Commit graph

568 commits

Author SHA1 Message Date
Aliaksandr Valialkin
acb7a95c64 app/vmselect: follow-up after aa11ef6d3b
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1413
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/4
2021-07-07 17:45:09 +03:00
tony
aa11ef6d3b
add vmui for vmselect component (#1431)
Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
2021-07-07 17:04:23 +03:00
Aliaksandr Valialkin
9c19719ad6 app/{vminsert,vmselect}: export vminsert_request_duration_seconds and vmselect_request_duration_seconds histograms 2021-07-07 13:27:23 +03:00
Aliaksandr Valialkin
ceda2b1df4 lib/httpserver: print full requestURI in httpserver.Errorf
This should simplify debugging.
2021-07-07 13:11:29 +03:00
Aliaksandr Valialkin
44855f0c9b app/{vmselect,vmstorage}: clarify the description for -dedup.minScrapeInterval command-line flag
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1426
2021-07-02 15:06:41 +03:00
Aliaksandr Valialkin
856aecae05 app/vmselect/promql: return the last timestamp for the max / min value from tmax_over_time() and tmin_over_time() function as most users expect 2021-06-23 14:18:37 +03:00
Aliaksandr Valialkin
b92d110cad app/vmselect: log slow requests to all the /api/v1/* handlers if their execution time exceeds -search.logSlowQueryDuration 2021-06-18 19:07:03 +03:00
Roman Khavronenko
a15c947045 promql: fix increase_pure calculation for cases with stale series (#1381)
Due to staleness handling, increase_pure were using incorrect previous value
during calculation in cases where series disappears for period longer
than staleness period and then returns back. The fix suppose to account
for a real datapoint value before staleness takes place. The fix should
remove unexpected spikes while using `increase_pure` for staled series.
2021-06-15 17:37:51 +03:00
John Belmonte
3e79f3994e spelling fix: synonym (#1363) 2021-06-11 10:58:48 +03:00
Aliaksandr Valialkin
e8e7f03394 app/vmselect/promql: typo fix in the comment 2021-06-09 18:34:57 +03:00
Aliaksandr Valialkin
a0b001bfec app/vmselect/netstorage: remove duplicate limiter on concurrent queries
It duplicates the `-search.maxConcurrentRequests` limiter.
2021-05-24 19:13:04 +03:00
Aliaksandr Valialkin
180829b8c2 app/vmselect/promql: add timezone_offset(tz) function
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1306
2021-05-20 11:54:06 +03:00
Nikolay
be87be34a4 Adds tsdb match filters (#1282)
* init work on filters

