copyblocks: add option to bypass duration check for no-compact blocks

CHANGELOG.md

@@ -274,6 +274,7 @@
 * [ENHANCEMENT] `tsdb-series`: Added `-min-time` and `-max-time` options to filter samples that are used for computing data-points per minute. #8844
 * [ENHANCEMENT] `mimir-rules-action`: Added new input to support matching target namespaces by regex. #9244
 * [ENHANCEMENT] `mimir-rules-action`: Added new inputs to support ignoring namespaces and ignoring namespaces by regex. #9258 #9324
+* [ENHANCEMENT] `copyblocks`: Added `--any-no-compact-block-duration`, which defaults to `false`, to simplify targeting blocks that are not awaiting compaction. #9439
 * [BUGFIX] `copyblocks`, `undelete-blocks`, `copyprefix`: use a multipart upload to server-side copy objects greater than 5GiB in size on S3. #9357
 
 ## 2.13.0

tools/copyblocks/README.md

@@ -11,7 +11,7 @@ The currently supported services are Amazon Simple Storage Service (S3 and S3-co
 - Prevents copying blocks multiple times to the same destination bucket by uploading block marker files to the source bucket
 - Runs continuously with periodic checks when supplied a time duration with `--copy-period`, otherwise runs one check then exits
 - Include or exclude users from having blocks copied (`--enabled-users` and `--disabled-users`)
-- Configurable minimum block duration (`--min-block-duration`) to avoid copying blocks that will be compacted
+- Configurable minimum block duration (`--min-block-duration`) and (`--any-no-compact-block-duration`) to target blocks that are not awaiting compaction
 - Configurable time range (`--min-time` and `--max-time`) to only copy blocks inclusively within a provided range
 - Log what would be copied without actually copying anything with `--dry-run`
 

tools/copyblocks/main.go

@@ -35,17 +35,18 @@ import (
 )
 
 type config struct {
-	copyConfig        objtools.CopyBucketConfig
-	minBlockDuration  time.Duration
-	minTime           flagext.Time
-	maxTime           flagext.Time
-	tenantConcurrency int
-	blockConcurrency  int
-	copyPeriod        time.Duration
-	enabledUsers      flagext.StringSliceCSV
-	disabledUsers     flagext.StringSliceCSV
-	dryRun            bool
-	httpListen        string
+	copyConfig                objtools.CopyBucketConfig
+	minBlockDuration          time.Duration
+	minTime                   flagext.Time
+	maxTime                   flagext.Time
+	tenantConcurrency         int
+	blockConcurrency          int
+	copyPeriod                time.Duration
+	enabledUsers              flagext.StringSliceCSV
+	disabledUsers             flagext.StringSliceCSV
+	dryRun                    bool
+	anyNoCompactBlockDuration bool
+	httpListen                string
 }
 
 func (c *config) registerFlags(f *flag.FlagSet) {
@@ -59,6 +60,7 @@ func (c *config) registerFlags(f *flag.FlagSet) {
 	f.Var(&c.enabledUsers, "enabled-users", "If not empty, only blocks for these users are copied.")
 	f.Var(&c.disabledUsers, "disabled-users", "If not empty, blocks for these users are not copied.")
 	f.BoolVar(&c.dryRun, "dry-run", false, "Don't perform any copy; only log what would happen.")
+	f.BoolVar(&c.anyNoCompactBlockDuration, "any-no-compact-block-duration", false, "If set, blocks marked as no-compact are not checked against min-block-duration")
 	f.StringVar(&c.httpListen, "http-listen-address", ":8080", "HTTP listen address.")
 }
 
@@ -269,7 +271,8 @@ func copyBlocks(ctx context.Context, cfg config, logger log.Logger, m *metrics)
 				return nil
 			}
 
-			if cfg.minBlockDuration > 0 {
+			skipDurationCheck := cfg.anyNoCompactBlockDuration && markers[blockID].noCompact
+			if !skipDurationCheck && cfg.minBlockDuration > 0 {
 				blockDuration := time.Millisecond * time.Duration(blockMeta.MaxTime-blockMeta.MinTime)
 				if blockDuration < cfg.minBlockDuration {
 					level.Debug(logger).Log("msg", "skipping block, block duration is less than the configured minimum duration", "block_duration", blockDuration, "configured_min_duration", cfg.minBlockDuration)