Sending unfinished aggregate states tend to produce unexpected anomalies with lower values than expected.
The old behavior can be restored by specifying `flush_on_shutdown: true` setting in streaming aggregation config
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* app/vmalert: fix data race during hot-config reload
During hot-reload, the logic evokes the group update and rules evaluation
interruption simultaneously. Falsely assuming that interruption happens before
the update. However, it could happen that group will be updated first and only
after the rules evaluation will be cancelled. Which will result in permanent
interruption for all rules within the group.
The fix caches the cancel context function into local variable first. And only after
performs the group update. With cached cancel function we can safely call it without
worrying that we cancel the evaluation for already updated group.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* Revert "app/vmalert: fix data race during hot-config reload"
This reverts commit a4bb7e8932.
* app/vmalert: fix data race during hot-config reload
During hot-reload, the logic evokes the group update and rules evaluation
interruption simultaneously. Falsely assuming that interruption happens before
the update. However, it could happen that group will be updated first and only
after the rules evaluation will be cancelled. Which will result in permanent
interruption for all rules within the group.
The fix cancels the evaulation context before applying the update, making sure
that the context will be cancelled for old group always.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* wip
Signed-off-by: hagen1778 <roman@victoriametrics.com>
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
- Maintain a separate worker pool per each part type (in-memory, file, big and small).
Previously a shared pool was used for merging all the part types.
A single merge worker could merge parts with mixed types at once. For example,
it could merge simultaneously an in-memory part plus a big file part.
Such a merge could take hours for big file part. During the duration of this merge
the in-memory part was pinned in memory and couldn't be persisted to disk
under the configured -inmemoryDataFlushInterval .
Another common issue, which could happen when parts with mixed types are merged,
is uncontrolled growth of in-memory parts or small parts when all the merge workers
were busy with merging big files. Such growth could lead to significant performance
degradataion for queries, since every query needs to check ever growing list of parts.
This could also slow down the registration of new time series, since VictoriaMetrics
searches for the internal series_id in the indexdb for every new time series.
The third issue is graceful shutdown duration, which could be very long when a background
merge is running on in-memory parts plus big file parts. This merge couldn't be interrupted,
since it merges in-memory parts.
A separate pool of merge workers per every part type elegantly resolves both issues:
- In-memory parts are merged to file-based parts in a timely manner, since the maximum
size of in-memory parts is limited.
- Long-running merges for big parts do not block merges for in-memory parts and small parts.
- Graceful shutdown duration is now limited by the time needed for flushing in-memory parts to files.
Merging for file parts is instantly canceled on graceful shutdown now.
- Deprecate -smallMergeConcurrency command-line flag, since the new background merge algorithm
should automatically self-tune according to the number of available CPU cores.
- Deprecate -finalMergeDelay command-line flag, since it wasn't working correctly.
It is better to run forced merge when needed - https://docs.victoriametrics.com/#forced-merge
- Tune the number of shards for pending rows and items before the data goes to in-memory parts
and becomes visible for search. This improves the maximum data ingestion rate and the maximum rate
for registration of new time series. This should reduce the duration of data ingestion slowdown
in VictoriaMetrics cluster on e.g. re-routing events, when some of vmstorage nodes become temporarily
unavailable.
- Prevent from possible "sync: WaitGroup misuse" panic on graceful shutdown.
This is a follow-up for fa566c68a6 .
Thanks @misutoth to for the inspiration at https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5212
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5190
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3790
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3551
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3337
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3425
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3647
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3641
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/648
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/291
* app/vmalert: autogenerate `ALERTS_FOR_STATE` time series for alerting rules with `for: 0`
Previously, `ALERTS_FOR_STATE` was generated only for alerts with `for > 0`.
This behavior differs from Prometheus behavior - it generates ALERTS_FOR_STATE
time series for alerting rules with `for: 0` as well. Such time series can
be useful for tracking the moment when alerting rule became active.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5648https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3056
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* app/vmalert: support ALERTS_FOR_STATE in `replay` mode
Signed-off-by: hagen1778 <roman@victoriametrics.com>
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
This reverts cd4f641d32 , since it has been appeared that the disabled compression
for vmstorage->vmselect data increase network bandwidth usage by more than 10x on typical production workloads,
while it decreases CPU usage at vmstorage by up to 10% and improves query latency by up to 10%.
The 10x increase in network usage is too high price for 10% improvements on query latency and vmstorage CPU usage.
This may result in network bandwidth bottlenecks, which can reduce the overall performance and stability
of VictoriaMetrics cluster. That's why return back the vmstorage->vmselect data compression by default.
The vmstorage->vmselect compression can be disabled by passing -rpc.disableCompression command-line flag to vmstorage.
The vmselect->vmselect compression in multi-level cluster setup can be disabled by passing -clusternative.disableCompression command-line flag.
It has been appeared that the registration of new time series slows down linearly
with the number of indexdb parts, since VictoriaMetrics needs to check every indexdb part
when it searches for TSID by newly ingested metric name.
The number of in-memory parts grows when new time series are registered
at high rate. The number of in-memory parts grows faster on systems with big number
of CPU cores, because the mergeset maintains per-CPU buffers with newly added entries
for the indexdb, and every such entry is transformed eventually into a separate in-memory part.
