### Describe Your Changes
Follow-up to
https://github.com/VictoriaMetrics/VictoriaMetrics/pull/8462
Addressed review comments:
- Log panic with FATAL prefix to indicate possible on-disk data
corruption
- Moved version bump line to the tip block (v1.114.0 has already been
released) in changelog
- Removed duplicate vmagent entry from targets list from Makefile
### Checklist
The following checks are **mandatory**:
- [x] My change adheres to [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Co-authored-by: Roman Khavronenko <roman@victoriametrics.com>
This commit improves how vmagent selects the remote write protocol.
Previously, vmagent [performed a handshake
probe](0ff1a3b154/lib/protoparser/protoparserutil/vmproto_handshake.go (L11))
at
[startup](0ff1a3b154/app/vmagent/remotewrite/client.go (L173)):
- If the probe succeeded, it used the VictoriaMetrics (VM) protocol.
- If the probe failed, it downgraded to the Prometheus protocol.
- No protocol changes occurred after the initial probe at runtime.
However, this approach had limitations:
- If vmstorage was unavailable during vmagent startup, vmagent would
immediately downgrade to the Prometheus protocol, leading to higher
network usage unitl vmagent restarted. This case has been reported in
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7615.
- If the remote write server was updated or downgraded (e.g., during a
fallback or migration), vmagent would not detect the protocol change. It
would continue retrying failed requests and eventually drop them.
Require a restart of vmagent to pick up the new protocol.
This commit introduces a more adaptive mechanism.
vmagent always starts with the VM protocol and downgrades to the
Prometheus protocol only if an unsupported media type or bad request
response is received.
When this happens, the protocol is downgraded for all future requests.
In-flight requests are re-packed from Zstd to Snappy and retried
immediately.
Snappy-encoded requests are dropped if an unsupported media type or bad
request is received (no retrying).
Additionally, the in-memory and persisted queues could mix snappy and
zstd encoded blocks. The proper encoding is decided before sending by
encoding.IsZstd function.
TODO:
* [x] Add tests
* [x] Update documentation
* [x] Changelog
* [x] Research on
[content-type](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/8462#issuecomment-2786918054),
[accept-encoding](https://github.com/VictoriaMetrics/VictoriaMetrics/pull/8462#issuecomment-2786923382)
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7615#top
issue.
The following checks are **mandatory**:
- [x] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
This reduces CPU usage by up to 30% in exchange of the increased RAM usage by 10%
when scraping thousands of targets, which expose millions of metrics in summary.
This looks like a good tradeoff after the commit edac875179 ,
which reduced RAM usage by more than 10%, so the final RAM usage for vmagent
is still lower than the RAM usage at v1.114.0 by ~15%, while CPU usage drops by 30%.
### Describe Your Changes
Fix many spelling errors and some grammar, including misspellings in
filenames.
The change also fixes a typo in metric `vm_mmaped_files` to `vm_mmapped_files`.
While this is a breaking change, this metric isn't used in alerts or dashboards.
So it seems to have low impact on users.
The change also deprecates `cspell` as it is much heavier and less usable.
---------
Co-authored-by: Andrii Chubatiuk <achubatiuk@victoriametrics.com>
Co-authored-by: Andrii Chubatiuk <andrew.chubatiuk@gmail.com>
(cherry picked from commit 76d205feae)
Signed-off-by: hagen1778 <roman@victoriametrics.com>
- Properly decode protobuf-encoded Loki request if it has no Content-Encoding header.
Protobuf Loki message is snappy-encoded by default, so snappy decoding must be used
when Content-Encoding header is missing.
- Return back the previous signatures of parseJSONRequest and parseProtobufRequest functions.
This eliminates the churn in tests for these functions. This also fixes broken
benchmarks BenchmarkParseJSONRequest and BenchmarkParseProtobufRequest, which consume
the whole request body on the first iteration and do nothing on subsequent iterations.
- Put the CHANGELOG entries into correct places, since they were incorrectly put into already released
versions of VictoriaMetrics and VictoriaLogs.
- Add support for reading zstd-compressed data ingestion requests into the remaining protocols
at VictoriaLogs and VictoriaMetrics.
- Remove the `encoding` arg from PutUncompressedReader() - it has enough information about
the passed reader arg in order to properly deal with it.
