Commit graph

125 commits

Author SHA1 Message Date
Aliaksandr Valialkin
4e71914e3c
app/vmselect/netstorage: do not intern string representation of MetricName for time series received from vmstorage
It has been appeared that this interning may lead to increased memory usage and increased CPU usage
when vmselect performs queries, which select big number of time series.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3692
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3863
2023-03-12 01:35:23 -08:00
Aliaksandr Valialkin
be8fba9b6a
app/vmselect/netstorage: tune the number of blocks per series which should be unpacked by a single goroutine instead of spinning up multiple goroutines
This reduces overhead on time series data unpacking for typical cases,
this reducing CPU usage at vmselect
2023-01-12 09:31:44 -08:00
Aliaksandr Valialkin
53d871d0b1
app/vmselect/netstorage: reduce tail latency during query processing
Previously the selected time series were split evenly among available CPU cores
for further processing - e.g unpacking the data and applying the given rollup
function to the unpacked data.
Some time series could be processed slower than others.
This could result in uneven work distribution among available CPU cores,
e.g. some CPU cores could complete their work sooner than others.
This could slow down query execution.

The new algorithm allows stealing time series to process from other CPU cores
when all the local work is done. This should reduce the maximum time
needed for query execution (aka tail latency).

The new algorithm should also scale better on systems with many CPU cores,
since every CPU processes locally assigned time series without inter-CPU communications.

The inter-CPU communications are used only when all the local work is finished
and the pending work from other CPUs needs to be stealed.
2023-01-10 13:43:14 -08:00
Aliaksandr Valialkin
e640ff72f1
app/vmselect/netstorage: reduce memory allocations when unpacking time series
Unpack time series with less than 400K samples in the currently running goroutine.
Previously a new goroutine was being started for unpacking the samples.
This was requiring additional memory allocations.
2023-01-09 23:18:17 -08:00
Aliaksandr Valialkin
df2a494a7c
app/vmselect/netstorage: pre-allocate 4 block references per each time series during querying
Usually the number of blocks returned per each time series during queries is around 4.
So it is a good idea to pre-allocate 4 block references per time series
in order to reduce the number of memory allocations.
2023-01-09 22:03:23 -08:00
Aliaksandr Valialkin
c5e0f527bc
app/vmselect/netstorage: cache canonical MetricName for time series returned from the storage
This reduces memory allocations for repeated queries, which return (almost) the same set of time series.
2023-01-09 21:53:10 -08:00
Aliaksandr Valialkin
7afcca0c51
all: use metricsql.CompileRegexp instead of regexp.Compile for compiling regexps used in graphite queries
This should speed up repeated queries, since metricsql.CompileRegexp returns regexps from the cache
on subsequent calls for the same input regexp.
2023-01-09 21:43:08 -08:00
Aliaksandr Valialkin
c38a10e143
app/vmselect/netstorage: eliminate memory allocation for sortBlocksHeap arg when calling mergeSortBlocks() 2023-01-09 21:08:51 -08:00
Aliaksandr Valialkin
1f9d605988
app/vmselect/netstorage: consistently select the sample with the biggest value out of samples with identical timestamps
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3333

This fix is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/3620 ,
but doesn't slow down the common case with merging replicated data blocks so significantly.

Benchmark results:

Before the change:

BenchmarkMergeSortBlocks/replicationFactor-1-4         	   13968	     85643 ns/op	 956.53 MB/s	    1700 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-2-4         	   10806	    109171 ns/op	1500.77 MB/s	    2191 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-3-4         	    8887	    130623 ns/op	1881.45 MB/s	    2660 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-4-4         	    7440	    157348 ns/op	2082.52 MB/s	    3174 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-5-4         	    6534	    184473 ns/op	2220.38 MB/s	    3612 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/overlapped-blocks-bestcase-4  	   13419	     85205 ns/op	 961.44 MB/s	    2213 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/overlapped-blocks-worstcase-4 	     579	   1894900 ns/op	  43.23 MB/s	   46760 B/op	       1 allocs/op

After the change:

