Commit graph

1298 commits

Author SHA1 Message Date
Aliaksandr Valialkin
33d9d63393 lib/storage: speed up time series search for queries with multiple filters
Use optimized specialized binary search for uint64 metricIDs instead of generic sort.Search.
2019-11-21 18:43:40 +02:00
Aliaksandr Valialkin
a02a57fbe9 lib/storage: verify the number of returned metricIDs in BenchmarkHeadPostingForMatchers 2019-11-20 15:40:03 +02:00
Aliaksandr Valialkin
3d1f4408cf lib/decimal: increase decimal->float speed conversion for integer numbers 2019-11-20 14:09:10 +02:00
Aliaksandr Valialkin
f1f2eff08f lib/decimal: reduce rounding error when converting from decimal to float with negative exponent
While at it, slightly increase the conversion performance by moving fast path to the top of the loop.
2019-11-19 23:34:41 +02:00
Aliaksandr Valialkin
17eca31989 lib/backup: retrieve only the required metadata when reading GCS objects 2019-11-19 21:30:51 +02:00
Aliaksandr Valialkin
216a260ced app/{vmbackup,vmrestore}: add -maxBytesPerSecond command-line flag for limiting the used network bandwidth during backup / restore 2019-11-19 20:32:43 +02:00
Aliaksandr Valialkin
9d1ee1e2ae lib/backup: prevent from restoring to directory which is in use by VictoriaMetrics during the restore 2019-11-19 18:35:59 +02:00
Aliaksandr Valialkin
6ca4b94511 lib/storage: increase the number of created time series in BenchmarkHeadPostingForMatchers in order to be on par with Promethues
The previous commit was accidentally creating 10x smaller number of time series than Prometheus
and this led to invalid benchmark results.

The updated benchmark results:

benchmark                                                          old ns/op      new ns/op     delta
BenchmarkHeadPostingForMatchers/n="1"                              272756688      6194893       -97.73%
BenchmarkHeadPostingForMatchers/n="1",j="foo"                      138132923      10781372      -92.19%
BenchmarkHeadPostingForMatchers/j="foo",n="1"                      134723762      10632834      -92.11%
BenchmarkHeadPostingForMatchers/n="1",j!="foo"                     195823953      10679975      -94.55%
BenchmarkHeadPostingForMatchers/i=~".*"                            7962582919     100118510     -98.74%
BenchmarkHeadPostingForMatchers/i=~".+"                            7589543864     154955671     -97.96%
BenchmarkHeadPostingForMatchers/i=~""                              1142371741     258003769     -77.42%
BenchmarkHeadPostingForMatchers/i!=""                              9964150263     159783895     -98.40%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",j="foo"              216995884      10937895      -94.96%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",i!="2",j="foo"       202541348      10990027      -94.57%
BenchmarkHeadPostingForMatchers/n="1",i!=""                        486285711      87004349      -82.11%
BenchmarkHeadPostingForMatchers/n="1",i!="",j="foo"                350776931      53342793      -84.79%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",j="foo"              380888565      54256156      -85.76%
BenchmarkHeadPostingForMatchers/n="1",i=~"1.+",j="foo"             89500296       21823279      -75.62%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!="2",j="foo"       379529654      46671359      -87.70%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!~"2.*",j="foo"     424563825      53915842      -87.30%

VictoriaMetrics uses 1GB of RAM during the benchmark (vs 3.5GB of RAM for Prometheus)
2019-11-18 19:48:27 +02:00
Aliaksandr Valialkin
6f61fd367a lib/storage: add BenchmarkHeadPostingForMatchers similar to the benchmark from Prometheus
See the corresponding benchmark in Prometheus - 23c0299d85/tsdb/head_bench_test.go (L52)

The benchmark allows performing apples-to-apples comparison of time series search
in Prometheus and VictoriaMetrics. The following article - https://www.robustperception.io/evaluating-performance-and-correctness -
contains incorrect numbers for VictoriaMetrics, since there wasn't this benchmark yet. Fix it.

Benchmarks can be repeated with the following commands from Prometheus and VictoriaMetrics source code roots:

- Prometheus: GOMAXPROCS=1 go test ./tsdb/ -run=111 -bench=BenchmarkHeadPostingForMatchers
- VictoriaMetrics: GOMAXPROCS=1 go test ./lib/storage/ -run=111 -bench=BenchmarkHeadPostingForMatchers

Benchmark results:
benchmark                                                          old ns/op      new ns/op     delta
BenchmarkHeadPostingForMatchers/n="1"                              272756688      364977        -99.87%
BenchmarkHeadPostingForMatchers/n="1",j="foo"                      138132923      1181636       -99.14%
BenchmarkHeadPostingForMatchers/j="foo",n="1"                      134723762      1141578       -99.15%
BenchmarkHeadPostingForMatchers/n="1",j!="foo"                     195823953      1148056       -99.41%
BenchmarkHeadPostingForMatchers/i=~".*"                            7962582919     8716755       -99.89%
BenchmarkHeadPostingForMatchers/i=~".+"                            7589543864     12096587      -99.84%
BenchmarkHeadPostingForMatchers/i=~""                              1142371741     16164560      -98.59%
BenchmarkHeadPostingForMatchers/i!=""                              9964150263     12230021      -99.88%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",j="foo"              216995884      1173476       -99.46%
BenchmarkHeadPostingForMatchers/n="1",i=~".*",i!="2",j="foo"       202541348      1299743       -99.36%
BenchmarkHeadPostingForMatchers/n="1",i!=""                        486285711      11555193      -97.62%
BenchmarkHeadPostingForMatchers/n="1",i!="",j="foo"                350776931      5607506       -98.40%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",j="foo"              380888565      6380335       -98.32%
BenchmarkHeadPostingForMatchers/n="1",i=~"1.+",j="foo"             89500296       2078970       -97.68%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!="2",j="foo"       379529654      6561368       -98.27%
BenchmarkHeadPostingForMatchers/n="1",i=~".+",i!~"2.*",j="foo"     424563825      6757132       -98.41%

The first column (old) is for Prometheus, the second column (new) is for VictoriaMetrics.