* init propose for status filters

* fixes tsdb status
adds test

* fix bug

* removes checks from test
2021-05-12 17:16:58 +03:00
Aliaksandr Valialkin
56b08390f6 app/vmselect/promql: allow to use 2x more memory for query processing in cluster mode compared to single-node mode
`vmselect` has no `vmstorage`-related caches. So it can use more memory for query processing compared to single-node VictoriaMetrics.
2021-05-12 14:43:49 +03:00
Aliaksandr Valialkin
6dc5d3b357 all: rename https://victoriametrics.github.io to https://docs.victoriametrics.com 2021-04-20 20:20:01 +03:00
Aliaksandr Valialkin
e48fd5776b app/vmselect/promql: fix tests after d3fa0ccabd 2021-04-08 00:17:40 +03:00
Aliaksandr Valialkin
f90bf265f4 app/vmselect/promql: properly detect aggregate topk* and bottomk* aggregate functions in order to disable duplicate sorting
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1189
2021-04-08 00:10:33 +03:00
Aliaksandr Valialkin
9ce3b7e1dd app/vmselect: return data:null instead of data:[] from /api/v1/query_exemplars, since Grafana throws an error otherwise
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1186
2021-04-07 23:34:01 +03:00
Aliaksandr Valialkin
3d5f1f779f app/vmselect: do not sort series returned from topk* and bottomk* functions, since these series are already sorted in user-expected order
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1189
2021-04-07 14:16:02 +03:00
Aliaksandr Valialkin
7fce4e9fb4 app/vmselect: return dumb response on /api/v1/query_exemplars request
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1186
2021-04-05 23:28:21 +03:00
Aliaksandr Valialkin
2585058a5f Makefile: prepare arm64 and amd64 release archives for cluster version on make release command 2021-04-05 23:01:45 +03:00
Aliaksandr Valialkin
4aa59cae7c app/vmselect/promql: do not delete dst_label if src_label is empty in label_copy(q, src_label, dst_label) and label_move(q, src_label, dst_label) 2021-04-03 22:06:50 +03:00
Aliaksandr Valialkin
b9469de410 app/vmselect/promql: add ability to set label value additionally to label name for the remaining sum of time series returned from topk_* and bottomk_* functions in the form: topk_min(N, m, "label=value") 2021-04-02 23:56:30 +03:00
Aliaksandr Valialkin
4028d692f5 app: do not process non-GET requests on at / handler 2021-04-02 22:56:38 +03:00
Aliaksandr Valialkin
31d3d02d25 app/vmselect/promql: remove the limit on the number of time series that can be sorted, since it may confuse users
Always sort time series returned from `/api/v1/query` and `/api/v1/query_range` unless `sort_*` function is used at top level of the query.
2021-04-02 16:50:02 +03:00
Aliaksandr Valialkin
d8c769e6af app/vmselect: add -search.maxStatusRequestDuration command-line flag for limiting the duration of requests to /api/v1/status/* and /api/v1/series/count 2021-03-30 21:39:32 +03:00
Aliaksandr Valialkin
0e29d48628 app/vmselect: remove mentioning for -search.storageTimeout from command-line flag descriptions
This is a follow-up for ecfd6fe78d
2021-03-30 15:07:40 +03:00
Aliaksandr Valialkin
ecfd6fe78d app/vmselect: remove -search.storageTimeout command-line flag, since it has the same meaning as -search.maxQueryDuration
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711
2021-03-30 14:54:36 +03:00
Aliaksandr Valialkin
7bafaad46d app/vmselect: prevent from possible incomplete query results after timed out query
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/711
2021-03-30 13:34:57 +03:00
Nikolay
022fe4efd0
makes tenant metrics names consistent with enterprise version (#1152) 2021-03-29 11:59:04 +03:00
Aliaksandr Valialkin
698ccca6ad app/vmselect/prometheus: explain why partial responses are disallowed during data export
See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1148
2021-03-25 18:39:07 +02:00
Aliaksandr Valialkin
b473c21915 app/vmselect/promql: do not merge time series during requests to /api/v1/query
See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1141
2021-03-25 13:56:23 +02:00
Aliaksandr Valialkin
d1e773266f app/vmselect/graphite: accept and enforce extra_label in all the Graphite APIs 2021-03-23 15:30:15 +02:00
Aliaksandr Valialkin
27bb614016 app/vmselect: move getEnforcedTagFiltersFromRequest to searchtuils, since it will be used in Graphite functions soon 2021-03-23 14:57:57 +02:00
Aliaksandr Valialkin
7f9863254d app/vmselect: improve description for -search.maxPointsPerTimeseries command-line flag 2021-03-22 16:44:36 +02:00
Aliaksandr Valialkin
92531a38c4 app/vmselect/promql: increment key prefix for faster reset for rollup result cache 2021-03-22 11:59:39 +02:00
Aliaksandr Valialkin
799461d8bf app/vmselect/graphite: follow-up after 529d7be26b 2021-03-18 16:33:00 +02:00
Nikolay
8807410a00 changes metricsFind api (#1137)
it should be able mitigate crash if label value contains *,[ or { symbols
2021-03-18 16:32:59 +02:00
Aliaksandr Valialkin
667d129e1e app/vmselect/promql: typo fix after 9666834045 2021-03-17 15:11:54 +02:00
Aliaksandr Valialkin
69201806f8 app/vmselect/promql: merge adjancent buckets with the smallest summary number of hits in buckets_limit() function
This should improve accuracy for the returned buckets
2021-03-17 14:31:29 +02:00
Aliaksandr Valialkin
05e301cfa0 app/vmselect/netstorage: reduce mutex contention when unpacking data on a system with high number of CPU cores 2021-03-16 21:52:05 +02:00
Aliaksandr Valialkin
f54ece438d app/vmselect/promql: do not crash if histogram_over_time() function name contains uppercase letters such as Histogram_over_time() 2021-03-16 12:23:55 +02:00
Aliaksandr Valialkin
e2717d84c0 all: various fixes in command-line flag descriptions 2021-03-15 22:03:49 +02:00
Aliaksandr Valialkin
30ab0eec27 app/vmselect/promql: reduce overhead on scrape interval estimation
It should be enough to use the first 20 datapoints instead of 100 datapoints for scrape interval estimation.
2021-03-15 20:32:35 +02:00
Aliaksandr Valialkin
8844b38745 app/vmselect/promql: fix tests after 2dae0a2c47 2021-03-15 20:20:27 +02:00
Aliaksandr Valialkin
6b9bba7448 app/vmselect: add round_digits query arg to /api/v1/query and /api/v1/query_range handlers for limiting the number of decimal digits after the point 2021-03-15 12:38:59 +02:00
Aliaksandr Valialkin
b6dcb37fca app/vmselect/prometheus: do not include datapoints with timestamps matching t-d when returning results from /api/v1/query?query=m[d]&time=t as Prometheus does 2021-03-12 12:18:37 +02:00
Aliaksandr Valialkin
942890b1bb app/vmselect/promql: follow up for 433fff0006 2021-03-09 12:54:23 +02:00
Nikolay
28e450cd7c duplicate timeseries fix for prometheus_buckets function (#1119)
* try fix for prometheus_buckets

