update docs

Signed-off-by: Ben Ye <benye@amazon.com>

Author: yeya24

CHANGELOG.md

@@ -9,6 +9,7 @@
 * [CHANGE] Ruler: Remove `cortex_ruler_write_requests_total`, `cortex_ruler_write_requests_failed_total`, `cortex_ruler_queries_total`, `cortex_ruler_queries_failed_total`, and `cortex_ruler_query_seconds_total` metrics for the tenant when the ruler deletes the manager for the tenant. #5772
 * [CHANGE] Main: Mark `mem-ballast-size-bytes` flag as deprecated. #5816
 * [CHANGE] Querier: Mark `-querier.ingester-streaming` flag as deprecated. Now query ingester streaming is always enabled. #5817
+* [CHANGE] Compactor/Bucket Store: Added `-blocks-storage.bucket-store.block-discovery-strategy` to configure different block listing strategy. Reverted the current recursive block listing mechanism and use the strategy `Concurrent` as in 1.15. #5828
 * [FEATURE] Ingester: Add per-tenant new metric `cortex_ingester_tsdb_data_replay_duration_seconds`. #5477
 * [FEATURE] Query Frontend/Scheduler: Add query priority support. #5605
 * [FEATURE] Tracing: Add `kuberesolver` to resolve endpoints address with `kubernetes://` prefix as Kubernetes service. #5731

docs/blocks-storage/querier.md

@@ -1314,6 +1314,17 @@ blocks_storage:
       # CLI flag: -blocks-storage.bucket-store.bucket-index.max-stale-period
       [max_stale_period: <duration> | default = 1h]
 
+    # One of concurrent, recursive, bucket_index. When set to concurrent, stores
+    # will concurrently issue one call per directory to discover active blocks
+    # in the bucket. The recursive strategy iterates through all objects in the
+    # bucket, recursively traversing into each directory. This avoids N+1 calls
+    # at the expense of having slower bucket iterations. bucket_index strategy
+    # can be used in Compactor only and utilizes the existing bucket index to
+    # fetch block IDs to sync. This avoids iterating the bucket but can be
+    # impacted by delays of cleaner creating bucket index.
+    # CLI flag: -blocks-storage.bucket-store.block-discovery-strategy
+    [block_discovery_strategy: <string> | default = "concurrent"]
+
     # 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.
     # CLI flag: -blocks-storage.bucket-store.max-chunk-pool-bytes

docs/blocks-storage/store-gateway.md

@@ -1423,6 +1423,17 @@ blocks_storage:
       # CLI flag: -blocks-storage.bucket-store.bucket-index.max-stale-period
       [max_stale_period: <duration> | default = 1h]
 
+    # One of concurrent, recursive, bucket_index. When set to concurrent, stores
+    # will concurrently issue one call per directory to discover active blocks
+    # in the bucket. The recursive strategy iterates through all objects in the
+    # bucket, recursively traversing into each directory. This avoids N+1 calls
+    # at the expense of having slower bucket iterations. bucket_index strategy
+    # can be used in Compactor only and utilizes the existing bucket index to
+    # fetch block IDs to sync. This avoids iterating the bucket but can be
+    # impacted by delays of cleaner creating bucket index.
+    # CLI flag: -blocks-storage.bucket-store.block-discovery-strategy
+    [block_discovery_strategy: <string> | default = "concurrent"]
+
     # 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.
     # CLI flag: -blocks-storage.bucket-store.max-chunk-pool-bytes

docs/configuration/config-file-reference.md

@@ -1850,6 +1850,17 @@ bucket_store:
     # CLI flag: -blocks-storage.bucket-store.bucket-index.max-stale-period
     [max_stale_period: <duration> | default = 1h]
 
+  # One of concurrent, recursive, bucket_index. When set to concurrent, stores
+  # will concurrently issue one call per directory to discover active blocks in
+  # the bucket. The recursive strategy iterates through all objects in the
+  # bucket, recursively traversing into each directory. This avoids N+1 calls at
+  # the expense of having slower bucket iterations. bucket_index strategy can be
+  # used in Compactor only and utilizes the existing bucket index to fetch block
+  # IDs to sync. This avoids iterating the bucket but can be impacted by delays
+  # of cleaner creating bucket index.
+  # CLI flag: -blocks-storage.bucket-store.block-discovery-strategy
+  [block_discovery_strategy: <string> | default = "concurrent"]
+
   # 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.
   # CLI flag: -blocks-storage.bucket-store.max-chunk-pool-bytes

pkg/storage/tsdb/config.go

@@ -2,7 +2,6 @@ package tsdb
 
 import (
 	"flag"
-	"github.com/cortexproject/cortex/pkg/util"
 	"path/filepath"
 	"strings"
 	"time"
@@ -15,6 +14,7 @@ import (
 	"github.com/thanos-io/thanos/pkg/store"
 
 	"github.com/cortexproject/cortex/pkg/storage/bucket"
+	"github.com/cortexproject/cortex/pkg/util"
 )
 
 const (