Prometheus was using 3.5GB of RAM during the benchmark, while VictoriaMetrics was using 400MB of RAM.
2019-11-18 18:47:02 +02:00
Aliaksandr Valialkin
d297b65089 lib/storage: add vm_cache_size_bytes{type="storage/hour_metric_ids"} metric 2019-11-13 20:26:05 +02:00
Aliaksandr Valialkin
494ad0fdb3 lib/storage: remove inmemory index for recent hour, since it uses too much memory
Production workload shows that the index requires ~4Kb of RAM per active time series.
This is too much for high number of active time series, so let's delete this index.

Now the queries should fall back to the index for the current day instead of the index
for the recent hour. The query performance for the current day index should be good enough
given the 100M rows/sec scan speed per CPU core.
2019-11-13 18:08:58 +02:00
Aliaksandr Valialkin
633dd81bb5 lib/storage: add -disableRecentHourIndex flag for disabling inmemory index for recent hour
This may be useful for saving RAM on high number of time series aka high cardinality
2019-11-13 15:10:12 +02:00
Aliaksandr Valialkin
f1620ba7c0 lib/storage: fix inmemory inverted index issues found in v1.29
Issues fixed:
- Slow startup times. Now the index is loaded from cache during start.
- High memory usage related to superflouos index copies every 10 seconds.
2019-11-13 13:35:38 +02:00
Aliaksandr Valialkin
87b39222be Revert "lib/fs: do not postpone directory removal on NFS error"
This reverts commit 21aeb02b46649ac9906cb37733f7b155a77a0db9.
2019-11-12 16:29:50 +02:00
Mike Poindexter
955a592106 Add test for invalid caching of tsids (#232)
* Add test for invalid caching of tsids

* Clean up error handling
2019-11-12 15:52:46 +02:00
Oleg Kovalov
74ba42d111 fix misspelled words (#229) 2019-11-12 00:18:24 +02:00
Aliaksandr Valialkin
c48e39eea9 lib/storage: add tests for dateMetricIDCache 2019-11-11 13:21:05 +02:00
Aliaksandr Valialkin
6bdde0d6d4 lib/storage: eliminate data race when updating lastSyncTime in dateMetricIDCache.Has 2019-11-10 22:04:23 +02:00
Aliaksandr Valialkin
5f52eb7653 lib/fs: do not postpone directory removal on NFS error
Continue trying to remove NFS directory on temporary errors for up to a minute.

The previous async removal process breaks in the following case during VictoriaMetrics start

- VictoriaMetrics opens index, finds incomplete merge transactions and starts replaying them.
- The transaction instructs removing old directories for parts, which were already merged into bigger part.
- VictoriaMetrics removes these directories, but their removal is delayed due to NFS errors.
- VictoriaMetrics scans partition directory after all the incomplete merge transactions are finished
  and finds directories, which should be removed, but weren't still removed due to NFS errors.
- VictoriaMetrics panics when it finds unexpected empty directory.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/162
2019-11-10 13:27:16 +02:00
Aliaksandr Valialkin
9ea2bd822e lib/storage: implement per-day inverted index 2019-11-10 00:20:32 +02:00
Aliaksandr Valialkin
dea2f3efed lib/storage: use specialized cache for (date, metricID) entries
This improves ingestion performance.
2019-11-09 23:09:18 +02:00
Aliaksandr Valialkin
9a43902bd8 lib/storage: remove unused code from getMetricIDsForTimeRange: it is expected that time range is always non-zero 2019-11-09 19:03:51 +02:00
Aliaksandr Valialkin
c16e17dede lib/storage: properly set time range when deleting time series 2019-11-09 18:50:02 +02:00
Aliaksandr Valialkin
8126007c15 lib/storage: obtain all the time series ids from (tag->metricIDs) rows instead of (metricID->TSID) rows, since this much faster 2019-11-09 18:04:26 +02:00
Aliaksandr Valialkin
50773348d3 lib/storage: small code prettifying 2019-11-09 14:01:24 +02:00
Aliaksandr Valialkin
44fa8226df lib/uint64set: remove superflouos check for item existence before deleting it in Set.Subtract 2019-11-09 14:01:24 +02:00
Aliaksandr Valialkin
0bc54c23ce lib/storage: inmemoryInvertedIndex prettifying 2019-11-09 14:01:24 +02:00
Aliaksandr Valialkin
46e67bb78c lib/storage: export vm_new_timeseries_created_total metric for determining time series churn rate 2019-11-08 19:58:21 +02:00
Aliaksandr Valialkin
0063c857f5 lib/storage: add inmemory inverted index for the last hour
It should improve performance for `last N hours` dashboards with update intervals smaller than 1 hour.
2019-11-08 19:37:46 +02:00
Aliaksandr Valialkin
7d7fbf890e app/{vmbackup,vmrestore}: add vmbackup and vmrestore tools for creating backups on s3 or gcs from instant snapshots
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/203
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/38
2019-11-07 21:26:43 +02:00
Aliaksandr Valialkin
89c03a5464 lib/storage: populate partition names from both small and big directories
Certain partition directories may be missing after restoring from backups
if they had no data. Re-create such directories on start.
2019-11-06 19:50:21 +02:00
Aliaksandr Valialkin
1c777e0245 lib/storage: substitute error message about unsorted items in the index block after metricIDs merge with counter
The origin of the error has been detected and documented in the code,
so it is enough to export a counter for such errors at `vm_index_blocks_with_metric_ids_incorrect_order_total`,
so it could be monitored and alerted on high error rates.

