metric name when inserted via Influx line protocol")
skipSingleField=flag.Bool("influxSkipSingleField",false,"Uses '{measurement}' instead of '{measurement}{separator}{field_name}' for metic name if Influx line contains only a single field")
skipMeasurement=flag.Bool("influxSkipMeasurement",false,"Uses '{field_name}' as a metric name while ignoring '{measurement}' and '-influxMeasurementFieldSeparator'")
dbLabel=flag.String("influxDBLabel","db","Default label for the DB name sent over '?db={db_name}' query parameter")
httpListenAddrs=flagutil.NewArrayString("httpListenAddr","TCP address to listen for incoming http requests. "+
httpListenAddr=flag.String("httpListenAddr",":8429","TCP address to listen for http connections. "+
"Set this flag to empty value in order to disable listening on any port. This mode may be useful for running multiple vmagent instances on the same server. "+
"Note that /targets and /metrics pages aren't available if -httpListenAddr=''. See also -tls and -httpListenAddr.useProxyProtocol")
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")
influxListenAddr=flag.String("influxListenAddr","","TCP and UDP address to listen for InfluxDB line protocol data. Usually :8089 must be set. Doesn't work if empty. "+
"This flag isn't needed when ingesting data over HTTP - just send it to http://<vmagent>:8429/write . "+
"See also -influxListenAddr.useProxyProtocol")
influxUseProxyProtocol=flag.Bool("influxListenAddr.useProxyProtocol",false,"Whether to use proxy protocol for connections accepted at -influxListenAddr . "+
"See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt")
graphiteListenAddr=flag.String("graphiteListenAddr","","TCP and UDP address to listen for Graphite plaintext data. Usually :2003 must be set. Doesn't work if empty. "+
"See also -graphiteListenAddr.useProxyProtocol")
graphiteUseProxyProtocol=flag.Bool("graphiteListenAddr.useProxyProtocol",false,"Whether to use proxy protocol for connections accepted at -graphiteListenAddr . "+
"See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt")
opentsdbListenAddr=flag.String("opentsdbListenAddr","","TCP and UDP address to listen for OpenTSDB metrics. "+
"Note that /targets and /metrics pages aren't available if -httpListenAddr=''")
influxListenAddr=flag.String("influxListenAddr","","TCP and UDP address to listen for Influx line protocol data. Usually :8189 must be set. Doesn't work if empty. "+
"This flag isn't needed when ingesting data over HTTP - just send it to http://<vmagent>:8429/write")
graphiteListenAddr=flag.String("graphiteListenAddr","","TCP and UDP address to listen for Graphite plaintext data. Usually :2003 must be set. Doesn't work if empty")
opentsdbListenAddr=flag.String("opentsdbListenAddr","","TCP and UDP address to listen for OpentTSDB metrics. "+
"Telnet put messages and HTTP /api/put messages are simultaneously served on TCP port. "+
"Usually :4242 must be set. Doesn't work if empty. See also -opentsdbListenAddr.useProxyProtocol")
opentsdbUseProxyProtocol=flag.Bool("opentsdbListenAddr.useProxyProtocol",false,"Whether to use proxy protocol for connections accepted at -opentsdbListenAddr . "+
"See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt")
opentsdbHTTPListenAddr=flag.String("opentsdbHTTPListenAddr","","TCP address to listen for OpenTSDB HTTP put requests. Usually :4242 must be set. Doesn't work if empty. "+
"See also -opentsdbHTTPListenAddr.useProxyProtocol")
opentsdbHTTPUseProxyProtocol=flag.Bool("opentsdbHTTPListenAddr.useProxyProtocol",false,"Whether to use proxy protocol for connections accepted "+
"at -opentsdbHTTPListenAddr . See https://www.haproxy.org/download/1.8/doc/proxy-protocol.txt")
configAuthKey=flagutil.NewPassword("configAuthKey","Authorization key for accessing /config page. It must be passed via authKey query arg. It overrides -httpAuth.*")