The solution has been suggested in https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5212
by @misutoth - to limit the number of in-memory parts with buffered channel.
This solution is implemented in this commit. Additionally, this commit merges per-CPU parts
into a single part before adding it to the list of in-memory parts. This reduces CPU load
when searching for TSID by newly ingested metric name.
The https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5212 recommends setting the limit on the number
of in-memory parts to 100, but my internal testing shows that much lower limit 15 works with the same efficiency
on a system with 16 CPU cores while reducing memory usage for `indexdb/dataBlocks` cache by up to 50%.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5190
* lib/promscrape: respect `0` value for `series_limit` param
Respect `0` value for `series_limit` param in `scrape_config`
even if global limit was set via `-promscrape.seriesLimitPerTarget`.
Previously, `0` value will be ignored in favor of `-promscrape.seriesLimitPerTarget`.
This behavior aligns with possibility to override `series_limit` value via
relabeling with `__series_limit__` label.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* Update docs/CHANGELOG.md
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
Previously, it was not possible to determine which tenant sends metrics with excessive amount of labels of label values.
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
* vmui: fix the logic of closing the popper #5470
* vmui: fix the logic of caching autocomplete results #5472
---------
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
* lib/promscrape/discovery/kubernetes: fix watcher start order for roles endpoints and endpointslice
Previously the groupWatcher could be mistakenly stopped when requests for pod or services resources take too long.
* remove mislead comment
* docs/sd_configs.md: mention -promscrape.kubernetes.attachNodeMetadataAll flag in the description for attach_metadata section
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4640
* wip
* lib/promscrape/kubernetes: prevent from stopping groupWatcher when there are in-flight apiWatcher.mustStart() calls
groupWatcher is stopped if it has zero registered apiWatchers during 14 seconds.
But such a groupWatcher can be still in use if apiWatcher for `role: endpoints` or `role: endpointslice`
is being registered and the discovery of the associated `pod` and/or `service` objects takes longer
than 14 seconds - see the beginning of groupWatcher.startWatchersForRole() function for details.
Track the number of in-flight calls to apiWatcher.mustStart() and prevent from stopping the associated groupWatcher
if the number of in-flight calls is non-zero.
P.S. postponing the discovery of `pod` and/or `service` objects associated with `endpoints` or `endpointslice` roles
isn't the best solution, since it slows down initial discovery of `endpoints` and `endpointslice` targets.
* typo fix
---------
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
Examples:
1) -metricsAuthKey=file:///abs/path/to/file - reads flag value from the given absolute filepath
2) -metricsAuthKey=file://./relative/path/to/file - reads flag value from the given relative filepath
3) -metricsAuthKey=http://some-host/some/path?query_arg=abc - reads flag value from the given url
The flag value is automatically updated when the file contents changes.
* app/vmauth: adds metric_labels and backend_errors counter
it must improve observability for user requests with new metric - per user backend errors counter.
it's needed to calculate requests fail rate to the configured backends.
metric_labels configuration allows to perform additional aggregations on top of multiple users from configuration section.
It could be multiple clients or clients with separate read/write tokens
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5565
* wip
---------
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
* app/vmselect/promql: properly handle possible negative results caused by float operations precision error in rollup functions like rate() or increase()
* fix test
- docs/sd_configs.md: moved hetzner_sd_configs docs to the correct place according to alphabetical order of SD names,
document missing __meta_hetzner_role label.
- lib/promscrape/config.go: added missing MustStop() call for Hetzner SD,
and moved the code to the correct place according to alphabetical order of SD names.
- lib/promscrape/discovery/hetzner: properly handle pagination for hloud API responses,
populate missing __meta_hetzner_role label like Prometheus does.
- Properly populate __meta_hetzner_public_ipv6_network label like Prometheus does.
- Remove unused SDConfig.Token.
- Remove "omitempty" annotation from SDConfig.Role field, since this field is mandatory.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5550
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3154
Properly determine time range search for instant queries with too big look-behind window like `foo[100y]`.
Previously, such queries could return empty responses even if `foo` is present in database.
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5553
Signed-off-by: hagen1778 <roman@victoriametrics.com>
- Clarify the bugfix description at docs/CHANGELOG.md
- Simplify the code by accessing prefetchedMetricIDs struct under the lock
instead of using lockless access to immutable struct.
This shouldn't worsen code scalability too much on busy systems with many CPU cores,
since the code executed under the lock is quite small and fast.
This allows removing cloning of prefetchedMetricIDs struct every time
new metric names are pre-fetched. This should reduce load on Go GC,
since the cloning of uin64set.Set struct allocates many new objects.
Before, this cache was limited only by size.
Cache invalidation by time happens with jitter to prevent thundering herd problem.
Signed-off-by: hagen1778 <roman@victoriametrics.com>
It was calculating the number of dropped time series instead of the number of dropped samples.
While at it, drop vmalert_remotewrite_dropped_bytes_total metric, since it was inconsistently calculated -
at one place it was calculating raw protobuf-encoded sample sizes, while at another place it was calculating
the size of snappy-compressed prompbmarshal.WriteRequest protobuf message.
Additionally, this metric has zero practical sense, so just drop it in order to reduce the level of confusion.