Export also the counter for processed index blocks with metricIDs - `vm_index_blocks_with_metric_ids_processed_total`,
so its' rate could be compared to `rate(vm_index_blocks_with_metric_ids_incorrect_order_total)`.
2019-11-06 14:32:41 +02:00
Aliaksandr Valialkin
c567a4353a lib/storage: take into account the requested time range when caching TSIDs for the given tag filters 2019-11-06 14:32:41 +02:00
Aliaksandr Valialkin
c6564c5d26 lib/storage: dump incorrectly sorted items on a single line; this should simplify error reporting 2019-11-05 18:41:50 +02:00
Aliaksandr Valialkin
a10c4cad85 lib/storage: return back finalPartsToMerge from 2 to 3 in order to prevent from excessive merges in old partitions 2019-11-05 17:28:57 +02:00
Aliaksandr Valialkin
e5b1fa0c38 lib/storage: separate the max inverted index scan loops per metric into fast and slow loops
Slow loops could require seeks and expensive regexp matching, while fast loops just scans
all the metricIDs for the given `tag=value` prefix. So these operations must have separate
max loops multiplier.
2019-11-05 17:28:57 +02:00
Aliaksandr Valialkin
f93c4f2493 lib/storage: skip repeated useless work when intersection of metricIDs with the given filter is too expensive
This should improve performance for query filters over big number of time series.
2019-11-05 14:35:55 +02:00
Aliaksandr Valialkin
f48e97263c lib/storage: reduce the maximum inverted index scans before giving up to label filters matching by metric name
The new value reduces the amount of wasted work during index scans over big number of time series.
2019-11-05 14:35:53 +02:00
Aliaksandr Valialkin
d2f688c550 lib/storage: try potentially faster tag filters at first, then apply slower tag filters
The fastest tag filters are non-negative non-regexp, since they are the most specific.
The slowest tag filters are negative regexp, since they require scanning
all the entries for the given label.
2019-11-05 14:35:48 +02:00
Aliaksandr Valialkin
2a38d30f93 lib/storage: pass pointer to MetricName in Fatalf, so it is properly detected as an interface with String() method
This fixes lint errors
2019-11-04 01:06:45 +02:00
Artem Navoiev
e05500cbd4 add unittests for bytesutil and storage (#221) 2019-11-04 00:57:24 +02:00
Aliaksandr Valialkin
f5fbc3ffd7 lib/{storage,uint64set}: add Set.Union() function and use it 2019-11-04 00:48:32 +02:00
Aliaksandr Valialkin
23e078261e lib/storage: tune the returned value from adjustMaxMetricsAdaptive 2019-11-04 00:45:28 +02:00
Aliaksandr Valialkin
386c349c8c lib/storage: remove interface conversion in hot path during block merging
This should improve merge speed a bit for parts with big number of small blocks.
2019-11-03 12:33:48 +02:00
Aliaksandr Valialkin
26ffc77622 lib/{storage,mergeset}: create missing partition directories after restoring from backups
Backup tools could skip empty directories. So re-create such directories on the first run.
2019-11-02 02:27:19 +02:00
Aliaksandr Valialkin
5d439cc6f2 lib/{decimal,encoding}: optimize float64<->decimal conversion for arrays with zeros or ones
Time series with only zeros or ones frequently occur in monitoring, so it is worth optimizing their handling.
2019-11-01 16:46:08 +02:00
Aliaksandr Valialkin
1037053fed lib/{encoding,decimal}: add benchmarks for blocks containing zeros or ones
Time series with such values are quite common in monitoring space,
so it would be great to have benchmarks for them.
2019-11-01 16:46:08 +02:00
Aliaksandr Valialkin
44fab198e2 lib/uint64set: return an emptry set instead of nil set from Set.Clone, since the caller may add data to the cloned set
This fixes the following panic in v1.28.1:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x10 pc=0x783a7e]