- Add ReadUncompressedData to lib/protoparser/common for reading uncompressed data from the reader until EOF.
This allows removing repeated code across request-based protocol parsers without streaming mode.
- Consistently limit data ingestion request sizes, which can be read by ReadUncompressedData function.
Previously this wasn't the case for all the supported protocols.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/8416
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8380
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8300
### Describe Your Changes
revert doc change in
815bad3687
and move new content to another section.
### Checklist
The following checks are **mandatory**:
- [x] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit 9548b7e442)
This commit introduces common readers for multiple compression encoding algorithms.
Currently, supported encodings are:
* zstd
* gzip
* deflat
* snappy
It adds new common reader to the all VictoriaLogs ingestion protocols.
And updates opentelemetry metrics parsing for VictoriaMetrics components.
Also, it ports zstd stream parses from cluster branch.
Related issues:
fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8380
fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8300
---------
Co-authored-by: Aliaksandr Valialkin <valyala@victoriametrics.com>
Co-authored-by: f41gh7 <nik@victoriametrics.com>
Previously, if the command-line flag value `-remoteWrite.showURL` changed, vmagent dropped content of persistent queues. It's not expected behavior and may lead to data-loss at queue.
Further more if command-line flag value `-remoteWrite.showURL` is set to `true`, any changes to url query arguments will lead to persistent queue drop. The most common uses is kafka and gcp pub-sub integration. It uses url query arguments for client configuration.
Also, it complicates copy content of persistent queue between vmagents. Since it requires to properly change name inside metainfo.json.
This commit removes persistent queue name equality check from `lib/persistentqueue`. This check was added as an additional protection from on-disk data corruption.
It's safe to skip this check for vmagent, because vmagent encodes remoteWrite.url as part of path to the queue. It guarantees that there will be no collision.
related issue:
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/8477.
### Checklist
The following checks are **mandatory**:
- [x] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Signed-off-by: f41gh7 <nik@victoriametrics.com>
Co-authored-by: f41gh7 <nik@victoriametrics.com>
`MustParsePromMetrics` imports `lib/protoparser/prometheus`, and this
package exposes the following metrics:
```
vm_protoparser_rows_read_total{type="promscrape"}
vm_rows_invalid_total{type="prometheus"}
```
It means every package that uses `lib/prompbmarshal` will start exposing
these metrics. For example, vlogs imports `lib/protoparser/common` which
uses `lib/prompbmarshal.Label`. And only because of this vlogs starts
exposing unrelated prometheus metrics on /metrics page.
Moving `MustParsePromMetrics` to `lib/protoparser/prometheus` seems like
the leas intrusive change.
-----------
Depends on another change
https://github.com/VictoriaMetrics/VictoriaMetrics/pull/8403
Signed-off-by: hagen1778 <roman@victoriametrics.com>
### Describe Your Changes
By default, stream aggregation and deduplication stores a single state
per each aggregation output result.
The data for each aggregator is flushed independently once per
aggregation interval. But there's no guarantee that
incoming samples with timestamps close to the aggregation interval's end
will get into it. For example, when aggregating
with `interval: 1m` a data sample with timestamp 1739473078 (18:57:59)
can fall into aggregation round `18:58:00` or `18:59:00`.
It depends on network lag, load, clock synchronization, etc. In most
scenarios it doesn't impact aggregation or
deduplication results, which are consistent within margin of error. But
for metrics represented as a collection of series,
like
[histograms](https://docs.victoriametrics.com/keyconcepts/#histogram),
such inaccuracy leads to invalid aggregation results.
For this case, streaming aggregation and deduplication support mode with
aggregation windows for current and previous state. With this mode,
flush doesn't happen immediately but is shifted by a calculated samples
lag that improves correctness for delayed data.
Enabling of this mode has increased resource usage: memory usage is
expected to double as aggregation will store two states
instead of one. However, this significantly improves accuracy of
calculations. Aggregation windows can be enabled via
the following settings:
- `-streamAggr.enableWindows` at [single-node
VictoriaMetrics](https://docs.victoriametrics.com/single-server-victoriametrics/)
and [vmagent](https://docs.victoriametrics.com/vmagent/). At
[vmagent](https://docs.victoriametrics.com/vmagent/)
`-remoteWrite.streamAggr.enableWindows` flag can be specified
individually per each `-remoteWrite.url`.
