Set defaults to query ingesters, not store, for recent data

CHANGELOG.md

@@ -5,6 +5,9 @@
 ### Grafana Mimir
 
 * [CHANGE] Increased default configuration for `-server.grpc-max-recv-msg-size-bytes` and `-server.grpc-max-send-msg-size-bytes` from 4MB to 100MB. #1883
+* [CHANGE] Default values have changed for the following settings. This improves query performance for recent data (within 12h) by only reading from ingesters: #1909
+    - `-blocks-storage.bucket-store.ignore-blocks-within` now defaults to `10h` (previously `0`)
+    - `-querier.query-store-after` now defaults to `12h` (previously `0`)
 * [ENHANCEMENT] Store-gateway: Add the experimental ability to run requests in a dedicated OS thread pool. This feature can be configured using `-store-gateway.thread-pool-size` and is disabled by default. Replaces the ability to run index header operations in a dedicated thread pool. #1660 #1812
 * [BUGFIX] Fix regexp parsing panic for regexp label matchers with start/end quantifiers. #1883
 * [BUGFIX] Ingester: fixed deceiving error log "failed to update cached shipped blocks after shipper initialisation", occurring for each new tenant in the ingester. #1893

cmd/mimir/config-descriptor.json

@@ -1385,7 +1385,7 @@
           "required": false,
           "desc": "The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'.",
           "fieldValue": null,
-          "fieldDefaultValue": 0,
+          "fieldDefaultValue": 43200000000000,
           "fieldFlag": "querier.query-store-after",
           "fieldType": "duration"
         },
@@ -5004,7 +5004,7 @@
               "required": false,
               "desc": "Blocks with minimum time within this duration are ignored, and not loaded by store-gateway. Useful when used together with -querier.query-store-after to prevent loading young blocks, because there are usually many of them (depending on number of ingesters) and they are not yet compacted. Negative values or 0 disable the filter.",
               "fieldValue": null,
-              "fieldDefaultValue": 0,
+              "fieldDefaultValue": 36000000000000,
               "fieldFlag": "blocks-storage.bucket-store.ignore-blocks-within",
               "fieldType": "duration",
               "fieldCategory": "advanced"

cmd/mimir/help-all.txt.tmpl

@@ -305,7 +305,7 @@ Usage of ./cmd/mimir/mimir:
   -blocks-storage.bucket-store.consistency-delay duration
     	Minimum age of a block before it's being read. Set it to safe value (e.g 30m) if your object storage is eventually consistent. GCS and S3 are (roughly) strongly consistent.
   -blocks-storage.bucket-store.ignore-blocks-within duration
-    	Blocks with minimum time within this duration are ignored, and not loaded by store-gateway. Useful when used together with -querier.query-store-after to prevent loading young blocks, because there are usually many of them (depending on number of ingesters) and they are not yet compacted. Negative values or 0 disable the filter.
+    	Blocks with minimum time within this duration are ignored, and not loaded by store-gateway. Useful when used together with -querier.query-store-after to prevent loading young blocks, because there are usually many of them (depending on number of ingesters) and they are not yet compacted. Negative values or 0 disable the filter. (default 10h0m0s)
   -blocks-storage.bucket-store.ignore-deletion-marks-delay duration
     	Duration after which the blocks marked for deletion will be filtered out while fetching blocks. The idea of ignore-deletion-marks-delay is to ignore blocks that are marked for deletion with some delay. This ensures store can still serve blocks that are meant to be deleted but do not have a replacement yet. (default 1h0m0s)
   -blocks-storage.bucket-store.index-cache.backend string
@@ -1082,7 +1082,7 @@ Usage of ./cmd/mimir/mimir:
   -querier.query-ingesters-within duration
     	Maximum lookback beyond which queries are not sent to ingester. 0 means all queries are sent to ingester. (default 13h0m0s)
   -querier.query-store-after duration
-    	The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'.
+    	The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'. (default 12h0m0s)
   -querier.scheduler-address string
     	Address of the query-scheduler component, in host:port format. Only one of -querier.frontend-address or -querier.scheduler-address can be set. If neither is set, queries are only received via HTTP endpoint.
   -querier.shuffle-sharding-ingesters-lookback-period duration