goroutine 1155 [running]:
github.com/VictoriaMetrics/VictoriaMetrics/lib/uint64set.(*Set).Add(0x0, 0x15b3bfb41e8b71ec)
  github.com/VictoriaMetrics/VictoriaMetrics@/lib/uint64set/uint64set.go:57 +0x2e
github.com/VictoriaMetrics/VictoriaMetrics/lib/storage.(*indexSearch).getMetricIDsForRecentHours(0xc5bdc0dd40, 0x16e273f6b50, 0x16e2745d3f0, 0x5b8d95, 0x10, 0x4a2f51, 0xaa01000000000000)
  github.com/VictoriaMetrics/VictoriaMetrics@/lib/storage/index_db.go:1951 +0x260
github.com/VictoriaMetrics/VictoriaMetrics/lib/storage.(*indexSearch).getMetricIDsForTimeRange(0xc5bdc0dd40, 0x16e273f6b50, 0x16e2745d3f0, 0x5b8d95, 0x10, 0xb296c0, 0xc00009cd80, 0x9bc640)
2019-11-01 16:12:21 +02:00
Aliaksandr Valialkin
6ab9c98a1e app/vmstorage: add -bigMergeConcurrency and -smallMergeConcurrency flags for tuning the maximum number of CPU cores used during merges 2019-10-31 16:17:29 +02:00
Aliaksandr Valialkin
6a22727676 lib/storage: optimize getMetricIDsForRecentHours for per-tenant lookups 2019-10-31 15:51:09 +02:00
Aliaksandr Valialkin
ca480915ca lib/storage: small cleanup in Storage.add 2019-10-31 14:30:22 +02:00
Aliaksandr Valialkin
22030b558f lib/decimal: speed up FromFloat for common case with integers 2019-10-31 13:25:09 +02:00
Aliaksandr Valialkin
6510258a80 lib/decimal: increase float64->decimal conversion precision a bit
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/213
2019-10-30 02:04:28 +02:00
Aliaksandr Valialkin
a27e034a40 lib/storage: get parts to merge after applying the limit on the number of concurrent merges
This should reduce write amplification under high ingestion rate.
2019-10-30 00:09:44 +02:00
Aliaksandr Valialkin
5d2276dbf7 lib/{mergeset,storage}: limit the maximum number of concurrent merges; leave smaller number of parts during final merge 2019-10-29 12:45:37 +02:00
Aliaksandr Valialkin
f581b2736a lib/fs: typo fix in comment to WriteFileAtomically 2019-10-29 11:31:34 +02:00
Aliaksandr Valialkin
eb513e7ba3 lib/decimal: increase float->decimal conversion precision for big numbers
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/213
2019-10-28 13:23:54 +02:00
hanzai
52778da1f3 warns during rows addition (#214) 2019-10-20 23:38:51 +03:00
Aliaksandr Valialkin
78fc35c9b1 all: make fmt 2019-10-17 20:05:12 +03:00
Aliaksandr Valialkin
5b01b7fb01 all: add support for GOARCH=386 and fix all the issues related to 32-bit architectures such as GOARCH=arm
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/212
2019-10-17 18:27:49 +03:00
Aliaksandr Valialkin
12c8afc3f2 lib/memory: properly handle int overflow in sysTotalMemory
This should fix builds on 32-bit architectures such as arm.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/212
2019-10-17 18:27:49 +03:00
Aliaksandr Valialkin
dc2f822577 lib/prompb: removed outdated README.md 2019-10-14 22:16:36 +03:00
Aliaksandr Valialkin
661b8ede5b lib/storage: harden the check that the original items are sorted after mergeTagToMetricIDsRows fails to preserve sort order 2019-10-09 12:13:43 +03:00
Aliaksandr Valialkin
7f4a04ee6a lib/storage: typo fix in comment to maxRowsPerSmallPart. 2019-10-08 18:51:56 +03:00
Aliaksandr Valialkin
7e410e1412 lib/storage: add tests for mergeTagToMetricIDsRows and return the original items if the function breaks items` ordering.
This should save from data corruption issues revealed in the previous releases up to v1.28.0-beta5.
2019-10-08 16:35:39 +03:00
Aliaksandr Valialkin
9fce611fbb lib/mergeset: reduce the maximum number of cached blocks, since there are reports on OOMs due to too big caches
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/189
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/195
2019-09-30 12:27:30 +03:00
Aliaksandr Valialkin
d0b4590099 lib/storage: optimize TSID comparison 2019-09-26 14:20:02 +03:00
Aliaksandr Valialkin
95e3d648cb lib/storage: verify whether items are sorted in the end of call to mergeTagToMetricIDsRows
This should prevent from inverted index corruption if bug in mergeTagToMetricIDsRows is discovered.
2019-09-26 13:13:58 +03:00
Aliaksandr Valialkin
2b8358726f lib/storage: properly match labels against regexp with (?i) flag
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/161
2019-09-26 11:03:26 +03:00
Aliaksandr Valialkin
4e3871ac1e lib/storage: add missing break in removeDuplicateMetricIDs 2019-09-25 18:23:13 +03:00
Aliaksandr Valialkin
4468f9f966 lib/storage: remove duplicate MetricIDs in tag->metricIDs items before writing them into inverted index 2019-09-25 17:57:36 +03:00
Aliaksandr Valialkin
adc18c3ee6 lib/{mergeset,storage}: do not cache inverted index blocks containing tag->metricIDs items
This should reduce the amounts of used RAM during queries with filters over big number of time series.
2019-09-25 13:48:24 +03:00
Aliaksandr Valialkin
c64fb91a43 lib/uint64set: optimize Set.AppendTo 2019-09-25 00:34:31 +03:00
Aliaksandr Valialkin
de0e4eee2c lib/storage: create and use lib/uint64set instead of map[uint64]struct{}
This should improve inverted index search performance for filters matching big number of time series,
since `lib/uint64set.Set` is faster than `map[uint64]struct{}` for both `Add` and `Has` calls.
See the corresponding benchmarks in `lib/uint64set`.
2019-09-24 21:18:04 +03:00
Aliaksandr Valialkin
2212d0e421 lib/storage: typo fix: return dstData instead of data from mergeTagToMetricIDsRows 2019-09-24 19:32:58 +03:00
Aliaksandr Valialkin
9307de1b92 lib/storage: limit the number of metricIDs in tag->metricIDs row
This reduces the overhead on index and metaindex in lib/mergeset
2019-09-24 00:50:47 +03:00
Aliaksandr Valialkin
7734fc8012 lib/storage: share tsids across all the partSearch instances
This should reduce memory usage when big number of time series matches the given query.
2019-09-23 22:36:16 +03:00
Aliaksandr Valialkin
67a2bcb98a lib/{storage,mergeset}: verify PrepareBlock callback results
Do not touch the first and the last item passed to PrepareBlock
in order to preserve sort order of mergeset blocks.
2019-09-23 20:46:33 +03:00
Aliaksandr Valialkin
3304dc1e85 lib/mergeset: detect whether we are in test by executable suffix 2019-09-22 23:12:35 +03:00
Aliaksandr Valialkin
d2ed8cb0b2 lib/storage: generate the first tag->metricIDs item in a mergeset block with a single metricID
The first item from each mergeset block goes into index (lib/mergeset.blockHeader),
so it must be short in order to reduce index size.
2019-09-22 19:37:50 +03:00
Aliaksandr Valialkin
0a9cb6368e lib/workingsetcache: remove data race when resetting c.misses 2019-09-22 19:37:09 +03:00
Aliaksandr Valialkin
7d13c31566 lib/{storage,mergeset}: merge tag->metricID rows into tag->metricIDs rows for common tag values
This should improve lookup performance if the same `label=value` pair exists
in big number of time series.
This should also reduce memory usage for mergeset data cache, since `tag->metricIDs` rows
occupy less space than the original `tag->metricID` rows.
2019-09-20 22:06:23 +03:00
Aliaksandr Valialkin
272e2f77c9 lib/encoding: optimize UnmarshalUint* and UnmarshalInt* 2019-09-20 13:08:24 +03:00
Aliaksandr Valialkin
7e0c6d4ca6 lib/storage: optimize selecting all the metricIDs by scanning MetricID->TSID entries instead of tag->MetricID entries
The number of MetricID->TSID entries is smaller than the number of tag->MetricID entries
and MetricID->TSID entries are usually shorter than tag->MetricID entries.
This should improve performance when selecting all the metricIDs.
2019-09-20 11:57:57 +03:00
Aliaksandr Valialkin
bf8505353a lib/mergeset: rename misleading mergeSmallParts to mergeExistingParts 2019-09-19 21:48:36 +03:00
Aliaksandr Valialkin
ebbef20535 lib/mergeset: use sort.IsSorted instead of sort.SliceIsSorted in inmemoryBlock.isSorted in order to reduce memory allocations 2019-09-19 20:13:54 +03:00
Aliaksandr Valialkin
89234f395d lib/storage: use sort.Sort instead of sort.slice in getSortedMetricIDs 2019-09-19 20:08:13 +03:00
Aliaksandr Valialkin
6e586fa09c lib/storage: skip duplicate call to intersectMetricIDsWithTagFilter on zero successful intersects 2019-09-19 17:51:10 +03:00
Aliaksandr Valialkin
410f993bf6 lib/mergeset: fill partHeader.firstItem on first block flush 2019-09-19 17:48:22 +03:00
Aliaksandr Valialkin
c05885fb5f lib/storage: mark tag filter returning errFallbackToMetricNameMatch as useless
This will save CPU on subsequent calls for this filter
2019-09-18 19:11:44 +03:00
Aliaksandr Valialkin
db71c940ea lib/storage: properly construct keys for uselessTagFiltersCache and register useless negative tag filters there 2019-09-17 23:18:37 +03:00
Aliaksandr Valialkin
41ef6b060e lib/mergeset: properly check for sorted block headers
Fix a typo for https://github.com/VictoriaMetrics/VictoriaMetrics/issues/181
2019-09-13 21:59:38 +03:00
hanzai
08cde5e3f4 lib/workingsetcache: adjust switching from mode=split to mode=whole smoothly and load cachefile successfully 2019-09-13 19:13:16 +03:00
Aliaksandr Valialkin
b101064f8b all: report the number of bytes read on io.ReadFull error
This should simplify error investigation similar to https://github.com/VictoriaMetrics/VictoriaMetrics/issues/175
2019-09-11 14:50:24 +03:00
Aliaksandr Valialkin
568ff61dcf lib/mergeset: dynamically calculate the maximum number of items per part, which can be cached in OS page cache 2019-09-09 11:42:45 +03:00
Aliaksandr Valialkin
2c654258ef lib/fs: add MustStopDirRemover for waiting until pending directories are removed on graceful shutdown
This patch is mainly required for laggy NFS. See https://github.com/VictoriaMetrics/VictoriaMetrics/issues/162
2019-09-05 11:17:17 +03:00
Aliaksandr Valialkin
2c2bd897dd lib/storage: remove duplicate tag keys on MetricName.Marshal call
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/172
2019-09-04 18:13:51 +03:00
Aliaksandr Valialkin
82bfe818d0 lib/fs: try harder with directory removal on NFS in the event of temporary lock
Do not give up after 11 attempts of directory removal on laggy NFS.