If one of these flags is set, then all aggregators will be using fixed
windows. In conjunction with `-remoteWrite.streamAggr.dedupInterval` or
`-streamAggr.dedupInterval` fixed aggregation windows are enabled on
deduplicator as well.
- `enable_windows` option in [aggregation
config](https://docs.victoriametrics.com/stream-aggregation/#stream-aggregation-config).
It allows enabling aggregation windows for a specific aggregator.
### Checklist
The following checks are **mandatory**:
- [ ] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit c8fc903669)
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Historically some of VictoriaMetrics components were optimized for the low rate of memory allocations.
These are: vmagent, single-node VictoriaMetrics and vmstorage. These components benefit from the low
GOGC value, since this allow reducing their memory usage in steady state on typical workloads.
Other VictoriaMetrics components aren't optimized for the reduced rate of memory allocations.
This results in the increased CPU usage spent on garbage collection (GC) in these components,
since it must be triggered at higher rate. See https://tip.golang.org/doc/gc-guide#GOGC for details.
These components do not use too much memory, so it is OK increasing the GOGC for these components
from 30 to 100 - this won't affect the most users.
Keep GOGC to 30 only for vmagent, single-node VictoriaMetrics and vmstorage components.
See 077193d87c and 54b9e1d3cb .
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7902
Previously, time series with labels exceeding the configured limits were truncated and written to storage, potentially causing data inconsistency. This could lead to collisions between time series and make it difficult to identify the source due to truncated labels.
This commit changes the behavior:
* Such time series are now rejected outright.
* Rejected time series are logged to stdout, and corresponding counters are incremented.
* removes `vm_too_long_label_values_total`, `vm_too_long_label_names_total`, `vm_metrics_with_dropped_labels_total` metrics.
* adds new values `[too_many_labels,too_long_label_name,too_long_label_value]` to `reason` label of the `vm_rows_ignored_total` metric name
related issues:
- https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6928
- https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7661
Removes global defaultAuthToken, since it's no longer needed.
It was added as fallback for 'remoteWrite.multitenantURL' feature.
This feature was deprecated at v1.102 version and removed.
Updates newRemoteWriteCtxs function, it shouldn't accept auth.Token no longer.
This was also a part of remove feature.
Signed-off-by: f41gh7 <nik@victoriametrics.com>
Previously, vmagent produced parsing error for 'multitenant' auth token
value for the cases:
* data ingestion with enableMultitentEndpoints
* data scrapping at promscrape
It's inconsistent to the other VictoriaMetrics components.
Since 'multitenant' is well-known token value for multitenancy via
labels. And vmagent is intended to be compatible with vminsert ingestion
endpoints.
This commit replaces NewToken with NewTokenPossibleMultitenant function
for token parsing. It allows to use multitenant value for it. And it
makes token values consistent for the all components.
Related issue:
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/7694
Previously unit `m` is not correctly supported.
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
### Describe Your Changes
The flags docs mention the flag that does not exist (and never existed).
Perhaps that was a typo.
`s/retryMaxInterval/retryMaxTime/g`
### Checklist
The following checks are **mandatory**:
- [x] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
Signed-off-by: Artem Fetishev <rtm@victoriametrics.com>
(cherry picked from commit e2c73dc89f)
….url` using `-remoteWrite.streamAggr.dropInputLabels`
Before, labels were set to all the `remoteWrite.url`.
address https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6780
---------
Co-authored-by: Roman Khavronenko <roman@victoriametrics.com>
(cherry picked from commit fbde238cdc)
Recent versions of `docker build` started generating the InvalidDefaultArgInFrom warning if Dockerfile contains
an ARG without default value. While this warning doesn't affect building Docker packages via `make package-*` commands,
it is better suppressing the warning, so it doesn't clutter `make package-*` output with the noise,
which can hide real issues in the future.
