Merge branch 'upstream/main' into enforce-query-tenancy

Signed-off-by: Jacob Baungard Hansen <jacobbaungard@redhat.com>

.busybox-versions

@@ -1,7 +1,7 @@
 # Auto generated by busybox-updater.sh. DO NOT EDIT
-amd64=393d14abb68b8b2d88304c72ac25b5ce130aa3a1d57ba7363e2c4d07d294513d
-arm64=9fe410fe5b8f283d057939a5b0a6f464ecb4bfe4a07d132d2846cfbe82cf43ea
-arm=a237b18458d6bcc8964e59ced627ea46eb9aae68875ea833c61d5050a742e624
-ppc64le=cbb9892625fd0d4c625afe8255fe35699a163bc4d74925dfcca74ee7cc43d4ba
-riscv64=fa1350d80e4481d3671d808fbe239e4075205f69c940e7e85711bdc39bf8e181
-s390x=1e3e5a05847ad67da2b148d952931cf6f716a334ab06ea00742560a2ff985c7d
+amd64=f173c44fab35484fa0e940e42929efe2a2f506feda431ba72c5f0d79639d7f55
+arm64=6277ab6abe348994989b3959d7c125d7a487012aedb80570ec28652a012c69d6
+arm=31533906c9eadc190de436bcbc021207d90839777c1b95991edd15e7df5d34ad
+ppc64le=ace2ea29bf8e4267c293ed5570df249039a0bc3949d3371429a71cf114e8a9e2
+riscv64=ff38cae5b5ed16251631e55156a7d92977bf5a8c4714ff6e3c333f7acb0297e0
+s390x=59d0ed3060aef57d1b23bc353a2223af24a6e1d035486647eb599a77ff2d446e

CHANGELOG.md

@@ -17,7 +17,13 @@ We use *breaking :warning:* to mark changes that are not backward compatible (re
 ### Added
 
 - [#6756](https://github.com/thanos-io/thanos/pull/6756) Query: Add the following options to allow enforcement of tenancy on the query path: `query.enable-tenancy`, `query.tenant-label-name`.
+- [#6944](https://github.com/thanos-io/thanos/pull/6944) Receive: Added a new flag for maximum retention bytes.
 - [#6891](https://github.com/thanos-io/thanos/pull/6891) Objstore: Bump `objstore` which adds support for Azure Workload Identity.
+- [#6453](https://github.com/thanos-io/thanos/pull/6453) Sidecar: Added `--reloader.method` to support configuration reloads via SIHUP signal.
+- [#6925](https://github.com/thanos-io/thanos/pull/6925) Store Gateway: Support float native histogram.
+- [#6954](https://github.com/thanos-io/thanos/pull/6954) Index Cache: Support tracing for fetch APIs.
+- [#6943](https://github.com/thanos-io/thanos/pull/6943) Ruler: Added `keep_firing_for` field in alerting rule.
+- [#6972](https://github.com/thanos-io/thanos/pull/6972) Store Gateway: Apply series limit when streaming series for series actually matched if lazy postings is enabled.
 
 ### Changed
 

MAINTAINERS.md

@@ -10,8 +10,7 @@
 | Prem Saraswat | prmsrswt@gmail.com | `@Prem Saraswat` | [@onprem](https://github.com/onprem) | Red Hat |
 | Matthias Loibl | mail@matthiasloibl.com | `@metalmatze` | [@metalmatze](https://github.com/metalmatze) | Polar Signals |
 | Ben Ye | yb532204897@gmail.com | `@yeya24` | [@yeya24](https://github.com/yeya24) | Amazon Web Services |
-| Wiard van Rij | wiard@outlook.com | `@wiard van Rij` | [@wiardvanrij](https://github.com/wiardvanrij) | Roku |
-| Matej Gera | matejgera@gmail.com | `@Matej Gera` | [@matej-g](https://github.com/matej-g) | Red Hat |
+| Matej Gera | matejgera@gmail.com | `@Matej Gera` | [@matej-g](https://github.com/matej-g) | Coralogix |
 | Filip Petkovski | filip.petkovsky@gmail.com | `@Filip Petkovski` | [@fpetkovski](https://github.com/fpetkovski) | Shopify |
 | Saswata Mukherjee | saswata.mukhe@gmail.com | `@saswatamcode` | [@saswatamcode](https://github.com/saswatamcode) | Red Hat |
 | Michael Hoffmann | mhoffm@posteo.de | `@Michael Hoffmann` | [@MichaHoffmann](https://github.com/MichaHoffmann) | Aiven |

cmd/thanos/compact.go

@@ -239,7 +239,8 @@ func runCompact(
  consistencyDelayMetaFilter := block.NewConsistencyDelayMetaFilter(logger, conf.consistencyDelay, extprom.WrapRegistererWithPrefix("thanos_", reg))
  timePartitionMetaFilter := block.NewTimePartitionMetaFilter(conf.filterConf.MinTime, conf.filterConf.MaxTime)
 