BenchmarkMergeSortBlocks/replicationFactor-1-4         	   13832	     85298 ns/op	 960.40 MB/s	    1716 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-2-4         	    8833	    134222 ns/op	1220.66 MB/s	    2675 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-3-4         	    6487	    184830 ns/op	1329.65 MB/s	    3636 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-4-4         	    4977	    236318 ns/op	1386.61 MB/s	    4733 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/replicationFactor-5-4         	    4088	    296734 ns/op	1380.36 MB/s	    5761 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/overlapped-blocks-bestcase-4  	   14083	     84067 ns/op	 974.47 MB/s	    2110 B/op	       1 allocs/op
BenchmarkMergeSortBlocks/overlapped-blocks-worstcase-4 	     536	   2043534 ns/op	  40.09 MB/s	   50511 B/op	       1 allocs/op
2023-01-09 13:01:48 -08:00
Roman Khavronenko
7c0ae3a86a
lib/storage: keep sample with the biggest value on timestamp conflict (#3421)
The change leaves raw sample with the biggest value for identical
timestamps per each `-dedup.minScrapeInterval` discrete interval
when the deduplication is enabled.

```
benchstat old.txt new.txt
name                                         old time/op    new time/op    delta
DeduplicateSamples/minScrapeInterval=1s-10      817ns ± 2%     832ns ± 3%      ~     (p=0.052 n=10+10)
DeduplicateSamples/minScrapeInterval=2s-10     1.56µs ± 1%    2.12µs ± 0%   +35.19%  (p=0.000 n=9+7)
DeduplicateSamples/minScrapeInterval=5s-10     1.32µs ± 3%    1.65µs ± 2%   +25.57%  (p=0.000 n=10+10)
DeduplicateSamples/minScrapeInterval=10s-10    1.13µs ± 2%    1.50µs ± 1%   +32.85%  (p=0.000 n=10+10)

name                                         old speed      new speed      delta
DeduplicateSamples/minScrapeInterval=1s-10   10.0GB/s ± 2%   9.9GB/s ± 3%      ~     (p=0.052 n=10+10)
DeduplicateSamples/minScrapeInterval=2s-10   5.24GB/s ± 1%  3.87GB/s ± 0%   -26.03%  (p=0.000 n=9+7)
DeduplicateSamples/minScrapeInterval=5s-10   6.22GB/s ± 3%  4.96GB/s ± 2%   -20.37%  (p=0.000 n=10+10)
DeduplicateSamples/minScrapeInterval=10s-10  7.28GB/s ± 2%  5.48GB/s ± 1%   -24.74%  (p=0.000 n=10+10)
```

https://github.com/VictoriaMetrics/VictoriaMetrics/issues/3333
Signed-off-by: hagen1778 <roman@victoriametrics.com>

Signed-off-by: hagen1778 <roman@victoriametrics.com>
2022-12-08 18:06:11 -08:00
Aliaksandr Valialkin
cae0f37edd
app/vmselect/netstorage: remove superflouos map lookup at ProcessSearchQuery
This should reduce CPU usage a bit during querying
2022-11-18 13:40:04 +02:00
Aliaksandr Valialkin
c53b7e66ef
app/vmselect: improve performance scalability on multi-CPU systems for /api/v1/export/... endpoints 2022-10-01 22:05:43 +03:00
Aliaksandr Valialkin
343241680b
all: remove the remaining bits of io/ioutil
The io/ioutil package is deprecated since Go1.16 - see https://tip.golang.org/doc/go1.16#ioutil

VictoriaMetrics requires at least Go1.18, so it is time to remove the io/ioutil from source code

This is a follow-up for 02ca2342ab
2022-08-22 00:20:58 +03:00
Aliaksandr Valialkin
7478d423c5
app/vmselect/netstorage: cleanup after 92630c1ab4
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2896
2022-08-04 18:28:11 +03:00
Aliaksandr Valialkin
c2bd75926b
app/vmselect/netstorage: initialize tsw.rowsProcessed before calling tsw.f, since tsw.f can modify r.Timestamps and r.Values lengths 2022-07-30 00:39:36 +03:00
Aliaksandr Valialkin
19a0b4679a
app/vmselect/netstorage: re-use random generator used for series shuffle in Result.RunParallel
This should reduce CPU usage needed for rand.Rand initialization
2022-07-30 00:30:37 +03:00
Aliaksandr Valialkin
92630c1ab4
app/vmselect/netstorage: improve the speed of queries over big number of time series on multi-CPU system
Reduce inter-CPU communications when processing the query over big number of time series.
This should improve performance for queries over big number of time series
on systems with many CPU cores.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2896