…specifying `-streamAggr.dedupInterval` or
`-remoteWrite.streamAggr.dedupInterval` command-line flag
[The
documentation](https://docs.victoriametrics.com/stream-aggregation/)
contains conflicting descriptions regarding deduplication for
non-matched series when `-remoteWrite.streamAggr.config` and / or
`-streamAggr.config` are set:
1. Statement below says **all the received data** is deduplicated:
>[vmagent](https://docs.victoriametrics.com/vmagent/) supports
relabeling, deduplication and stream aggregation for all the received
data, scraped or pushed. Then, the collected data will be forwarded to
specified -remoteWrite.url destinations. The data processing order is
the following:
>1. all the received data is relabeled according to the specified
[-remoteWrite.relabelConfig](https://docs.victoriametrics.com/vmagent/#relabeling)
(if it is set)
>2. all the received data is deduplicated according to specified
[-streamAggr.dedupInterval](https://docs.victoriametrics.com/stream-aggregation/#deduplication)
(if it is set to duration bigger than 0)
2. Another statement says the deduplication is performed individually
for the **matching samples**
>The de-deduplication is performed after applying
[relabeling](https://docs.victoriametrics.com/vmagent/#relabeling) and
before performing the aggregation. If the -remoteWrite.streamAggr.config
and / or -streamAggr.config is set, then the de-duplication is performed
individually per each [stream aggregation
config](https://docs.victoriametrics.com/stream-aggregation/#stream-aggregation-config)
for the matching samples after applying
[input_relabel_configs](https://docs.victoriametrics.com/stream-aggregation/#relabeling).
Considering the following deduplication use cases:
1. To apply deduplication(globally or for specific remoteWrite
destination) for all the received data, scraped or pushed
--- using `-streamAggr.dedupInterval` or
`-remoteWrite.streamAggr.dedupInterval`.
2. To deduplicate and aggregate metrics that match the rule `match`
filters
--- using `-remoteWrite.streamAggr.config` and specifiying
`dedup_interval` option in [stream aggregation
config](https://docs.victoriametrics.com/stream-aggregation/#stream-aggregation-config).
3. To deduplicate all the received data while having `streamAggr.config`
for some metrics
--- no way for a single vmagent now, need to set up two level vmagents
This PR implements case3.
---------
Co-authored-by: Roman Khavronenko <roman@victoriametrics.com>
(cherry picked from commit d523015f27)
## Describe Your Changes
Add RemoteWrite Retry Controls
This PR introduces two new flags to the remote write functionality:
- remoteWrite.retryMinInterval
- remoteWrite.retryMaxTime
These flags provide finer control over the retry behavior for
remoteWrite operations, allowing users to customize the minimum interval
between retries and the maximum duration for retry attempts.
Fixes#5486.
## Checklist
- [x] The following checks are mandatory:
My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Signed-off-by: Yury Akudovich <ya@matterlabs.dev>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit d0f5a9d77a)
Describe Your Changes
When I use usePromCompatibleNaming with vmagent to process data that
needs to be formatted from different sources such as InfluxDB, I find
that it doesn’t work
However, it works in vminsert. I found that vminsert uses the
HasRelabeling method to determine whether to relabel.
```go
func HasRelabeling() bool {
pcs := pcsGlobal.Load()
return pcs.Len() > 0 || *usePromCompatibleNaming
}
```
in vmagent, the decision to relabel is determined only by
pcsGlobal.Len() > 0. However, in the applyRelabeling method, the
usePromCompatibleNaming logic is also used to determine whether to
relabel in the error handling.
```go
func (rctx *relabelCtx) applyRelabeling(tss []prompbmarshal.TimeSeries, pcs *promrelabel.ParsedConfigs) []prompbmarshal.TimeSeries {
if pcs.Len() == 0 && !*usePromCompatibleNaming {
// Nothing to change.
return tss
}
```
So I think that the logic for determining whether to relabel in vmagent
is not as expected.
Checklist
The following checks are mandatory:
[✅]My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Co-authored-by: Roman Khavronenko <hagen1778@gmail.com>
(cherry picked from commit d134a310f3)
…eep_metric_names` options in stream aggregation config together
With aggregated data and raw data under the same metric, results would
be confusing.
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit 62d19369a3)
Make `-remoteWrite.streamAggr.ignoreFirstIntervals` of array type so it could
accept multiple values which can be applied to the corresponding`-remoteWrite.url`.
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit 8f5c26d788)
Fix `-streamAggr.dropInputLabels` behavior when global deduplication is enabled without `-streamAggr.config`.
Previously, `-remoteWrite.streamAggr.dropInputLabels` is misapplied.
---------
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit 4863605469)
### Describe Your Changes
This is useful for clients which validate InfluxDB is available before
data ingestion can be started.