cmd/mimir/help.txt.tmpl

@@ -340,7 +340,7 @@ Usage of ./cmd/mimir/mimir:
   -querier.query-ingesters-within duration
     	Maximum lookback beyond which queries are not sent to ingester. 0 means all queries are sent to ingester. (default 13h0m0s)
   -querier.query-store-after duration
-    	The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'.
+    	The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'. (default 12h0m0s)
   -querier.scheduler-address string
     	Address of the query-scheduler component, in host:port format. Only one of -querier.frontend-address or -querier.scheduler-address can be set. If neither is set, queries are only received via HTTP endpoint.
   -querier.timeout duration

docs/sources/operators-guide/configuring/reference-configuration-parameters/index.md

@@ -801,7 +801,7 @@ The `querier` block configures the querier.
 # the time range of the query sent to the store-gateway will be manipulated to
 # ensure the query end is not more recent than 'now - query-store-after'.
 # CLI flag: -querier.query-store-after
-[query_store_after: <duration> | default = 0s]
+[query_store_after: <duration> | default = 12h]
 
 # (advanced) Maximum duration into the future you can query. 0 to disable.
 # CLI flag: -querier.max-query-into-future
@@ -3331,7 +3331,7 @@ bucket_store:
   # are usually many of them (depending on number of ingesters) and they are not
   # yet compacted. Negative values or 0 disable the filter.
   # CLI flag: -blocks-storage.bucket-store.ignore-blocks-within
-  [ignore_blocks_within: <duration> | default = 0s]
+  [ignore_blocks_within: <duration> | default = 10h]
 
   # (advanced) Max size - in bytes - of a chunks pool, used to reduce memory
   # allocations. The pool is shared across all tenants. 0 to disable the limit.

docs/sources/operators-guide/running-production-environment/production-tips/index.md

@@ -46,14 +46,15 @@ When running Grafana Mimir at scale, querying non-compacted blocks might be inef
 - Non compacted blocks contain duplicated samples, as a result of the ingesters replication.
 - Querying many small TSDB indexes is slower than querying a few compacted TSDB indexes.
 
-Configure Grafana Mimir to ensure only compacted blocks are queried:
+The default values for `-querier.query-store-after`, `-querier.query-ingesters-within`, and `-blocks-storage.bucket-store.ignore-blocks-within` are set such that only compacted blocks are queried. In most cases, no additional configuration is required.
+
+Configure Grafana Mimir so large tenants are parallelized by the compactor:
 
 1. Configure compactor's `-compactor.split-and-merge-shards` and `-compactor.split-groups` for every tenant with more than 20 million active series. For more information about configuring the compactor's split and merge shards, refer to [compactor]({{< relref "../../architecture/components/compactor/index.md" >}}).
-1. Configure querier's `-querier.query-store-after` equal to `-querier.query-ingesters-within` minus five minutes. The five-minute delta is recommended to ensure the time range on the boundary is queried both from ingesters and queriers.
 
 #### How to estimate `-querier.query-store-after`
 
-Set the `-querier.query-store-after` to a duration that is large enough to give compactor enough time to compact newly uploaded blocks, and queriers and store-gateways to discover and synchronize newly compacted blocks.
+If not using the defaults, set the `-querier.query-store-after` to a duration that is large enough to give compactor enough time to compact newly uploaded blocks, and queriers and store-gateways to discover and synchronize newly compacted blocks.
 