Based on b596ac3745
Thanks to @zqyzyq for the idea.
2022-07-25 09:18:44 +03:00
Aliaksandr Valialkin
159c2e15e3
app/vmselect/netstorage: optimize mergeSortBlocks() for the worst case when blocks contain interleaved samples 2022-07-12 12:31:38 +03:00
Aliaksandr Valialkin
8429d4af5a
app/vmselect/netstorage: add mergeSortBlocks benchmark for the worstcase 2022-07-12 12:31:36 +03:00
Aliaksandr Valialkin
cd09f583fe
app/vmselect/netstorage: add benchmarks for mergeSortBlocks
This is a follow-up for 743ff84863
2022-07-11 12:54:48 +03:00
Aliaksandr Valialkin
743ff84863
app/vmselect/netstorage: optimize mergeSortBlocks function
- Use binary search instead of linear scan when locating the run of smallest timestamps
  in blocks with intersected time ranges. This should improve performance
  when merging blocks with big number of samples

- Skip samples with duplicate timestamps. This should increase query performance
  in cluster version of VictoriaMetrics with the enabled replication.
2022-07-09 00:34:42 +03:00
Aliaksandr Valialkin
77cbbacfdb
lib/vmselectapi: pass storage.SearchQuery to API calls instead of []*storage.TagFilters + storage.TimeRange + maxMetrics
This reduces the number of args to vmselectapi calls
2022-07-06 12:37:54 +03:00
Aliaksandr Valialkin
f435924ab3
lib/vmselectapi: pass maxSuffixes arg to tagValueSuffixes RPC call 2022-07-06 12:37:54 +03:00
Aliaksandr Valialkin
e1b8059086
lib/vmselectapi: rename deleteMetrics to more correct deleteSeries 2022-07-06 12:37:54 +03:00
Aliaksandr Valialkin
a60e03b3a7
lib/vmselectapi: use string type for tagKey and tagValuePrefix args at TagValueSuffixes()
This improves the API consistency
2022-07-06 12:37:53 +03:00
Aliaksandr Valialkin
a14188dd8e
app/vmselect: expose additional histograms at /metrics page, which may help get more insights for the query workload
This commit is based on https://github.com/VictoriaMetrics/VictoriaMetrics/pull/2792
2022-06-28 20:18:13 +03:00
Aliaksandr Valialkin
a350d1e81c
lib/storage: return marshaled metric names from SearchMetricNames
Previously SearchMetricNames was returning unmarshaled metric names.
This wasn't great for vmstorage, which should spend additional CPU time
for marshaling the metric names before sending them to vmselect.

While at it, remove possible duplicate metric names, which could occur when
multiple samples for new time series are ingested via concurrent requests.

Also sort the metric names before returning them to the client.
This simplifies debugging of the returned metric names across repeated requests to /api/v1/series
2022-06-28 18:17:15 +03:00
Aliaksandr Valialkin
e578549b8a
app/vmselect: optimize /api/v1/series a bit for time ranges smaller than one day 2022-06-28 13:02:47 +03:00
Aliaksandr Valialkin
ba514284f1
lib/storage: add querytracer to more contexts
querytracer has been added to the following storage.Storage methods:
- RegisterMetricNames
- DeleteMetrics
- SearchTagValueSuffixes
- SearchGraphitePaths
2022-06-27 13:45:51 +03:00
Aliaksandr Valialkin
9a314106ca
app/vmselect/netstorage: remove Get prefix from netstorage functions
This makes these function names more consistent with the server side
2022-06-27 00:45:05 +03:00
Aliaksandr Valialkin
ec7963208d
app/vmselect: accept focusLabel query arg at /api/v1/status/tsdb
This allows filling the seriesCountByFocusLabelValue list in the /api/v1/status/tsdb response
with label values for the specified focusLabel, which contain the highest number of time series.

TODO: add this to Cardinality explorer at VMUI - https://docs.victoriametrics.com/#cardinality-explorer
2022-06-14 18:36:54 +03:00
Aliaksandr Valialkin
374beb350e
app/vmselect: optimize /api/v1/labels and /api/v1/label/.../values handlers when match[] query arg is passed to them 2022-06-12 04:32:13 +03:00
Aliaksandr Valialkin
89b778902b
app/vmselect: add optional limit query arg to /api/v1/labels and /api/v1/label_values endpoints
This arg allows limiting the number of sample values returned from these APIs
2022-06-10 09:50:33 +03:00
Aliaksandr Valialkin
2bcb960f17
all: improve query tracing coverage for indexdb search
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1403
2022-06-09 20:07:07 +03:00
Aliaksandr Valialkin
12ac255dae
lib/querytracer: make it easier to use by passing trace context message to New and NewChild
The context message can be extended by calling Donef.
If there is no need to extend the message, then just call Done.
2022-06-08 21:06:52 +03:00
Aliaksandr Valialkin
41958ed5dd
all: add initial support for query tracing
See https://docs.victoriametrics.com/Single-server-VictoriaMetrics.html#query-tracing