See: https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6653
### Checklist
The following checks are **mandatory**:
- [x] My change adheres [VictoriaMetrics contributing
guidelines](https://docs.victoriametrics.com/contributing/).
---------
Signed-off-by: Zakhar Bessarab <z.bessarab@victoriametrics.com>
Signed-off-by: hagen1778 <roman@victoriametrics.com>
Co-authored-by: hagen1778 <roman@victoriametrics.com>
(cherry picked from commit 9877a5e7d5)
Signed-off-by: hagen1778 <roman@victoriametrics.com>
The %q formatter may result in incorrectly formatted JSON string if the original string
contains special chars such as \x1b . They must be encoded as \u001b , otherwise the resulting JSON string
cannot be parsed by JSON parsers.
This is a follow-up for c0caa69939
See https://github.com/VictoriaMetrics/victorialogs-datasource/issues/24
- Rename GetStatDialFunc to NewStatDialFunc, since it returns new function with every call
- NewStatDialFunc isn't related to http in any way, so it must be moved from lib/httputils to lib/netutil
- Simplify the implementation of NewStatDialFunc by removing sync.Map from there.
- Use netutil.NewStatDialFunc at app/vmauth and lib/promscrape/discoveryutils
- Use gauge instead of counter type for *_conns metric
This is a follow-up for d7b5062917
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6299
- Move the remaining code responsible for stream aggregation initialization from remotewrite.go to streamaggr.go .
This improves code maintainability a bit.
- Properly shut down streamaggr.Aggregators initialized inside remotewrite.CheckStreamAggrConfigs().
This prevents from potential resource leaks.
- Use separate functions for initializing and reloading of global stream aggregation and per-remoteWrite.url stream aggregation.
This makes the code easier to read and maintain. This also fixes INFO and ERROR logs emitted by these functions.
- Add an ability to specify `name` option in every stream aggregation config. This option is used as `name` label
in metrics exposed by stream aggregation at /metrics page. This simplifies investigation of the exposed metrics.
- Add `path` label additionally to `name`, `url` and `position` labels at metrics exposed by streaming aggregation.
This label should simplify investigation of the exposed metrics.
- Remove `match` and `group` labels from metrics exposed by streaming aggregation, since they have little practical applicability:
it is hard to use these labels in query filters and aggregation functions.
- Rename the metric `vm_streamaggr_flushed_samples_total` to less misleading `vm_streamaggr_output_samples_total` .
This metric shows the number of samples generated by the corresponding streaming aggregation rule.
This metric has been added in the commit 861852f262 .
See https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6462
- Remove the metric `vm_streamaggr_stale_samples_total`, since it is unclear how it can be used in practice.
This metric has been added in the commit 861852f262 .
See https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6462
- Remove Alias and aggrID fields from streamaggr.Options struct, since these fields aren't related to optional params,
which could modify the behaviour of the constructed streaming aggregator.
Convert the Alias field to regular argument passed to LoadFromFile() function, since this argument is mandatory.
- Pass Options arg to LoadFromFile() function by reference, since this structure is quite big.
This also allows passing nil instead of Options when default options are enough.
- Add `name`, `path`, `url` and `position` labels to `vm_streamaggr_dedup_state_size_bytes` and `vm_streamaggr_dedup_state_items_count` metrics,
so they have consistent set of labels comparing to the rest of streaming aggregation metrics.
- Convert aggregator.aggrStates field type from `map[string]aggrState` to `[]aggrOutput`, where `aggrOutput` contains the corresponding
`aggrState` plus all the related metrics (currently only `vm_streamaggr_output_samples_total` metric is exposed with the corresponding
`output` label per each configured output function). This simplifies and speeds up the code responsible for updating per-output
metrics. This is a follow-up for the commit 2eb1bc4f81 .
See https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6604
- Added missing urls to docs ( https://docs.victoriametrics.com/stream-aggregation/ ) in error messages. These urls help users
figuring out why VictoriaMetrics or vmagent generates the corresponding error messages. The urls were removed for unknown reason
in the commit 2eb1bc4f81 .
- Fix incorrect update for `vm_streamaggr_output_samples_total` metric in flushCtx.appendSeriesWithExtraLabel() function.