reloadAuthKey=flagutil.NewPassword("reloadAuthKey","Auth key for /-/reload http endpoint. It must be passed via authKey query arg. It overrides -httpAuth.*")
dryRun=flag.Bool("dryRun",false,"Whether to check config files without running vmagent. The following files are checked: "+
"Unknown config entries aren't allowed in -promscrape.config by default. This can be changed by passing -promscrape.config.strictParse=false command-line flag")
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")
"Usually :4242 must be set. Doesn't work if empty")
opentsdbHTTPListenAddr=flag.String("opentsdbHTTPListenAddr","","TCP address to listen for OpentTSDB HTTP put requests. Usually :4242 must be set. Doesn't work if empty")
dryRun=flag.Bool("dryRun",false,"Whether to check only config files without running vmagent. The following files are checked: "+
fmt.Fprintf(w,"See docs at <a href='https://docs.victoriametrics.com/victoriametrics/vmagent/'>https://docs.victoriametrics.com/victoriametrics/vmagent/</a></br>")
fmt.Fprintf(w,"Useful endpoints:</br>")
httpserver.WriteAPIHelp(w,[][2]string{
{"targets","status for discovered active targets"},
{"service-discovery","labels before and after relabeling for discovered targets"},
rateLimit=flagutil.NewArrayInt("remoteWrite.rateLimit",0,"Optional rate limit in bytes per second for data sent to the corresponding -remoteWrite.url. "+
"By default, the rate limit is disabled. It can be useful for limiting load on remote storage when big amounts of buffered data "+
"is sent after temporary unavailability of the remote storage. See also -maxIngestionRate")
sendTimeout=flagutil.NewArrayDuration("remoteWrite.sendTimeout",time.Minute,"Timeout for sending a single block of data to the corresponding -remoteWrite.url")
retryMinInterval=flagutil.NewArrayDuration("remoteWrite.retryMinInterval",time.Second,"The minimum delay between retry attempts to send a block of data to the corresponding -remoteWrite.url. Every next retry attempt will double the delay to prevent hammering of remote database. See also -remoteWrite.retryMaxInterval")
// deprecated in the future. use -remoteWrite.retryMaxInterval instead
retryMaxTime=flagutil.NewArrayDuration("remoteWrite.retryMaxTime",time.Minute,"The max time spent on retry attempts to send a block of data to the corresponding -remoteWrite.url. This flag is deprecated, use -remoteWrite.retryMaxInterval instead")
retryMaxInterval=flagutil.NewArrayDuration("remoteWrite.retryMaxInterval",time.Minute,"The maximum delay between retry attempts to send a block of data to the corresponding -remoteWrite.url. The delay doubles with each retry until this maximum is reached, after which it remains constant. See also -remoteWrite.retryMinInterval")
proxyURL=flagutil.NewArrayString("remoteWrite.proxyURL","Optional proxy URL for writing data to the corresponding -remoteWrite.url. "+
tlsInsecureSkipVerify=flagutil.NewArrayBool("remoteWrite.tlsInsecureSkipVerify","Whether to skip tls verification when connecting to -remoteWrite.url")
tlsCertFile=flagutil.NewArray("remoteWrite.tlsCertFile","Optional path to client-side TLS certificate file to use when connecting to -remoteWrite.url. "+
"If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
tlsKeyFile=flagutil.NewArray("remoteWrite.tlsKeyFile","Optional path to client-side TLS certificate key to use when connecting to -remoteWrite.url. "+
"If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
tlsCAFile=flagutil.NewArray("remoteWrite.tlsCAFile","Optional path to TLS CA file to use for verifying connections to -remoteWrite.url. "+
"By default system CA is used. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
tlsServerName=flagutil.NewArray("remoteWrite.tlsServerName","Optional TLS server name to use for connections to -remoteWrite.url. "+
"By default the server name from -remoteWrite.url is used. If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