* merge possible end of the bucket collision
2021-03-09 12:54:21 +02:00
Aliaksandr Valialkin
5277507932 app/vmselect/promql: add histogram_avg(), histogram_stddev() and histogram_stdvar() functions to MetricsQL 2021-03-04 14:12:52 +02:00
Aliaksandr Valialkin
3b1c2f03c3 app/vmselect/querystats: show the number of matching queries in the top by average duration and in the top by summary duration
This should help debugging slow queries.
2021-02-28 19:42:29 +02:00
Nikolay
673b10dd7f adds enforced tag filters into cache key (#1095) 2021-02-27 00:23:38 +02:00
Aliaksandr Valialkin
e38ca28d99 app/vmselect/promql: increase accuracy for buckets_limit() function for small limits by skipping the first and the last buckets during merge
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.
2021-02-26 22:56:04 +02:00
Aliaksandr Valialkin
0144b164c7 app/vmselect: add sign(q) and clamp(q, min, max) functions, which will be added in the upcoming Prometheus release
See https://twitter.com/roidelapluie/status/1363428376162295811

The `last_over_time(m[d])` function already exists in MetricsQL.
2021-02-24 17:26:27 +02:00
Aliaksandr Valialkin
1b90a091cf app/vmselect/netstorage: cleanup after 4805b80977 2021-02-24 11:43:09 +02:00
weng zhao
4805b80977
vmselect: fix vmselect cache partial result when timeout (#1085) 2021-02-24 11:39:56 +02:00
Aliaksandr Valialkin
d16effc29e app/vmselect/promql: properly calculate histogram_quantile() over zero buckets and only a single non-zero le="+Inf"` bucket like Prometheus does 2021-02-24 00:41:30 +02:00
Aliaksandr Valialkin
5dbe88a1c6 app/vmselect/promql: add increase_pure() function to MetricsQL
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/962
2021-02-22 19:15:31 +02:00
Aliaksandr Valialkin
bd1d906eee app/vmselect/promql: reduce the probability of duplicate time series errors when querying Kubernetes metrics 2021-02-18 22:08:13 +02:00
Aliaksandr Valialkin
3062ff0fdb app/vmselect: export per-tenant stats on the number of requests and the cumulative request duration
The metrics are:
- vm_vmselect_http_requests_total{accountID="...",projectID="..."} - the total number of select requests per each tenant
- vm_vmselect_http_duration_ms_total{accountID="...",projectID="..."} - the total duration in milliseconds for per-tenant select requests