 The following diagram shows all of the timings involved in the estimation. This diagram should be used only as a template and you can modify the assumptions based on real measurements in your Mimir cluster. The example makes the following assumptions:
 

integration/getting_started_with_gossiped_ring_test.go

@@ -41,6 +41,7 @@ func TestGettingStartedWithGossipedRing(t *testing.T) {
 		"-ingester.ring.observe-period":                     "5s", // to avoid conflicts in tokens
 		"-blocks-storage.bucket-store.bucket-index.enabled": "false",
 		"-blocks-storage.bucket-store.sync-interval":        "1s", // sync continuously
+		"-blocks-storage.bucket-store.ignore-blocks-within": "0",
 		"-blocks-storage.backend":                           "s3",
 		"-blocks-storage.s3.bucket-name":                    blocksBucketName,
 		"-blocks-storage.s3.access-key-id":                  e2edb.MinioAccessKey,

integration/ingester_sharding_test.go

@@ -51,6 +51,7 @@ func TestIngesterSharding(t *testing.T) {
 
 			// Enable shuffle sharding on read path but not lookback, otherwise all ingesters would be
 			// queried being just registered.
+			flags["-querier.query-store-after"] = "0"
 			flags["-querier.shuffle-sharding-ingesters-lookback-period"] = "1ns"
 
 			// Start dependencies.

integration/querier_sharding_test.go

@@ -71,7 +71,6 @@ func runQuerierShardingTest(t *testing.T, cfg querierShardingTestConfig) {
 
 	flags := mergeFlags(BlocksStorageFlags(), map[string]string{
 		"-query-frontend.cache-results":                     "true",
-		"-querier.query-ingesters-within":                   "12h", // Required by the test on query /series out of ingesters time range
 		"-query-frontend.results-cache.backend":             "memcached",
 		"-query-frontend.results-cache.memcached.addresses": "dns+" + memcached.NetworkEndpoint(e2ecache.MemcachedPort),
 		"-query-frontend.results-cache.compression":         "snappy",

integration/querier_tenant_federation_test.go

@@ -66,7 +66,6 @@ func runQuerierTenantFederationTest(t *testing.T, cfg querierTenantFederationCon
 
 	flags := mergeFlags(BlocksStorageFlags(), map[string]string{
 		"-query-frontend.cache-results":                     "true",
-		"-querier.query-ingesters-within":                   "12h", // Required by the test on query /series out of ingesters time range
 		"-query-frontend.results-cache.backend":             "memcached",
 		"-query-frontend.results-cache.memcached.addresses": "dns+" + memcached.NetworkEndpoint(e2ecache.MemcachedPort),
 		"-tenant-federation.enabled":                        "true",

integration/querier_test.go

@@ -148,8 +148,10 @@ func TestQuerierWithBlocksStorageRunningInMicroservicesMode(t *testing.T) {
 				"-blocks-storage.bucket-store.index-cache.memcached.addresses": "dns+" + memcached.NetworkEndpoint(e2ecache.MemcachedPort),
 				"-blocks-storage.bucket-store.sync-interval":                   "1s",
 				"-blocks-storage.bucket-store.index-cache.backend":             testCfg.indexCacheBackend,
+				"-blocks-storage.bucket-store.ignore-blocks-within":            "0",
 				"-blocks-storage.bucket-store.bucket-index.enabled":            strconv.FormatBool(testCfg.bucketIndexEnabled),
 				"-store-gateway.tenant-shard-size":                             fmt.Sprintf("%d", testCfg.tenantShardSize),
+				"-querier.query-store-after":                                   "0",
 				"-query-frontend.query-stats-enabled":                          "true",
 				"-query-frontend.parallelize-shardable-queries":                strconv.FormatBool(testCfg.queryShardingEnabled),
 			})
@@ -348,10 +350,12 @@ func TestQuerierWithBlocksStorageRunningInSingleBinaryMode(t *testing.T) {
 				"-blocks-storage.tsdb.block-ranges-period":                     blockRangePeriod.String(),
 				"-blocks-storage.tsdb.ship-interval":                           "1s",
 				"-blocks-storage.bucket-store.sync-interval":                   "1s",
+				"-blocks-storage.bucket-store.ignore-blocks-within":            "0",
 				"-blocks-storage.tsdb.retention-period":                        ((blockRangePeriod * 2) - 1).String(),
 				"-blocks-storage.bucket-store.index-cache.backend":             testCfg.indexCacheBackend,
 				"-blocks-storage.bucket-store.index-cache.memcached.addresses": "dns+" + memcached.NetworkEndpoint(e2ecache.MemcachedPort),
 				"-blocks-storage.bucket-store.bucket-index.enabled":            strconv.FormatBool(testCfg.bucketIndexEnabled),
+
 				// Ingester.
 				"-ingester.ring.store":           "consul",
 				"-ingester.ring.consul.hostname": consul.NetworkHTTPEndpoint(),
@@ -366,6 +370,8 @@ func TestQuerierWithBlocksStorageRunningInSingleBinaryMode(t *testing.T) {
 				"-compactor.ring.store":                           "consul",
 				"-compactor.ring.consul.hostname":                 consul.NetworkHTTPEndpoint(),
 				"-compactor.cleanup-interval":                     "2s", // Update bucket index often.
+				// Querier.
+				"-querier.query-store-after": "0",
 			})
 