tlsHandshakeTimeout=flagutil.NewArrayDuration("remoteWrite.tlsHandshakeTimeout",20*time.Second,"The timeout for establishing tls connections to the corresponding -remoteWrite.url")
tlsInsecureSkipVerify=flagutil.NewArrayBool("remoteWrite.tlsInsecureSkipVerify","Whether to skip tls verification when connecting to the corresponding -remoteWrite.url")
tlsCertFile=flagutil.NewArrayString("remoteWrite.tlsCertFile","Optional path to client-side TLS certificate file to use when connecting "+
"to the corresponding -remoteWrite.url")
tlsKeyFile=flagutil.NewArrayString("remoteWrite.tlsKeyFile","Optional path to client-side TLS certificate key to use when connecting to the corresponding -remoteWrite.url")
tlsCAFile=flagutil.NewArrayString("remoteWrite.tlsCAFile","Optional path to TLS CA file to use for verifying connections to the corresponding -remoteWrite.url. "+
"By default, system CA is used")
tlsServerName=flagutil.NewArrayString("remoteWrite.tlsServerName","Optional TLS server name to use for connections to the corresponding -remoteWrite.url. "+
"By default, the server name from -remoteWrite.url is used")
headers=flagutil.NewArrayString("remoteWrite.headers","Optional HTTP headers to send with each request to the corresponding -remoteWrite.url. "+
"For example, -remoteWrite.headers='My-Auth:foobar' would send 'My-Auth: foobar' HTTP header with every request to the corresponding -remoteWrite.url. "+
"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")
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")
bearerToken=flagutil.NewArrayString("remoteWrite.bearerToken","Optional bearer auth token to use for the corresponding -remoteWrite.url")
bearerTokenFile=flagutil.NewArrayString("remoteWrite.bearerTokenFile","Optional path to bearer token file to use for the corresponding -remoteWrite.url. "+
"The token is re-read from the file every second")
oauth2ClientID=flagutil.NewArrayString("remoteWrite.oauth2.clientID","Optional OAuth2 clientID to use for the corresponding -remoteWrite.url")
oauth2ClientSecret=flagutil.NewArrayString("remoteWrite.oauth2.clientSecret","Optional OAuth2 clientSecret to use for the corresponding -remoteWrite.url")
oauth2ClientSecretFile=flagutil.NewArrayString("remoteWrite.oauth2.clientSecretFile","Optional OAuth2 clientSecretFile to use for the corresponding -remoteWrite.url")
oauth2EndpointParams=flagutil.NewArrayString("remoteWrite.oauth2.endpointParams","Optional OAuth2 endpoint parameters to use for the corresponding -remoteWrite.url . "+
`The endpoint parameters must be set in JSON format: {"param1":"value1",...,"paramN":"valueN"}`)
oauth2TokenURL=flagutil.NewArrayString("remoteWrite.oauth2.tokenUrl","Optional OAuth2 tokenURL to use for the corresponding -remoteWrite.url")
oauth2Scopes=flagutil.NewArrayString("remoteWrite.oauth2.scopes","Optional OAuth2 scopes to use for the corresponding -remoteWrite.url. Scopes must be delimited by ';'")
awsUseSigv4=flagutil.NewArrayBool("remoteWrite.aws.useSigv4","Enables SigV4 request signing for the corresponding -remoteWrite.url. "+
"It is expected that other -remoteWrite.aws.* command-line flags are set if sigv4 request signing is enabled")
awsEC2Endpoint=flagutil.NewArrayString("remoteWrite.aws.ec2Endpoint","Optional AWS EC2 API endpoint to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set")
awsSTSEndpoint=flagutil.NewArrayString("remoteWrite.aws.stsEndpoint","Optional AWS STS API endpoint to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set")
awsRegion=flagutil.NewArrayString("remoteWrite.aws.region","Optional AWS region to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set")
awsRoleARN=flagutil.NewArrayString("remoteWrite.aws.roleARN","Optional AWS roleARN to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set")
awsAccessKey=flagutil.NewArrayString("remoteWrite.aws.accessKey","Optional AWS AccessKey to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set")