While at it, reduce memory usage by limiting the maximum number of samples per flush to 10K.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5467
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6268
The old link was changed globally to the new link in the commit f4b1cbfef0 .
Unfortunately, old links are still posted in new commits :(
This is a follow-up for 680b8c25c8 .
While at it, remove duplicate 'len(*remoteWriteURLs) > 0' check in the remotewrite.Init() functions,
since this check is already made at the beginning of the function.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6253
- Drop samples and return true from remotewrite.TryPush() at fast path when all the remote storage
systems are configured with the disabled on-disk queue, every in-memory queue is full
and -remoteWrite.dropSamplesOnOverload is set to true. This case is quite common,
so it should be optimized. Previously additional CPU time was spent on per-remoteWriteCtx
relabeling and other processing in this case.
- Properly count the number of dropped samples inside remoteWriteCtx.pushInternalTrackDropped().
Previously dropped samples were counted only if -remoteWrite.dropSamplesOnOverload flag is set.
In reality, the samples are dropped when they couldn't be sent to the queue because in-memory queue is full
and on-disk queue is disabled.
The remoteWriteCtx.pushInternalTrackDropped() function is called by streaming aggregation for pushing
the aggregated data to the remote storage. Streaming aggregation cannot wait until the remote storage
processes pending data, so it drops aggregated samples in this case.
- Clarify the description for -remoteWrite.disableOnDiskQueue command-line flag at -help output,
so it is clear that this flag can be set individually per each -remoteWrite.url.
- Make the -remoteWrite.dropSamplesOnOverload flag global. If some of the remote storage systems
are configured with the disabled on-disk queue, then there is no sense in keeping samples
on some of these systems, while dropping samples on the remaining systems, since this
will result in global stall on the remote storage system with the disabled on-disk queue
and with the -remoteWrite.dropSamplesOnOverload=false flag. vmagent will always return false
from remotewrite.TryPush() in this case. This will result in infinite duplicate samples
written to the remaining remote storage systems. That's why the -remoteWrite.dropSamplesOnOverload
is forcibly set to true if more than one -remoteWrite.disableOnDiskQueue flag is set.
This allows proceeding with newly scraped / pushed samples by sending them to the remaining
remote storage systems, while dropping them on overloaded systems with the -remoteWrite.disableOnDiskQueue flag set.
- Verify that the remoteWriteCtx.TryPush() returns true in the TestRemoteWriteContext_TryPush_ImmutableTimeseries test.
- Mention in vmagent docs that the -remoteWrite.disableOnDiskQueue command-line flag can be set individually per each -remoteWrite.url.
See https://docs.victoriametrics.com/vmagent/#disabling-on-disk-persistence
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6248
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6065
Consistently using t.Fatal* simplifies the test code and makes it less fragile, since it is common error
to forget to make proper cleanup after t.Error* call. Also t.Error* calls do not provide any practical
benefits when some tests fail. They just clutter test output with additional noise information,
which do not help in fixing failing tests most of the time.
This is a follow-up for a9525da8a4
'any' type is supported starting from Go1.18. Let's consistently use it
instead of 'interface{}' type across the code base, since `any` is easier to read than 'interface{}'.
Reason for revert:
There are many statsd servers exist:
- https://github.com/statsd/statsd - classical statsd server
- https://docs.datadoghq.com/developers/dogstatsd/ - statsd server from DataDog built into DatDog Agent ( https://docs.datadoghq.com/agent/ )
- https://github.com/avito-tech/bioyino - high-performance statsd server
- https://github.com/atlassian/gostatsd - statsd server in Go
- https://github.com/prometheus/statsd_exporter - statsd server, which exposes the aggregated data as Prometheus metrics
These servers can be used for efficient aggregating of statsd data and sending it to VictoriaMetrics
according to https://docs.victoriametrics.com/#how-to-send-data-from-graphite-compatible-agents-such-as-statsd (
the https://github.com/prometheus/statsd_exporter can be scraped as usual Prometheus target
according to https://docs.victoriametrics.com/#how-to-scrape-prometheus-exporters-such-as-node-exporter ).
Adding support for statsd data ingestion protocol into VictoriaMetrics makes sense only if it provides
significant advantages over the existing statsd servers, while has no significant drawbacks comparing
to existing statsd servers.
The main advantage of statsd server built into VictoriaMetrics and vmagent - getting rid of additional statsd server.