 			// Start Mimir replicas.
@@ -765,10 +771,12 @@ func TestQuerierWithBlocksStorageOnMissingBlocksFromStorage(t *testing.T) {
 	// blocks (less than 3*sync-interval age) as they could be unnoticed by the store-gateway and ingesters
 	// have them anyway. We turn down the sync-interval to speed up the test.
 	storeGateway := e2emimir.NewStoreGateway("store-gateway", consul.NetworkHTTPEndpoint(), mergeFlags(flags, map[string]string{
-		"-blocks-storage.bucket-store.sync-interval": "1s",
+		"-blocks-storage.bucket-store.sync-interval":        "1s",
+		"-blocks-storage.bucket-store.ignore-blocks-within": "0",
 	}))
 	querier := e2emimir.NewQuerier("querier", consul.NetworkHTTPEndpoint(), mergeFlags(flags, map[string]string{
 		"-blocks-storage.bucket-store.sync-interval": "1s",
+		"-querier.query-store-after":                 "0",
 	}))
 	require.NoError(t, s.StartAndWaitReady(querier, storeGateway))
 

integration/query_frontend_cache_test.go

@@ -35,7 +35,6 @@ func TestQueryFrontendUnalignedQuery(t *testing.T) {
 	flags = mergeFlags(flags, map[string]string{
 		"-query-frontend.cache-results":                     "true",
 		"-query-frontend.split-queries-by-interval":         "2m",
-		"-querier.query-ingesters-within":                   "12h", // Required by the test on query /series out of ingesters time range
 		"-query-frontend.align-querier-with-step":           "true",
 		"-query-frontend.max-cache-freshness":               "0", // Cache everything.
 		"-query-frontend.results-cache.backend":             "memcached",

integration/query_frontend_test.go

@@ -169,7 +169,6 @@ func runQueryFrontendTest(t *testing.T, cfg queryFrontendTestConfig) {
 
 	flags = mergeFlags(flags, map[string]string{
 		"-query-frontend.cache-results":                     "true",
-		"-querier.query-ingesters-within":                   "12h", // Required by the test on query /series out of ingesters time range
 		"-query-frontend.results-cache.backend":             "memcached",
 		"-query-frontend.results-cache.memcached.addresses": "dns+" + memcached.NetworkEndpoint(e2ecache.MemcachedPort),
 		"-query-frontend.query-stats-enabled":               strconv.FormatBool(cfg.queryStatsEnabled),