Add `vm_nfs_dir_remove_failed_attempts_total` metric for counting the number of failed attempts
on directory removal.

Log failed attempts on directory removal after long sleep times.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/162
2019-09-04 12:24:41 +03:00
Aliaksandr Valialkin
0b0153ba3d lib/storage: invalidate tagFilters -> TSIDS cache when newly added index data becomes visible to search
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/163
2019-08-29 15:08:44 +03:00
Aliaksandr Valialkin
8504a38214 lib/prombp: apply ba06b47c16
The following commands used:

gofmt -r '(uint64(x)&0x7F)<<shift -> uint64(x&0x7F)<<shift' -w ./lib/prompb/
gofmt -r '(int64(x)&0x7F)<<shift -> int64(x&0x7F)<<shift' -w ./lib/prompb/
2019-08-29 13:35:54 +03:00
Aliaksandr Valialkin
604a4312f9 all: port to FreeBSD on GOARCH=amd64 2019-08-28 01:46:09 +03:00
Aliaksandr Valialkin
da07a6fb38 lib/storage: go fmt 2019-08-27 14:28:24 +03:00
Aliaksandr Valialkin
a63b69e9e2 lib/storage: report proper maxMetrics limit when more than -search.maxUniqueTimeseries series match the given filters 2019-08-27 14:21:31 +03:00
Aliaksandr Valialkin
82e813bad3 lib/storage: properly handle (?i) in the tag filter regexp
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/161
2019-08-26 00:44:56 +03:00
Aliaksandr Valialkin
e2eac858b5 lib/storage: calculate the maximum number of rows per small part from -memory.allowedPercent
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/159