- baseMetaFetcher, err := block.NewBaseFetcher(logger, conf.blockMetaFetchConcurrency, insBkt, conf.dataDir, extprom.WrapRegistererWithPrefix("thanos_", reg))
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ baseMetaFetcher, err := block.NewBaseFetcher(logger, conf.blockMetaFetchConcurrency, insBkt, baseBlockIDsFetcher, conf.dataDir, extprom.WrapRegistererWithPrefix("thanos_", reg))
  if err != nil {
  return errors.Wrap(err, "create meta fetcher")
  }

cmd/thanos/config.go

@@ -114,8 +114,18 @@ type reloaderConfig struct {
  ruleDirectories []string
  watchInterval time.Duration
  retryInterval time.Duration
+ method string
+ processName string
 }
 
+const (
+ // HTTPReloadMethod reloads the configuration using the HTTP reload endpoint.
+ HTTPReloadMethod = "http"
+
+ // SignalReloadMethod reloads the configuration sending a SIGHUP signal to the process.
+ SignalReloadMethod = "signal"
+)
+
 func (rc *reloaderConfig) registerFlag(cmd extkingpin.FlagClause) *reloaderConfig {
  cmd.Flag("reloader.config-file",
  "Config file watched by the reloader.").
@@ -132,6 +142,12 @@ func (rc *reloaderConfig) registerFlag(cmd extkingpin.FlagClause) *reloaderConfi
  cmd.Flag("reloader.retry-interval",
  "Controls how often reloader retries config reload in case of error.").
  Default("5s").DurationVar(&rc.retryInterval)
+ cmd.Flag("reloader.method",
+ "Method used to reload the configuration.").
+ Default(HTTPReloadMethod).EnumVar(&rc.method, HTTPReloadMethod, SignalReloadMethod)
+ cmd.Flag("reloader.process-name",
+ "Executable name used to match the process being reloaded when using the signal method.").
+ Default("prometheus").StringVar(&rc.processName)
 
  return rc
 }

cmd/thanos/downsample.go

@@ -90,7 +90,8 @@ func RunDownsample(
  insBkt := objstoretracing.WrapWithTraces(objstore.WrapWithMetrics(bkt, extprom.WrapRegistererWithPrefix("thanos_", reg), bkt.Name()))
 
  // While fetching blocks, filter out blocks that were marked for no downsample.
- metaFetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix("thanos_", reg), []block.MetadataFilter{
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ metaFetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix("thanos_", reg), []block.MetadataFilter{
  block.NewDeduplicateFilter(block.FetcherConcurrency),
  downsample.NewGatherNoDownsampleMarkFilter(logger, insBkt, block.FetcherConcurrency),
  })

cmd/thanos/main_test.go

@@ -157,7 +157,8 @@ func TestRegression4960_Deadlock(t *testing.T) {
 
  metrics := newDownsampleMetrics(prometheus.NewRegistry())
  testutil.Equals(t, 0.0, promtest.ToFloat64(metrics.downsamples.WithLabelValues(meta.Thanos.GroupKey())))
- metaFetcher, err := block.NewMetaFetcher(nil, block.FetcherConcurrency, bkt, "", nil, nil)
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, bkt)
+ metaFetcher, err := block.NewMetaFetcher(nil, block.FetcherConcurrency, bkt, baseBlockIDsFetcher, "", nil, nil)
  testutil.Ok(t, err)
 
  metas, _, err := metaFetcher.Fetch(ctx)
@@ -196,7 +197,8 @@ func TestCleanupDownsampleCacheFolder(t *testing.T) {
 
  metrics := newDownsampleMetrics(prometheus.NewRegistry())
  testutil.Equals(t, 0.0, promtest.ToFloat64(metrics.downsamples.WithLabelValues(meta.Thanos.GroupKey())))
- metaFetcher, err := block.NewMetaFetcher(nil, block.FetcherConcurrency, bkt, "", nil, nil)
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, bkt)
+ metaFetcher, err := block.NewMetaFetcher(nil, block.FetcherConcurrency, bkt, baseBlockIDsFetcher, "", nil, nil)
  testutil.Ok(t, err)
 