pkg/querier/querier.go

@@ -68,7 +68,7 @@ func (cfg *Config) RegisterFlags(f *flag.FlagSet) {
 	f.BoolVar(&cfg.BatchIterators, "querier.batch-iterators", true, "Use batch iterators to execute query, as opposed to fully materialising the series in memory.  Takes precedent over the -querier.iterators flag.")
 	f.DurationVar(&cfg.QueryIngestersWithin, "querier.query-ingesters-within", 13*time.Hour, "Maximum lookback beyond which queries are not sent to ingester. 0 means all queries are sent to ingester.")
 	f.DurationVar(&cfg.MaxQueryIntoFuture, "querier.max-query-into-future", 10*time.Minute, "Maximum duration into the future you can query. 0 to disable.")
-	f.DurationVar(&cfg.QueryStoreAfter, "querier.query-store-after", 0, "The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'.")
+	f.DurationVar(&cfg.QueryStoreAfter, "querier.query-store-after", 12*time.Hour, "The time after which a metric should be queried from storage and not just ingesters. 0 means all queries are sent to store. If this option is enabled, the time range of the query sent to the store-gateway will be manipulated to ensure the query end is not more recent than 'now - query-store-after'.")
 	f.DurationVar(&cfg.ShuffleShardingIngestersLookbackPeriod, "querier.shuffle-sharding-ingesters-lookback-period", 0, "When distributor's sharding strategy is shuffle-sharding and this setting is > 0, queriers fetch in-memory series from the minimum set of required ingesters, selecting only ingesters which may have received series since 'now - lookback period'. The lookback period should be greater or equal than the configured -querier.query-store-after and -querier.query-ingesters-within. If this setting is 0, queriers always query all ingesters (ingesters shuffle sharding on read path is disabled).")
 
 	cfg.EngineConfig.RegisterFlags(f)

pkg/storage/tsdb/config.go

@@ -310,7 +310,7 @@ func (cfg *BucketStoreConfig) RegisterFlags(f *flag.FlagSet) {
 	f.DurationVar(&cfg.ConsistencyDelay, "blocks-storage.bucket-store.consistency-delay", 0, "Minimum age of a block before it's being read. Set it to safe value (e.g 30m) if your object storage is eventually consistent. GCS and S3 are (roughly) strongly consistent.")
 	f.DurationVar(&cfg.IgnoreDeletionMarksDelay, "blocks-storage.bucket-store.ignore-deletion-marks-delay", time.Hour*1, "Duration after which the blocks marked for deletion will be filtered out while fetching blocks. "+
 		"The idea of ignore-deletion-marks-delay is to ignore blocks that are marked for deletion with some delay. This ensures store can still serve blocks that are meant to be deleted but do not have a replacement yet.")
-	f.DurationVar(&cfg.IgnoreBlocksWithin, "blocks-storage.bucket-store.ignore-blocks-within", 0, "Blocks with minimum time within this duration are ignored, and not loaded by store-gateway. Useful when used together with -querier.query-store-after to prevent loading young blocks, because there are usually many of them (depending on number of ingesters) and they are not yet compacted. Negative values or 0 disable the filter.")
+	f.DurationVar(&cfg.IgnoreBlocksWithin, "blocks-storage.bucket-store.ignore-blocks-within", 10*time.Hour, "Blocks with minimum time within this duration are ignored, and not loaded by store-gateway. Useful when used together with -querier.query-store-after to prevent loading young blocks, because there are usually many of them (depending on number of ingesters) and they are not yet compacted. Negative values or 0 disable the filter.")
 	f.IntVar(&cfg.PostingOffsetsInMemSampling, "blocks-storage.bucket-store.posting-offsets-in-mem-sampling", DefaultPostingOffsetInMemorySampling, "Controls what is the ratio of postings offsets that the store will hold in memory.")
 	f.BoolVar(&cfg.IndexHeaderLazyLoadingEnabled, "blocks-storage.bucket-store.index-header-lazy-loading-enabled", true, "If enabled, store-gateway will lazy load an index-header only once required by a query.")
 	f.DurationVar(&cfg.IndexHeaderLazyLoadingIdleTimeout, "blocks-storage.bucket-store.index-header-lazy-loading-idle-timeout", 60*time.Minute, "If index-header lazy loading is enabled and this setting is > 0, the store-gateway will offload unused index-headers after 'idle timeout' inactivity.")