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1403
2022-06-01 02:29:23 +03:00
Aliaksandr Valialkin
6e364e19ef
app/vmselect: add fine-grained limits for the number of returned/scanned time series for various APIs 2022-03-26 11:29:49 +02:00
Aliaksandr Valialkin
4bdd10ab90
lib/bytesutil: split Resize* funcs to MayOverallocate and NoOverallocate for more fine-grained control over memory allocations
Follow-up for f4989edd96

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2007
2022-02-01 00:18:42 +02:00
Aliaksandr Valialkin
f4989edd96
lib/bytesutil: split Resize() into ResizeNoCopy() and ResizeWithCopy() functions
Previously bytesutil.Resize() was copying the original byte slice contents to a newly allocated slice.
This wasted CPU cycles and memory bandwidth in some places, where the original slice contents wasn't needed
after slize resizing. Switch such places to bytesutil.ResizeNoCopy().

Rename the original bytesutil.Resize() function to bytesutil.ResizeWithCopy() for the sake of improved readability.

Additionally, allocate new slice with `make()` instead of `append()`. This guarantees that the capacity of the allocated slice
exactly matches the requested size. The `append()` could return a slice with bigger capacity as an optimization for further `append()` calls.
This could result in excess memory usage when the returned byte slice was cached (for instance, in lib/blockcache).

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/2007
2022-01-25 15:24:44 +02:00
Aliaksandr Valialkin
92070cbb67
lib/storage: return dedup interval in milliseconds from GetDedupInterval()
This removes duplicate .Milliseconds() calls after GetDedupInterval() calls.
2021-12-15 13:26:38 +02:00
Aliaksandr Valialkin
1d20a19c7d
lib/storage: explicitly pass dedupInterval to DeduplicateSamples() and deduplicateSamplesDuringMerge()
This improves the code readability and debuggability, since the output of these functions
stops depending on global state.
2021-12-14 20:49:12 +02:00
Aliaksandr Valialkin
8ed95e82c6
app/vmselect/promql: follow-up after 57b3320478 2021-09-24 01:24:18 +03:00
Aliaksandr Valialkin
f4dead529f lib/storage: properly search series by multiple tag filters matching empty labels such as foo{bar=~"baz|",x=~"y|"}
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1601
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/395
2021-09-09 21:09:21 +03:00
Aliaksandr Valialkin
a1911e1330 app/vmselect/netstorage: unpack time series data in mostly local big chunks
This should improve performance on multi-CPU systems for queries selecting time series with big number of raw samples
2021-07-30 12:03:17 +03:00
Aliaksandr Valialkin
74ffaa45d9 app/vmselect/netstorage: do not query Go maps with unsafe string keys, since this breaks in Go 1.17 2021-07-30 09:57:53 +03:00
Aliaksandr Valialkin
8ee8660ac4 app/vmselect: follow-up for 626073bca8
* Rename -search.maxMetricsPointSearch to -search.maxSamplesPerQuery, so it is more consistent with the existing -search.maxSamplesPerSeries
* Move the -search.maxSamplesPerQuery from vmstorage to vmselect, so it could effectively limit the number of raw samples obtained from all the vmstorage nodes
* Document the -search.maxSamplesPerQuery in docs/CHANGELOG.md
2021-07-28 18:00:23 +03:00
Aliaksandr Valialkin
95aff47330 app/vmselect: prevent from possible deadlock when f callback blocks inside RunParallel 2021-07-26 15:47:30 +03:00
Aliaksandr Valialkin
ed10141ff8 app/vmselect/netstorage: use more scalable algorithm for ditributing the work among among multiple channels on systems with big number of CPU cores 2021-07-16 00:35:23 +03:00
Aliaksandr Valialkin
f4e81aef7e app/vmselect/netstorage: add -search.maxSamplesPerSeries command-line option for limiting the number of samples a query can process per each series
This should prevent from out of memory crashes like in https://github.com/VictoriaMetrics/VictoriaMetrics/issues/1067
2021-07-15 16:03:28 +03:00
Aliaksandr Valialkin
e6ef97a5ee app/vmselect/netstorage: improve scalability of series unpacking on multi-CPU systems 2021-07-15 15:41:58 +03:00