Previously url watchers for pod, service and node objects could be mistakenly closed
when service discovery was set up only for endpoints and endpointslice roles,
since watchers for these roles may start start pod, service and node url watchers
with nil apiWatcher passed to groupWatcher.startWatchersForRole().
Now all the url watchers, which belong to a particular groupWatcher, are stopped at once
when this groupWatcher has no apiWatcher subscribers.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5216
The issue has been introduced in v1.93.5 when addressing https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4850
* do not print redundant error logs when failed to scrape consul or nomad target
prometheus performs the same because it uses consul lib which just drops the error(1806bcb38c/api/api.go (L1134))
lib/promscrape/discovery/kubernetes: supress context.Cancelled error in logs
It is possible that context.Cancelled will appear after k8s watcher was closed due to reload(see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4850).
Logging an error misinforms user and looks like vmagent discovery will stop working even though this does not affect discovery.
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* lib/backup: fix issue with inconsistent copying of appliedRetention.txt
appliedRetention.txt can be modified in place, so it should be always copied just the same as parts.json
Updates: https://github.com/victoriaMetrics/victoriaMetrics/issues/5005
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* docs: add changelog entry for appliedRetention.txt copying fix
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
- Move the bugfix description to the correct place in docs/CHANGELOG.md
- Prevent from logging of 'context canceled' errors after the url watcher is stopped,
since these errors are expected and may confuse users.
- Remove unused urlWatcher.refCount field.
- Remove unused urlWatcher.close() method.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4850
* lib/promscrape/discovery/kubernetes: fix leaking api watcher
goroutine which was polling k8s API had no execution control. This leaded to leaking goroutines during config reload.
See: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4850
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* lib/promscrape/discovery/kubernetes: use reference counting for urlWatcher cleanup
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* lib/promscrape/discovery/kubernetes: remove waitgroup sync for goroutines polling API server
This is unnecessary since context will is cancelled and new requests will not be sent. Also, using waitgroup will increase time required to perform reload which might result in missed scrapes.
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* lib/promscrape/discovery/kubernetes: clarify comment
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* Apply suggestions from code review
* lib/promscrape/discovery/kubernetes: address review feedback
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Co-authored-by: Nikolay <nik@victoriametrics.com>
* lib/backup: force copying of parts.json
Copying of parts.json is required because `part.key()` comparison can create same key value for files with different contents. This will result in inconsistent backup being created or restored.
See: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5005
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* lib/backup: ensure parts.json is only copied once
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Co-authored-by: Nikolay <nik@victoriametrics.com>
* lib/vmselectapi: do not send empty label names for labelNames request
it breaks cluster communication, since vmselect incorrectly reads request buffer, leaving unread data on it
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4932
* typo fix
* wip
---------
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
Disallow parsing multitenant token at auth.NewToken().
Use auth.NewTokenPossibleMultitenant() at vminsert only. All the other callers should call auth.NewToken(),
since they do not support multitenant token.
This is a follow-up for f0c06b428e
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4910
Initially, stream parse mode was reading data from response and parsing it on flight. This was causing longer delay to read the whole response and required increasing timeout value to allow data processing while reading. So that 908e35affd increased timeout value to fix this.
But after 74c00a8762 response in stream parse mode is saved into memory and then parsed eliminating necessity of having timeout value higher that for usual scrape.
Updates: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4847
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Opentelemetry format allows histograms with non-counter buckets. In this case it makes no sense to add buckets into database
and save only counter with _count suffix.
It could be used as gauge.
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4814
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
* lib/promrelabel: fix relabeling if clause being applied to labels outside of current context
Relabeling is applied to each metric row separately, but in order to lower amount of memory allocations it is reusing labels.
Functions which are working on current metric row labels are supposed to use only current metric labels by using provided offset, but if clause matcher was using the whole labels set instead of local metrics.
This leaded to invalid relabeling results such as one described here: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4806
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* docs/CHANGELOG.md: document the bugfix
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1998
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4806
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
* app/vmbackupmanager/storage: fix path join for windows
See: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4704
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* lib/backup: fixes for windows support
- close dir before running os.RemoveAll. Windows FS does not allow to delete directory before all handles will be closed.
- add path "normalization" for local FS to use the same format of paths for both *unix and Windows
See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4704
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
* Revert "vmalert: unittest support stale datapoint (#4696)"
This reverts commit 0b44df7ec8.
* Revert "docs: specify min version and limitations for vmalert's unit tests"
This reverts commit a24541bd
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* Revert "vmalert: init unit test (#4596)"
This reverts commit da60a68d
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* docs: mention unittest revert in changelog
Signed-off-by: hagen1778 <roman@victoriametrics.com>
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit 9f1b9b86cc)
Signed-off-by: hagen1778 <roman@victoriametrics.com>
* lib/protoparser: adds opentelemetry parser
app/{vmagent,vminsert}: adds opentelemetry ingestion path
Adds ability to ingest data with opentelemetry protocol
protobuf and json encoding is supported
data converted into prometheus protobuf timeseries
each data type has own converter and it may produce multiple timeseries
from single datapoint (for summary and histogram).
only cumulative aggregationFamily is supported for sum(prometheus
counter) and histogram.
Apply suggestions from code review
Co-authored-by: Roman Khavronenko <roman@victoriametrics.com>
updates deps
fixes tests
wip
wip
wip
wip
lib/protoparser/opentelemetry: moves to vtprotobuf generator
go mod vendor
lib/protoparse/opentelemetry: reduce memory allocations
* wip
- Remove support for JSON parsing, since it is too fragile and is rarely used in practice.
The most clients send OpenTelemetry metrics in protobuf.
The JSON parser can be added in the future if needed.
- Remove unused code from lib/protoparser/opentelemetry/pb and lib/protoparser/opentelemetry/proto
- Do not re-use protobuf message between ParseStream() calls, since there is high chance
of high fragmentation of the re-used message because of too complex nested structure of the message.
* wip
* wip
* wip
---------
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