  metas, _, err := metaFetcher.Fetch(ctx)

cmd/thanos/query.go

@@ -261,7 +261,10 @@ func registerQuery(app *extkingpin.App) {
  Files: *fileSDFiles,
  RefreshInterval: *fileSDInterval,
  }
- fileSD = file.NewDiscovery(conf, logger)
+ var err error
+ if fileSD, err = file.NewDiscovery(conf, logger, reg); err != nil {
+ return err
+ }
  }
 
  if *webRoutePrefix == "" {

cmd/thanos/receive.go

@@ -10,6 +10,7 @@ import (
  "strings"
  "time"
 
+ "github.com/alecthomas/units"
  extflag "github.com/efficientgo/tools/extkingpin"
  "github.com/go-kit/log"
  "github.com/go-kit/log/level"
@@ -84,6 +85,7 @@ func registerReceive(app *extkingpin.App) {
  MaxBlockDuration: int64(time.Duration(*conf.tsdbMaxBlockDuration) / time.Millisecond),
  RetentionDuration: int64(time.Duration(*conf.retention) / time.Millisecond),
  OutOfOrderTimeWindow: int64(time.Duration(*conf.tsdbOutOfOrderTimeWindow) / time.Millisecond),
+ MaxBytes: int64(conf.tsdbMaxBytes),
  OutOfOrderCapMax: conf.tsdbOutOfOrderCapMax,
  NoLockfile: conf.noLockFile,
  WALCompression: wlog.ParseCompressionType(conf.walCompression, string(wlog.CompressionSnappy)),
@@ -165,6 +167,8 @@ func runReceive(
  // Has this thanos receive instance been configured to ingest metrics into a local TSDB?
  enableIngestion := receiveMode == receive.IngestorOnly || receiveMode == receive.RouterIngestor
 
+ isOOOEnabled := tsdbOpts.OutOfOrderTimeWindow > 0
+
  upload := len(confContentYaml) > 0
  if enableIngestion {
  if upload {
@@ -174,6 +178,9 @@ func runReceive(
  "Compaction needs to be disabled (tsdb.min-block-duration = tsdb.max-block-duration)", tsdbOpts.MaxBlockDuration, tsdbOpts.MinBlockDuration)
  }
  level.Warn(logger).Log("msg", "flag to ignore min/max block duration flags differing is being used. If the upload of a 2h block fails and a tsdb compaction happens that block may be missing from your Thanos bucket storage.")
+ if isOOOEnabled {
+ level.Warn(logger).Log("msg", "out-of-order support is also enabled which means that Receiver will now upload compacted blocks to not lose any data. Vertical compaction needs to be enabled on Compactor! See https://github.com/prometheus/prometheus/issues/13112")
+ }
  }
  // The background shipper continuously scans the data directory and uploads
  // new blocks to object storage service.
@@ -211,6 +218,7 @@ func runReceive(
  conf.tenantLabelName,
  bkt,
  conf.allowOutOfOrderUpload,
+ isOOOEnabled,
  hashFunc,
  )
  writer := receive.NewWriter(log.With(logger, "component", "receive-writer"), dbs, &receive.WriterOptions{
@@ -809,6 +817,7 @@ type receiveConfig struct {
  tsdbOutOfOrderCapMax int64
  tsdbAllowOverlappingBlocks bool
  tsdbMaxExemplars int64
+ tsdbMaxBytes units.Base2Bytes
  tsdbWriteQueueSize int64
  tsdbMemorySnapshotOnShutdown bool
  tsdbEnableNativeHistograms bool
@@ -916,6 +925,8 @@ func (rc *receiveConfig) registerFlag(cmd extkingpin.FlagClause) {
 
  cmd.Flag("tsdb.allow-overlapping-blocks", "Allow overlapping blocks, which in turn enables vertical compaction and vertical query merge. Does not do anything, enabled all the time.").Default("false").BoolVar(&rc.tsdbAllowOverlappingBlocks)
 