The main drawback is non-trivial and inconvenient streaming aggregation configs, which must be used for the ingested statsd metrics (
see https://docs.victoriametrics.com/stream-aggregation/ ). These configs are incompatible with the configs for standalone statsd servers.
So you need to manually translate configs of the used statsd server to stream aggregation configs when migrating
from standalone statsd server to statsd server built into VictoriaMetrics (or vmagent).
Another important drawback is that it is very easy to shoot yourself in the foot when using built-in statsd server
with the -statsd.disableAggregationEnforcement command-line flag or with improperly configured streaming aggregation.
In this case the ingested statsd metrics will be stored to VictoriaMetrics as is without any aggregation.
This may result in high CPU usage during data ingestion, high disk space usage for storing all the unaggregated
statsd metrics and high CPU usage during querying, since all the unaggregated metrics must be read, unpacked and processed
during querying.
P.S. Built-in statsd server can be added to VictoriaMetrics and vmagent after figuring out more ergonomic
specialized configuration for aggregating of statsd metrics. The main requirements for this configuration:
- easy to write, read and update (ideally it should work out of the box for most cases without additional configuration)
- hard to misconfigure (e.g. hard to shoot yourself in the foot)
It would be great if this configuration will be compatible with the configuration of the most widely used statsd server.
In the mean time it is recommended continue using external statsd server.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6265
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5053
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5052
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/206
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4600
This reverts commit 5a3abfa041.
Reason for revert: exemplars aren't in wide use because they have numerous issues which prevent their adoption (see below).
Adding support for examplars into VictoriaMetrics introduces non-trivial code changes. These code changes need to be supported forever
once the release of VictoriaMetrics with exemplar support is published. That's why I don't think this is a good feature despite
that the source code of the reverted commit has an excellent quality. See https://docs.victoriametrics.com/goals/ .
Issues with Prometheus exemplars:
- Prometheus still has only experimental support for exemplars after more than three years since they were introduced.
It stores exemplars in memory, so they are lost after Prometheus restart. This doesn't look like production-ready feature.
See 0a2f3b3794/content/docs/instrumenting/exposition_formats.md (L153-L159)
and https://prometheus.io/docs/prometheus/latest/feature_flags/#exemplars-storage
- It is very non-trivial to expose exemplars alongside metrics in your application, since the official Prometheus SDKs
for metrics' exposition ( https://prometheus.io/docs/instrumenting/clientlibs/ ) either have very hard-to-use API
for exposing histograms or do not have this API at all. For example, try figuring out how to expose exemplars
via https://pkg.go.dev/github.com/prometheus/client_golang@v1.19.1/prometheus .
- It looks like exemplars are supported for Histogram metric types only -
see https://pkg.go.dev/github.com/prometheus/client_golang@v1.19.1/prometheus#Timer.ObserveDurationWithExemplar .
Exemplars aren't supported for Counter, Gauge and Summary metric types.
- Grafana has very poor support for Prometheus exemplars. It looks like it supports exemplars only when the query
contains histogram_quantile() function. It queries exemplars via special Prometheus API -
https://prometheus.io/docs/prometheus/latest/querying/api/#querying-exemplars - (which is still marked as experimental, btw.)
and then displays all the returned exemplars on the graph as special dots. The issue is that this doesn't work
in production in most cases when the histogram_quantile() is calculated over thousands of histogram buckets
exposed by big number of application instances. Every histogram bucket may expose an exemplar on every timestamp shown on the graph.
This makes the graph unusable, since it is litterally filled with thousands of exemplar dots.
Neither Prometheus API nor Grafana doesn't provide the ability to filter out unneeded exemplars.
- Exemplars are usually connected to traces. While traces are good for some
I doubt exemplars will become production-ready in the near future because of the issues outlined above.
Alternative to exemplars:
Exemplars are marketed as a silver bullet for the correlation between metrics, traces and logs -
just click the exemplar dot on some graph in Grafana and instantly see the corresponding trace or log entry!
This doesn't work as expected in production as shown above. Are there better solutions, which work in production?
Yes - just use time-based and label-based correlation between metrics, traces and logs. Assign the same `job`
and `instance` labels to metrics, logs and traces, so you can quickly find the needed trace or log entry
by these labes on the time range with the anomaly on metrics' graph.
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/5982