awsService=flagutil.NewArrayString("remoteWrite.aws.service","Optional AWS Service to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set. "+
"Defaults to \"aps\"")
awsSecretKey=flagutil.NewArrayString("remoteWrite.aws.secretKey","Optional AWS SecretKey to use for the corresponding -remoteWrite.url if -remoteWrite.aws.useSigv4 is set")
basicAuthUsername=flagutil.NewArray("remoteWrite.basicAuth.username","Optional basic auth username to use for -remoteWrite.url. "+
"If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
basicAuthPassword=flagutil.NewArray("remoteWrite.basicAuth.password","Optional basic auth password to use for -remoteWrite.url. "+
"If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
bearerToken=flagutil.NewArray("remoteWrite.bearerToken","Optional bearer auth token to use for -remoteWrite.url. "+
"If multiple args are set, then they are applied independently for the corresponding -remoteWrite.url")
)
typeclientstruct{
sanitizedURLstring
remoteWriteURLstring
authHeaderstring
fq*persistentqueue.FastQueue
hc*http.Client
// Whether to use VictoriaMetrics remote write protocol for sending the data to remoteWriteURL
// Just drop block on 409 status code like Prometheus does.
// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/873
// and https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1149
_=resp.Body.Close()
c.packetsDropped.Inc()
returntrue
// - 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
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)
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
}
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.
// See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/873
// and https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1149
logBlockRejected(block,c.sanitizedURL,resp)
// drop block on 400 status code,
// not expected that remote server will be able to handle it on retry
// should fix https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1149
flushInterval=flag.Duration("remoteWrite.flushInterval",time.Second,"Interval for flushing the data to remote storage. "+
"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")
"This option takes effect only when less than 10K data points per second are pushed to -remoteWrite.url")
maxUnpackedBlockSize=flagutil.NewBytes("remoteWrite.maxBlockSize",8*1024*1024,"The maximum size in bytes of unpacked request to send to remote storage. "+
"It shouldn't exceed -maxInsertRequestSize from VictoriaMetrics")
)
// the maximum number of rows to send per each block.
// A single time series left. Recursively split its samples and metadata into smaller parts if possible.
samples:=wr.Timeseries[0].Samples
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)
unparsedLabelsGlobal=flagutil.NewArrayString("remoteWrite.label","Optional label in the form 'name=value' to add to all the metrics before sending them to -remoteWrite.url. "+
unparsedLabelsGlobal=flagutil.NewArray("remoteWrite.label","Optional label in the form 'name=value' to add to all the metrics before sending them to -remoteWrite.url. "+
"Pass multiple -remoteWrite.label flags in order to add multiple labels to metrics before sending them to remote storage")
relabelConfigPathGlobal=flag.String("remoteWrite.relabelConfig","","Optional path to file with relabeling configs, which are applied "+
"to all the metrics before sending them to -remoteWrite.url. See also -remoteWrite.urlRelabelConfig. "+
"The path can point either to local file or to http url. "+
"See https://docs.victoriametrics.com/victoriametrics/relabeling/")
relabelConfigPaths=flagutil.NewArrayString("remoteWrite.urlRelabelConfig","Optional path to relabel configs for the corresponding -remoteWrite.url. "+
"See also -remoteWrite.relabelConfig. The path can point either to local file or to http url. "+