This simplifies error detection additionally to the `vm_rows_ignored_total` counters.
2019-08-25 15:29:09 +03:00
Aliaksandr Valialkin
0a8dd9cc9a lib/storage: calculate the maximum number of rows per small part from -memory.allowedPercent
This should improve query speed over recent data on machines with big amounts of RAM
2019-08-25 14:41:32 +03:00
Aliaksandr Valialkin
bc576fb386 lib/storage: properly limit the number of output rows in small and big parts storage
Previously small parts storage didn't take into account the available disk space for big parts.
2019-08-25 14:41:32 +03:00
Aliaksandr Valialkin
947decb3dd lib/storage: remove outdated comment on maxRowsPerSmallPart
The commend became outdated after the commit ed6ac1a5df027f0dfc22448e3b27c26b6f77c67a,
which stops merging of small parts on graceful shutdown instead of waiting
for their completion.
2019-08-25 13:46:10 +03:00
Aliaksandr Valialkin
c197641978 all: return 503 http error if service is temporarily unavailable
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/156
2019-08-23 09:49:50 +03:00
Aliaksandr Valialkin
e734076f0f app/vminsert: allow setting the maximum number of labels per time series via -maxLabelsPerTimeseries 2019-08-23 08:47:18 +03:00
Aliaksandr Valialkin
4ed63d033a lib/storage: add benchmarks for regexp filter match / mismatch
These benchmarks allow estimate the performance of regexp filters in promql
2019-08-22 16:37:19 +03:00
Aliaksandr Valialkin
6ec6a8d7c1 lib/storage: try slower path for searching the tag filter with the minimum number of matching time series before giving up with increase -search.maxUniqueTimeseries error 2019-08-19 16:07:05 +03:00
Aliaksandr Valialkin
c59f5c4865 lib/storage: pre-allocate memory for blockHeader slice in unmarshalBlockHeaders
This reduces memory usage and memory fragmentation when working with big number of time series
2019-08-19 12:46:45 +03:00
Aliaksandr Valialkin
f696cc503a lib/workingsetcache: automatically detect when it is better to double cache capacity 2019-08-15 22:58:04 +03:00
Aliaksandr Valialkin
99eed2ca14 lib/storage: properly cache tagFilters -> TSIDs entries from historical index 2019-08-14 02:32:25 +03:00
Aliaksandr Valialkin
f1d81b9405 lib/storage: compress contents of cache for tagFilters -> TSIDs
This should increase cache capacity
2019-08-14 02:32:22 +03:00
Aliaksandr Valialkin
8c2158af24 all: use workingsetcache instead of fastcache
This should reduce the amount of RAM required for processing time series
with non-zero churn rate.

The previous cache behavior can be restored with `-cache.oldBehavior` command-line flag.
2019-08-13 21:40:28 +03:00
Aliaksandr Valialkin
51263b1a45 lib/fs: add test for IsTemporaryFileName 2019-08-13 21:33:54 +03:00
Aliaksandr Valialkin
5a7ab0d90b lib/storage: remove broken BenchmarkIndexDBSearchTSIDs 2019-08-13 20:21:23 +03:00
Aliaksandr Valialkin
39f3f3a517 lib: move common code for creating flock.lock file into fs.CreateFlockFile 2019-08-13 01:46:20 +03:00
Aliaksandr Valialkin
73f866d874 lib/fs: atomically create file with the given contents on WriteFileAtomically
This should prevent from `transaction` and `metadata.json` files corruption
on unclean shutdown such as OOM, `kill -9`, power loss, etc.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/148
2019-08-12 15:02:04 +03:00
Aliaksandr Valialkin
4fb635b0c9 lib/storage: do not change timestamps to constant rate if values are constant or have constant delta
This breaks the original timestamps, which results in issues like
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/120 and
https://github.com/VictoriaMetrics/VictoriaMetrics/issues/141 .
2019-08-06 15:40:17 +03:00
Aliaksandr Valialkin
f56c1298ad app/vmstorage: add vm_concurrent_addrows_* metrics for tracking concurrency for Storage.AddRows calls
Track also the number of dropped rows due to the exceeded timeout
on concurrency limit for Storage.AddRows. This number is tracked in `vm_concurrent_addrows_dropped_rows_total`
2019-08-06 15:08:43 +03:00
Aliaksandr Valialkin
a3ecf3c1f7 lib/storage: properly reset partSearch.fetchData in partSearch.reset 2019-08-05 09:55:50 +03:00
Aliaksandr Valialkin
880b1d80b1 app/vmselect: optimize /api/v1/series by skipping storage data
Fetch and process only time series metainfo.
2019-08-04 23:00:46 +03:00
Aliaksandr Valialkin
a99e89945e lib/decimal: modernize tests a bit 2019-07-31 21:09:54 +03:00
Aliaksandr Valialkin
b7c4b0c6d2 lib/storage: fix matching against tag filter with empty name
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/137
2019-07-30 15:15:21 +03:00
Aliaksandr Valialkin
c6bec48927 lib/storage: add metrics for calculating skipped rows outside the retention
The metrics are:

    - vm_too_big_timestamp_rows_total
    - vm_too_small_timestamp_rows_total
2019-07-26 14:11:56 +03:00
Aliaksandr Valialkin
f2e8d54fb0 lib/encoding/zstd: go fmt 2019-07-25 01:37:57 +03:00
Aliaksandr Valialkin
97b5dc7122 lib/encoding/zstd: disable CRC checks in pure Go build
This should give slightly better compression and decompressions performance.
Additionally this shaves off 4 bytes per each compressed block.
2019-07-24 19:17:32 +03:00
Aliaksandr Valialkin
54f035d4ce all: small updates after PR #114 2019-07-24 17:43:43 +03:00
Aliaksandr Valialkin
7a133567fb lib/encoding: small fixes in tests after the PR #114 2019-07-24 17:43:39 +03:00
Roman Khavronenko
fcf09aaa3c all: add Pure Go build (pull request #114)
Updates #94
2019-07-24 17:43:32 +03:00
Aliaksandr Valialkin
3fae34eeb4 lib/encoding: improve gauge series detection
- Series with negative values are always gauges
- Counters may only have increasing values with possible counter resets