+ cmd.Flag("tsdb.max-retention-bytes", "Maximum number of bytes that can be stored for blocks. A unit is required, supported units: B, KB, MB, GB, TB, PB, EB. Ex: \"512MB\". Based on powers-of-2, so 1KB is 1024B.").Default("0").BytesVar(&rc.tsdbMaxBytes)
+
  cmd.Flag("tsdb.wal-compression", "Compress the tsdb WAL.").Default("true").BoolVar(&rc.walCompression)
 
  cmd.Flag("tsdb.no-lockfile", "Do not create lockfile in TSDB data directory. In any case, the lockfiles will be deleted on next startup.").Default("false").BoolVar(&rc.noLockFile)

cmd/thanos/sidecar.go

@@ -5,7 +5,9 @@ package main
 
 import (
  "context"
+ "fmt"
  "math"
+ "net/http"
  "net/url"
  "sync"
  "time"
@@ -62,18 +64,44 @@ func registerSidecar(app *extkingpin.App) {
  return errors.Wrap(err, "error while parsing config for request logging")
  }
 
+ httpConfContentYaml, err := conf.prometheus.httpClient.Content()
+ if err != nil {
+ return errors.Wrap(err, "getting http client config")
+ }
+ httpClientConfig, err := httpconfig.NewClientConfigFromYAML(httpConfContentYaml)
+ if err != nil {
+ return errors.Wrap(err, "parsing http config YAML")
+ }
+
+ httpClient, err := httpconfig.NewHTTPClient(*httpClientConfig, "thanos-sidecar")
+ if err != nil {
+ return errors.Wrap(err, "Improper http client config")
+ }
+
+ opts := reloader.Options{
+ HTTPClient: *httpClient,
+ CfgFile: conf.reloader.confFile,
+ CfgOutputFile: conf.reloader.envVarConfFile,
+ WatchedDirs: conf.reloader.ruleDirectories,
+ WatchInterval: conf.reloader.watchInterval,
+ RetryInterval: conf.reloader.retryInterval,
+ }
+
+ switch conf.reloader.method {
+ case HTTPReloadMethod:
+ opts.ReloadURL = reloader.ReloadURLFromBase(conf.prometheus.url)
+ case SignalReloadMethod:
+ opts.ProcessName = conf.reloader.processName
+ opts.RuntimeInfoURL = reloader.RuntimeInfoURLFromBase(conf.prometheus.url)
+ default:
+ return fmt.Errorf("invalid reload method: %s", conf.reloader.method)
+ }
+
  rl := reloader.New(log.With(logger, "component", "reloader"),
  extprom.WrapRegistererWithPrefix("thanos_sidecar_", reg),
- &reloader.Options{
- ReloadURL: reloader.ReloadURLFromBase(conf.prometheus.url),
- CfgFile: conf.reloader.confFile,
- CfgOutputFile: conf.reloader.envVarConfFile,
- WatchedDirs: conf.reloader.ruleDirectories,
- WatchInterval: conf.reloader.watchInterval,
- RetryInterval: conf.reloader.retryInterval,
- })
+ &opts)
 
- return runSidecar(g, logger, reg, tracer, rl, component.Sidecar, *conf, grpcLogOpts, tagOpts)
+ return runSidecar(g, logger, reg, tracer, rl, component.Sidecar, *conf, httpClient, grpcLogOpts, tagOpts)
  })
 }
 
@@ -85,28 +113,13 @@ func runSidecar(
  reloader *reloader.Reloader,
  comp component.Component,
  conf sidecarConfig,
+ httpClient *http.Client,
  grpcLogOpts []grpc_logging.Option,
  tagOpts []tags.Option,
 ) error {
- httpConfContentYaml, err := conf.prometheus.httpClient.Content()
- if err != nil {
- return errors.Wrap(err, "getting http client config")
- }
- httpClientConfig, err := httpconfig.NewClientConfigFromYAML(httpConfContentYaml)
- if err != nil {
- return errors.Wrap(err, "parsing http config YAML")
- }
-
- httpClient, err := httpconfig.NewHTTPClient(*httpClientConfig, "thanos-sidecar")
- if err != nil {
- return errors.Wrap(err, "Improper http client config")
- }
-
- reloader.SetHttpClient(*httpClient)
 
  var m = &promMetadata{
  promURL: conf.prometheus.url,
-
  // Start out with the full time range. The shipper will constrain it later.
  // TODO(fabxc): minimum timestamp is never adjusted if shipping is disabled.
  mint: conf.limitMinTime.PrometheusTimestamp(),