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/932
2021-02-16 23:30:29 +02:00
Aliaksandr Valialkin
73c9da16b8 app/vmselect/netstorage: reuse timeseriesWork objects in order to reduce memory allocations 2021-02-16 16:09:00 +02:00
Aliaksandr Valialkin
a3a09a3c6e app/vmselect/netstorage: use unsafe string as a key for a map when the map already contains the given key
This should prevent from a memory allocation and a string copy.
2021-02-16 15:46:31 +02:00
Aliaksandr Valialkin
46e98ed490 vendor: update github.com/VictoriaMetrics/metrics from v1.13.1 to v1.14.0
The new version switches from log-linear histograms to log-based histograms,
which provide up to 3.6 times better accuracy.
2021-02-15 15:11:15 +02:00
Aliaksandr Valialkin
c769f8321d deployment/docker: embed tzdata into prod Go app instead of installing it into base docker image
While this increases app size by 700Kb, this allows using -loggerTimezone in a scratch base image
See https://github.com/golang/go/issues/38017
2021-02-12 04:56:27 +02:00
Aliaksandr Valialkin
ff7850aec0 deployment/docker: use docker buildx for creating multiarch builds
See https://github.com/docker/buildx/
2021-02-12 04:35:35 +02:00
Aliaksandr Valialkin
ee4288987b app/vmselect/prometheus: treat match query arg in the same way as match[] query arg 2021-02-11 15:01:55 +02:00
Aliaksandr Valialkin
54d2f67924 app/vmselect/promql: properly make copies of EvalConfig 2021-02-11 12:41:37 +02:00
Aliaksandr Valialkin
5973fd4067 app/vmselect/promql: make a copy of EvalConfig when executing q1 and q2 in parallel for q1 binary_op q2
This should prevent from data races if the underlying functions modify EvalConfig contents.
2021-02-10 23:06:58 +02:00
Aliaksandr Valialkin
985c3e301d app/vmselect: parallelize q1 <binary_op> q2 queries by running q1 and q2 in parallel
This should reduce query execution times.
2021-02-10 23:00:56 +02:00
Aliaksandr Valialkin
8de3a329ff app/vmselect: deprecate -search.treatDotsAsIsInRegexps in favor to {__graphite__="foo.*.bar"} syntax 2021-02-03 20:36:42 +02:00
Aliaksandr Valialkin
8249f13104 app/vmselect,lib/storage: properly parse Graphite selectors with inner wildcards
Example: foo{bar{x,yz},a[b-c],*de}
2021-02-03 20:16:28 +02:00
Aliaksandr Valialkin
4b930b9ffe app/vmselect: add ability to set Graphite-compatible filter via {__graphite__="foo.*.bar"} syntax 2021-02-03 01:17:19 +02:00
Aliaksandr Valialkin
7d23f3ff3a app/vmselect: add ability to set an additional label filters via extra_label query arg 2021-02-01 17:42:35 +02:00
Nikolay
b8bc1c2e0f Graphite vmalert wip (#112)
* 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
2021-02-01 15:28:30 +02:00
Aliaksandr Valialkin
5328a102e0 app/vmselect: unconditionally deny partial responses from /api/v1/export*
It is expected that `/api/v1/export*` returns full data, so there is no sense in partial responses there.
2021-01-27 14:39:53 +02:00
Aliaksandr Valialkin
44c74f1e79 app/vmselect/promql: improve documentation for -search.maxPointsPertimeseries command-line flag
This should reduce incorrect usage and assumptions for this flag.

See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1020
2021-01-22 13:00:35 +02:00
Aliaksandr Valialkin
e55205220b app/vmselect: add -search.maxStepForPointsAdjustment command-line flag, which can be used for disabling adjustment for points returned from /api/v1/query_range handler if they have timestamps closer than -search.latencyOffset to the current time 2021-01-19 22:57:50 +02:00
Aliaksandr Valialkin
5856611291 app/vmselect/graphite: extract getCanonicalPath() function from loop body inside getCanonicalPaths() 2021-01-18 17:31:27 +02:00
Aliaksandr Valialkin
c5bdab5a4c app/vmselect/promql: add ability to pass multiple labels to sort_by_label and sort_by_label_desc functions
See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/992
2021-01-13 12:43:47 +02:00
Aliaksandr Valialkin
8cae98aa78 app/vmselect/promql: properly parse escaped multibyte utf8 code sequences in metric names and labels names
See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/990
2021-01-13 10:59:32 +02:00
Aliaksandr Valialkin
df6e399f73 app/vmselect/promql: add tfirst_over_time(m[d]) and tlast_over_time(m[d]) MetricsQL functions for returning timestamps for the first and the last samples in m over d 2021-01-12 16:12:47 +02:00
Aliaksandr Valialkin
47872ada7e app/vmselect/promql: do not ajdust offset value provided in the query
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
2020-12-27 14:10:15 +02:00
Aliaksandr Valialkin
5bbf200de2 app/vmselect: add per-tenant /api/v1/status/top_queries handler 2020-12-27 12:53:50 +02:00
Aliaksandr Valialkin
0e739efc88 app/vmselect/promql: simplify defer call for querystats.RegisterQuery 2020-12-27 12:07:56 +02:00
Aliaksandr Valialkin
44932098b5 app/vmselect/querystats: reduce the default number of last queries to track from 100K to 20K
This should reduce memory usage in constrained environments
2020-12-25 17:40:32 +02:00
Aliaksandr Valialkin
e6deb39064 app/vmselect: refactor /api/v1/stats/top_queries
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/907
2020-12-25 17:24:25 +02:00
Nikolay
76d092c091 Adds query stats handler (#945)
* 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
2020-12-25 17:24:24 +02:00
Nikolay
0b87f02602
fixes panic (#979)
* fixes panic
https://github.com/VictoriaMetrics/helm-charts/issues/89