This should improve compression ratio for gauge series which
were previously mistakenly detected as counters.
2019-07-20 14:05:25 +03:00
Aliaksandr Valialkin
7abb96b454 lib/netutil: do not count timeouts as network errors 2019-07-15 23:06:13 +03:00
Aliaksandr Valialkin
73a47d2a53 lib/storage: remove unused function isTooBigTimeRangeForDateMetricIDs 2019-07-12 02:28:40 +03:00
Aliaksandr Valialkin
97f9397687 lib/storage: do not reduce maxMetrics on time ranges exceeding maxDaysForDateMetricIDs
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/95
2019-07-12 02:21:52 +03:00
Aliaksandr Valialkin
4ca66344ee lib/storage: do not pollute inverted index with data for samples outside the retention period 2019-07-11 17:11:33 +03:00
Aliaksandr Valialkin
0522efb2d6 lib/storage: add missing tagFilter.Marshal func 2019-07-11 15:01:01 +03:00
Aliaksandr Valialkin
12b1d67b41 lib/storage: use fast path for orSuffix when searching for metricIDs against plain tag value 2019-07-11 14:48:51 +03:00
Aliaksandr Valialkin
bf2e1b0ac1 lib/storage: remember and skip individual tag filters matching too many metrics
This saves CPU time by skipping useless matching for individual tag filters.
2019-07-11 14:48:47 +03:00
Aliaksandr Valialkin
ba8195c58e all: consistency renaming: bytesSize -> sizeBytes 2019-07-10 00:47:42 +03:00
Aliaksandr Valialkin
41f512af1c all: add vm_data_size_bytes metrics for easy monitoring of on-disk data size and on-disk inverted index size 2019-07-04 19:43:04 +03:00
Aliaksandr Valialkin
6682a35731 lib/memory: attempt #3 to determine memory limit for LXC container
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/84
2019-07-01 14:01:57 +03:00
Aliaksandr Valialkin
b6ea1a7d5e lib/mergeset: make fmt 2019-06-29 14:25:46 +03:00
Aliaksandr Valialkin
ffc1bb00f6 lib/storage: skip non-matching metricIDs in sortedFilter
This should improve performance for big sorteFilter lists.
2019-06-29 13:49:40 +03:00
Aliaksandr Valialkin
2257dcd278 lib/mergeset: speed up binarySearchKey by skipping the first item during binary search 2019-06-29 13:49:32 +03:00
Aliaksandr Valialkin
6ea12a079e lib/memory: remove TestReadLXCMemoryLimit, since it doesnt work in Travis 2019-06-28 18:23:06 +03:00
Aliaksandr Valialkin
d0732d3137 lib/memory: attempt #2 to determine memory limit inside LXC container
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/84
2019-06-28 18:08:57 +03:00
Aliaksandr Valialkin
ad436757c3 lib/memory: an attempt to read proper memory limit inside LXC container
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/84
2019-06-28 15:35:02 +03:00
Aliaksandr Valialkin
416d27ef11 lib/storage: optimize time series search by regexp filter
This should improve search speed on label filters like `{foo=~"bar.+baz"}`
2019-06-27 16:18:00 +03:00
Aliaksandr Valialkin
ee23a143b9 lib/storage: make sure non-nil args are passed to openIndexDB 2019-06-25 20:10:08 +03:00
Aliaksandr Valialkin
8b0a63722f lib/storage: reduce too big maxMetrics in getTagFilterWithMinMetricIDsCountAdaptive
This should improve performance on inverted index search for big amount of unique time series
when big -search.maxUniqueTimeseries is set.
2019-06-25 19:57:31 +03:00
Aliaksandr Valialkin
0263cb0adc lib/storage: free up memory from caches owned by indexDB when it is deleted 2019-06-25 14:41:16 +03:00
Aliaksandr Valialkin
362e187011 lib/storage: use unversioned keys for tag cache in extDB
Data in ExtDB cannot be changed, so it is OK to use unversioned keys for tag cache.
This should improve performance for index lookups over big amount of time series.
2019-06-25 13:15:42 +03:00
Aliaksandr Valialkin
51e2f3b48f lib/storage: skip searching in extDB if it doesn't contain items for the given time range
This should improve inverted index search performance for big amount
of unique time series when the search is performed only on recent data.
2019-06-25 12:57:56 +03:00
Aliaksandr Valialkin
fb9358635d lib/storage: mention source parts on merge error
This should improve determining broken source part.

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/76
2019-06-24 14:09:46 +03:00
Aliaksandr Valialkin
9164c223ec all: initial stubs for Windows support; see https://github.com/VictoriaMetrics/VictoriaMetrics/issues/70 2019-06-20 20:07:41 +03:00
Aliaksandr Valialkin
786beb8fc8 Makefile: enable golangci-lint in make check_all 2019-06-20 15:00:58 +03:00
Aliaksandr Valialkin
9cac11db64 lib/storage: typo fixes found by golangci-lint; updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/69 2019-06-20 14:38:45 +03:00
Aliaksandr Valialkin
7778030f9f lib/netutil: remove unused TCPListener.name; updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/69 2019-06-20 14:36:19 +03:00
Aliaksandr Valialkin
a0c22a6830 app/vmstorage: add vm_cache_entries{type="storage/hour_metric_ids"} metric for tracking active time series count 2019-06-19 18:37:38 +03:00
Aliaksandr Valialkin
24ae3ef532 lib/prompb: remove superflouos bytes copying in ReadSnappy 2019-06-18 21:02:02 +03:00
Aliaksandr Valialkin
e40224d5de lib/flagutil: add NewArray helper func 2019-06-18 10:44:09 +03:00
Aliaksandr Valialkin
f9e1d32168 lib/storage: persist metric ids for the current and the previous hour on graceful shutdown
This should improve performance after restart when the db contains a lot of time series
with high time series churn (i.e. metrics from Kubernetes with many pods and frequent deployments)
2019-06-14 07:55:09 +03:00
Aliaksandr Valialkin
b0b93e3d50 lib/fs: sync parent dir in MustRemoveAll only if it exists
The parent directory may be non-existing when the deleted directory
didn't exist before the MustRemoveAll call
2019-06-12 02:16:15 +03:00
Aliaksandr Valialkin
18d6f293f7 lib/fs: consolidate *RemoveAll* funcs into a single MustRemoveAll func
The func syncs parent dir in order to persist directory removal
in the event of power loss
2019-06-12 01:55:18 +03:00
Aliaksandr Valialkin
28d9904efc lib/fs: panic with fatal error when directories cannot be removed
Unremoved directories may lead to inconsistent data directory,
so VictoriaMetrics will fail to start next time.