"See https://docs.victoriametrics.com/victoriametrics/relabeling/")
usePromCompatibleNaming=flag.Bool("usePromCompatibleNaming",false,"Whether to replace characters unsupported by Prometheus with underscores "+
"in the ingested metric names and label names. For example, foo.bar{a.b='c'} is transformed into foo_bar{a_b='c'} during data ingestion if this flag is set. "+
"See https://prometheus.io/docs/concepts/data_model/#metric-names-and-labels")
relabelConfigPathGlobal=flag.String("remoteWrite.relabelConfig","","Optional path to file with relabel_config entries. These entries are applied to all the metrics "+
"before sending them to -remoteWrite.url. See https://victoriametrics.github.io/vmagent.html#relabeling for details")
relabelConfigPaths=flagutil.NewArray("remoteWrite.urlRelabelConfig","Optional path to relabel config for the corresponding -remoteWrite.url")
// extraLabels must be added before applying relabeling according to https://prometheus.io/docs/prometheus/latest/configuration/configuration/#remote_write
t.Fatalf("average active time series before: %d, after: %d, changed: %.2f. threshold: %.2f",avgActiveTimeSeries1,avgActiveTimeSeries2,changed,threshold)
streamAggrGlobalConfig=flag.String("streamAggr.config","","Optional path to file with stream aggregation config. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/ . "+
"See also -streamAggr.keepInput, -streamAggr.dropInput and -streamAggr.dedupInterval")
streamAggrGlobalKeepInput=flag.Bool("streamAggr.keepInput",false,"Whether to keep all the input samples after the aggregation "+
"with -streamAggr.config. By default, only aggregates samples are dropped, while the remaining samples "+
"are written to remote storages write. See also -streamAggr.dropInput and https://docs.victoriametrics.com/victoriametrics/stream-aggregation/")
streamAggrGlobalDropInput=flag.Bool("streamAggr.dropInput",false,"Whether to drop all the input samples after the aggregation "+
"with -remoteWrite.streamAggr.config. By default, only aggregates samples are dropped, while the remaining samples "+
"are written to remote storages write. See also -streamAggr.keepInput and https://docs.victoriametrics.com/victoriametrics/stream-aggregation/")
streamAggrGlobalDedupInterval=flag.Duration("streamAggr.dedupInterval",0,"Input samples are de-duplicated with this interval on "+
"aggregator before optional aggregation with -streamAggr.config . "+
"See also -dedup.minScrapeInterval and https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#deduplication")
streamAggrGlobalIgnoreOldSamples=flag.Bool("streamAggr.ignoreOldSamples",false,"Whether to ignore input samples with old timestamps outside the "+
"current aggregation interval for aggregator. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#ignoring-old-samples")
streamAggrGlobalIgnoreFirstIntervals=flag.Int("streamAggr.ignoreFirstIntervals",0,"Number of aggregation intervals to skip after the start for "+
"aggregator. Increase this value if you observe incorrect aggregation results after vmagent restarts. It could be caused by receiving unordered delayed data from "+
"clients pushing data into the vmagent. See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#ignore-aggregation-intervals-on-start")
streamAggrGlobalDropInputLabels=flagutil.NewArrayString("streamAggr.dropInputLabels","An optional list of labels to drop from samples for aggregator "+
"before stream de-duplication and aggregation . See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#dropping-unneeded-labels")
streamAggrGlobalEnableWindows=flag.Bool("streamAggr.enableWindows",false,"Enables aggregation within fixed windows for all global aggregators. "+
"This allows to get more precise results, but impacts resource usage as it requires twice more memory to store two states. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#aggregation-windows.")