* add fast-path

* Apply suggestions from code review

Co-authored-by: Aliaksandr Valialkin <valyala@gmail.com>
2020-12-24 09:01:40 +02:00
Aliaksandr Valialkin
262cf81757 app/vmselect: properly parse negative combined offsets such as -1h2m3s
Previously such offsets were parsed as `-1h + 2m + 3s`. Now they are parsed as `-(1h + 2m + 3s)`.
2020-12-19 01:25:03 +02:00
Aliaksandr Valialkin
8d1031c29a app/vmselect/promql: return expected increase() result for the first point on the graph with value not exceeding 100 2020-12-15 14:10:50 +02:00
Aliaksandr Valialkin
a2eb451de4 app/{vmagent,vminsert}: follow-up for ce8c2dd1f1: return /targets page in HTML when requested via web browser 2020-12-14 14:13:01 +02:00
Aliaksandr Valialkin
d6f9bf2d19 app/vmselect/graphite: properly handle wildcards and charsets inside curly braces
For example, `foo{bar*,[a-f]a*b}` should match `foobar`, `foobar123`, `foofab`, etc.

See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/952
2020-12-11 17:26:32 +02:00
Aliaksandr Valialkin
1a237c6903 all: properly handle CPU limits set on the host system/container
This can reduce memory usage on systems with enabled CPU limits.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/946
2020-12-08 21:07:03 +02:00
Aliaksandr Valialkin
bd8eef2528 app/vmselect/promql: do not reduce lookbehind window for any_rollup_func(m) to -search.maxStalenessInterval. It should equal to step value passed to /api/v1/query_range as most users expect 2020-12-08 15:17:05 +02:00
Aliaksandr Valialkin
9660774fd1 app/vmselect/graphite: remove duplicate name tag from /tags/autoComplete/tags handler
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/942
2020-12-07 01:10:02 +02:00
Aliaksandr Valialkin
d242c2f2bd app/vmselect/promql: add count_eq_over_time(m[d], N) and count_ne_over_time(m[d], N) for calculating the number of samples in m over d that are equal / not equal to N 2020-12-05 12:31:01 +02:00
Aliaksandr Valialkin
bdac2171f1 all: do not print usage info for all the flags when incorrect command-line flag is passed
This should improve usability for VictoriaMetrics apps that have big number of command-line flags,
i.e. all the apps.
2020-12-03 21:46:19 +02:00
Aliaksandr Valialkin
8cf76d8747 app/vmselect/promql: add label_uppercase(q, label1, ... labelN) and label_lowercase(q, label1, ... labelN) functions
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/936
2020-12-03 21:46:18 +02:00
Aliaksandr Valialkin
11bbb3552d app/vmselect/promql: make fmt 2020-12-02 21:34:15 +02:00
Aliaksandr Valialkin
9e98a8f3d3 app/vmselect/promql: return nan from minute(m) when m equals to nan
This aligns VictoriaMetrics behaviour with Prometheus behaviour.

The issue has been spotted in https://promlabs.com/promql-compliance-test-results/2020-12-01/victoriametrics/
2020-12-02 20:16:40 +02:00