So panic on the first error when trying to remove directory in order
to simplify recover process.
2019-06-12 01:20:10 +03:00
Aliaksandr Valialkin
d897bc3f08 lib/fs: attempt #2 to work around NFS issue with directory removal
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/61
2019-06-12 01:07:29 +03:00
Aliaksandr Valialkin
51e2e255a6 lib/fs: consistency renaming SyncPath -> MustSyncPath, since it doesnt return error 2019-06-11 23:13:45 +03:00
Aliaksandr Valialkin
3fa4c28f6b lib/fs: make sure the created directory remains visible in the fs in the event of power loss
Updates https://github.com/VictoriaMetrics/VictoriaMetrics/issues/63
2019-06-11 23:08:17 +03:00
Aliaksandr Valialkin
0b7f751f60 lib/fs: use filepath.Dir instead of filepath.Split, since the filename is unused 2019-06-11 22:54:23 +03:00
Aliaksandr Valialkin
b491045a4b lib/{storage,mergeset}: sync filenames inside part when finalizing the part
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/63
2019-06-11 21:51:19 +03:00
Aliaksandr Valialkin
3437c30180 all: try hard removing directory with contents
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/61
2019-06-11 01:58:08 +03:00
Aliaksandr Valialkin
75a0acf72d app/vmselect: add /api/v1/labels/count handler for quick detection of labels with the maximum number of distinct values 2019-06-10 19:54:55 +03:00
Aliaksandr Valialkin
0ccedbdfd2 lib/storage: mention the accountID and projectID in error message when filtering out other (accountID, projectID) entries 2019-06-10 14:43:53 +03:00
Aliaksandr Valialkin
d54f5fec0b lib/storage: skip adaptive searching for tag filter matching the minimum number of metrics if the identical previous search didn't found such filter
This should improve speed for searching metrics among high number of time series
with high churn rate like in big Kubernetes clusters with frequent deployments.
2019-06-10 14:07:47 +03:00
Aliaksandr Valialkin
27e50e86f4 lib/storage: factor out getTagFilterWithMinMetricIDsCountAdaptive from updateMetricIDsForTagFilters 2019-06-10 13:26:00 +03:00
Aliaksandr Valialkin
b69d3dbd0c lib/storage: filter out metricIDs from another (AccountID, ProjectID) in getMetricIDsForRecentHours 2019-06-10 13:05:16 +03:00
Aliaksandr Valialkin
3059ae7be0 lib/storage: give clearer names to more functions 2019-06-10 12:59:33 +03:00
Aliaksandr Valialkin
d3a024d2d6 lib/storage: give more clear names to functions 2019-06-10 12:50:22 +03:00
Aliaksandr Valialkin
00e0760608 lib/storage: test GetSeriesCount 2019-06-10 12:40:33 +03:00
Aliaksandr Valialkin
e4cba5a7ed lib/storage: make getSeriesCount func indexSearch method 2019-06-10 12:29:24 +03:00
Aliaksandr Valialkin
d882afa905 lib/storage: optimize time series lookup for recent hours when the db contains many millions of time series with high churn rate (aka frequent deployments in Kubernetes) 2019-06-09 19:14:04 +03:00
Aliaksandr Valialkin
8cf0a0e59c app/vminsert: split vm_rows_inserted_total into per-(accountID, projectID) metrics
Fixes https://github.com/VictoriaMetrics/VictoriaMetrics/issues/59
2019-06-07 22:11:20 +03:00
Aliaksandr Valialkin
1d87abc8eb lib/procutil: typo fix in comment to WaitForSigterm 2019-06-03 16:54:37 +03:00
Aliaksandr Valialkin
a2986cde70 lib/storage: tune updating a map with today`s metric ids
- Increase update iterval from 1s to 10s. This should reduce CPU usage
  for large amounts of metric ids with constant churn.
- Reduce pendingTodayMetricIDsLock lock duration during the update.
2019-06-02 22:00:13 +03:00
Aliaksandr Valialkin
e27fd5148a lib/storage: speed up checking metricID existence in the list for the current date 2019-06-02 18:34:20 +03:00
Aliaksandr Valialkin
a6d02ff275 lib/timerpool: use timer pool in concurrency limiters
This should reduce the number of memory allocations in highly loaded system
2019-05-28 17:30:10 +03:00
Aliaksandr Valialkin
bdf696ef18 all: fix misspellings 2019-05-25 21:51:24 +03:00
Aliaksandr Valialkin
121a920a18 Makefile: add -s flag to go fmt in make fmt command 2019-05-25 21:44:36 +03:00
Aliaksandr Valialkin
9b64dfee4b lib/encoding: add vm_zstd_block_{compress|decompress}_calls_total for determining the number CompressZSTD / DecompressZSTD calls 2019-05-24 13:01:15 +03:00
Aliaksandr Valialkin
b23352dc9e lib/encoding: add vm_zstd_block_{original|compressed}_bytes_total metrics for rough estimation of block compression ratio 2019-05-24 12:34:51 +03:00
Aliaksandr Valialkin
f67f40d63a lib/encoding: substitute CompressZSTD with CompressZSTDLevel 2019-05-24 12:32:49 +03:00
Aliaksandr Valialkin
a26e774eca lib/httpserver: add -http.disableResponseCompression flag, which may help saving CPU resources at the cost of higher network bandwidth usage 2019-05-24 12:20:36 +03:00
Aliaksandr Valialkin
820cdae88d lib/decimal: add a comment explaining weird code in maxUpExponent. Fixes #29 2019-05-23 17:18:50 +03:00
Aliaksandr Valialkin
24578b4bb1 all: open-sourcing cluster version 2019-05-23 00:25:38 +03:00
Aliaksandr Valialkin
1836c415e6 all: open-sourcing single-node version 2019-05-23 00:18:06 +03:00