// Per URL config
streamAggrConfig=flagutil.NewArrayString("remoteWrite.streamAggr.config","Optional path to file with stream aggregation config for the corresponding -remoteWrite.url. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/ . "+
"See also -remoteWrite.streamAggr.keepInput, -remoteWrite.streamAggr.dropInput and -remoteWrite.streamAggr.dedupInterval")
streamAggrDropInput=flagutil.NewArrayBool("remoteWrite.streamAggr.dropInput","Whether to drop all the input samples after the aggregation "+
"with -remoteWrite.streamAggr.config at the corresponding -remoteWrite.url. By default, only aggregates samples are dropped, while the remaining samples "+
"are written to the corresponding -remoteWrite.url . See also -remoteWrite.streamAggr.keepInput and https://docs.victoriametrics.com/victoriametrics/stream-aggregation/")
streamAggrKeepInput=flagutil.NewArrayBool("remoteWrite.streamAggr.keepInput","Whether to keep all the input samples after the aggregation "+
"with -remoteWrite.streamAggr.config at the corresponding -remoteWrite.url. By default, only aggregates samples are dropped, while the remaining samples "+
"are written to the corresponding -remoteWrite.url . See also -remoteWrite.streamAggr.dropInput and https://docs.victoriametrics.com/victoriametrics/stream-aggregation/")
streamAggrDedupInterval=flagutil.NewArrayDuration("remoteWrite.streamAggr.dedupInterval",0,"Input samples are de-duplicated with this interval before optional aggregation "+
"with -remoteWrite.streamAggr.config at the corresponding -remoteWrite.url. See also -dedup.minScrapeInterval and https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#deduplication")
streamAggrIgnoreOldSamples=flagutil.NewArrayBool("remoteWrite.streamAggr.ignoreOldSamples","Whether to ignore input samples with old timestamps outside the current "+
"aggregation interval for the corresponding -remoteWrite.streamAggr.config at the corresponding -remoteWrite.url. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#ignoring-old-samples")
streamAggrIgnoreFirstIntervals=flagutil.NewArrayInt("remoteWrite.streamAggr.ignoreFirstIntervals",0,"Number of aggregation intervals to skip after the start "+
"for the corresponding -remoteWrite.streamAggr.config at the corresponding -remoteWrite.url. Increase this value if "+
"you observe incorrect aggregation results after vmagent restarts. It could be caused by receiving buffered delayed data from clients pushing data into the vmagent. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#ignore-aggregation-intervals-on-start")
streamAggrDropInputLabels=flagutil.NewArrayString("remoteWrite.streamAggr.dropInputLabels","An optional list of labels to drop from samples "+
"before stream de-duplication and aggregation with -remoteWrite.streamAggr.config and -remoteWrite.streamAggr.dedupInterval at the corresponding -remoteWrite.url. "+
"Multiple labels per remoteWrite.url must be delimited by '^^': -remoteWrite.streamAggr.dropInputLabels='replica^^az,replica'. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#dropping-unneeded-labels")
streamAggrEnableWindows=flagutil.NewArrayBool("remoteWrite.streamAggr.enableWindows","Enables aggregation within fixed windows for all remote write's aggregators. "+
"This allows to get more precise results, but impacts resource usage as it requires twice more memory to store two states. "+
"See https://docs.victoriametrics.com/victoriametrics/stream-aggregation/#aggregation-windows.")
)
// CheckStreamAggrConfigs checks -remoteWrite.streamAggr.config and -streamAggr.config.
funcCheckStreamAggrConfigs()error{
// Check global config
sas,err:=newStreamAggrConfigGlobal()
iferr!=nil{
returnerr
}
sas.MustStop()
iflen(*streamAggrConfig)>len(*remoteWriteURLs){
returnfmt.Errorf("too many -remoteWrite.streamAggr.config args: %d; it mustn't exceed the number of -remoteWrite.url args: %d",len(*streamAggrConfig),len(*remoteWriteURLs))
}
pushNoop:=func(_[]prompb.TimeSeries){}
foridx:=range*streamAggrConfig{
sas,err:=newStreamAggrConfigPerURL(idx,pushNoop)
iferr!=nil{
returnerr
}
sas.MustStop()
}
returnnil
}
funcreloadStreamAggrConfigs(){
reloadStreamAggrConfigGlobal()
for_,rwctx:=rangerwctxsGlobal{
rwctx.reloadStreamAggrConfig()
}
}
funcreloadStreamAggrConfigGlobal(){
path:=*streamAggrGlobalConfig
ifpath==""{
return
}
logger.Infof("reloading stream aggregation configs pointed by -streamAggr.config=%q",path)
File diff suppressed because one or more lines are too long
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.