cmd/thanos/store.go

@@ -339,7 +339,8 @@ func runStore(
  }
 
  ignoreDeletionMarkFilter := block.NewIgnoreDeletionMarkFilter(logger, insBkt, time.Duration(conf.ignoreDeletionMarksDelay), conf.blockMetaFetchConcurrency)
- metaFetcher, err := block.NewMetaFetcher(logger, conf.blockMetaFetchConcurrency, insBkt, dataDir, extprom.WrapRegistererWithPrefix("thanos_", reg),
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ metaFetcher, err := block.NewMetaFetcher(logger, conf.blockMetaFetchConcurrency, insBkt, baseBlockIDsFetcher, dataDir, extprom.WrapRegistererWithPrefix("thanos_", reg),
  []block.MetadataFilter{
  block.NewTimePartitionMetaFilter(conf.filterConf.MinTime, conf.filterConf.MaxTime),
  block.NewLabelShardedMetaFilter(relabelConfig),

cmd/thanos/tools_bucket.go

@@ -350,7 +350,8 @@ func registerBucketVerify(app extkingpin.AppClause, objStoreConfig *extflag.Path
 
  // We ignore any block that has the deletion marker file.
  filters := []block.MetadataFilter{block.NewIgnoreDeletionMarkFilter(logger, insBkt, 0, block.FetcherConcurrency)}
- fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg), filters)
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg), filters)
  if err != nil {
  return err
  }
@@ -407,7 +408,8 @@ func registerBucketLs(app extkingpin.AppClause, objStoreConfig *extflag.PathOrCo
  ignoreDeletionMarkFilter := block.NewIgnoreDeletionMarkFilter(logger, insBkt, 0, block.FetcherConcurrency)
  filters = append(filters, ignoreDeletionMarkFilter)
  }
- fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg), filters)
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg), filters)
  if err != nil {
  return err
  }
@@ -508,7 +510,8 @@ func registerBucketInspect(app extkingpin.AppClause, objStoreConfig *extflag.Pat
  }
  insBkt := objstoretracing.WrapWithTraces(objstore.WrapWithMetrics(bkt, extprom.WrapRegistererWithPrefix("thanos_", reg), bkt.Name()))
 
- fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg), nil)
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg), nil)
  if err != nil {
  return err
  }
@@ -651,7 +654,8 @@ func registerBucketWeb(app extkingpin.AppClause, objStoreConfig *extflag.PathOrC
  return err
  }
  // TODO(bwplotka): Allow Bucket UI to visualize the state of block as well.
- fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg),
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ fetcher, err := block.NewMetaFetcher(logger, block.FetcherConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg),
  []block.MetadataFilter{
  block.NewTimePartitionMetaFilter(filterConf.MinTime, filterConf.MaxTime),
  block.NewLabelShardedMetaFilter(relabelConfig),
@@ -829,7 +833,8 @@ func registerBucketCleanup(app extkingpin.AppClause, objStoreConfig *extflag.Pat
 
  var sy *compact.Syncer
  {
- baseMetaFetcher, err := block.NewBaseFetcher(logger, tbc.blockSyncConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg))
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ baseMetaFetcher, err := block.NewBaseFetcher(logger, tbc.blockSyncConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg))
  if err != nil {
  return errors.Wrap(err, "create meta fetcher")
  }
@@ -1371,7 +1376,8 @@ func registerBucketRetention(app extkingpin.AppClause, objStoreConfig *extflag.P
 
  var sy *compact.Syncer
  {
- baseMetaFetcher, err := block.NewBaseFetcher(logger, tbc.blockSyncConcurrency, insBkt, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg))
+ baseBlockIDsFetcher := block.NewBaseBlockIDsFetcher(logger, insBkt)
+ baseMetaFetcher, err := block.NewBaseFetcher(logger, tbc.blockSyncConcurrency, insBkt, baseBlockIDsFetcher, "", extprom.WrapRegistererWithPrefix(extpromPrefix, reg))
  if err != nil {
  return errors.Wrap(err, "create meta fetcher")
  }

docs/components/receive.md

@@ -425,6 +425,11 @@ Flags:
  ingesting a new exemplar will evict the oldest
  exemplar from storage. 0 (or less) value of
  this flag disables exemplars storage.
+ --tsdb.max-retention-bytes=0
+ Maximum number of bytes that can be stored for
+ blocks. A unit is required, supported units: B,
+ KB, MB, GB, TB, PB, EB. Ex: "512MB". Based on
+ powers-of-2, so 1KB is 1024B.
  --tsdb.no-lockfile Do not create lockfile in TSDB data directory.
  In any case, the lockfiles will be deleted on
  next startup.