Cortex - v1.15.0

This release contains 177 contributions from 24 contributors. We also have 13 new contributors. Thank you all for the contribution!

Some notable changes release are:
* Out of order samples ingestion
* MultiKey KV ring for DynamoDB
* Snappy-block gRPC compression
* Redis as index cache and caching bucket backend
* Arm images support
* Thanos PromQL engine support
* Multiples performance improvements and bug fixes

Cortex

• [CHANGE] Storage: Make Max exemplars config per tenant instead of global configuration. #5080 #5122
• [CHANGE] Alertmanager: Local file disclosure vulnerability in OpsGenie configuration has been fixed. #5045
• [CHANGE] Rename oltp_endpoint to otlp_endpoint to match opentelemetry spec and lib name. #5068
• [CHANGE] Distributor/Ingester: Log warn level on push requests when they have status code 4xx. Do not log if status is 429. #5103
• [CHANGE] Tracing: Use the default OTEL trace sampler when -tracing.otel.exporter-type is set to awsxray. #5141
• [CHANGE] Ingester partial error log line to debug level. #5192
• [CHANGE] Change HTTP status code from 503/422 to 499 if a request is canceled. #5220
• [CHANGE] Store gateways summary metrics have been converted to histograms cortex_bucket_store_series_blocks_queried, cortex_bucket_store_series_data_fetched, cortex_bucket_store_series_data_size_touched_bytes, cortex_bucket_store_series_data_size_fetched_bytes, cortex_bucket_store_series_data_touched, cortex_bucket_store_series_result_series #5239
• [FEATURE] Querier/Query Frontend: support Prometheus /api/v1/status/buildinfo API. #4978
• [FEATURE] Ingester: Add active series to all_user_stats page. #4972
• [FEATURE] Ingester: Added -blocks-storage.tsdb.head-chunks-write-queue-size allowing to configure the size of the in-memory queue used before flushing chunks to the disk . #5000
• [FEATURE] Query Frontend: Log query params in query frontend even if error happens. #5005
• [FEATURE] Ingester: Enable snapshotting of In-memory TSDB on disk during shutdown via -blocks-storage.tsdb.memory-snapshot-on-shutdown. #5011
• [FEATURE] Query Frontend/Scheduler: Add a new counter metric cortex_request_queue_requests_total for total requests going to queue. #5030
• [FEATURE] Build ARM docker images. #5041
• [FEATURE] Query-frontend/Querier: Create spans to measure time to merge promql responses. #5041
• [FEATURE] Querier/Ruler: Support the new thanos promql engine. This is an experimental feature and might change in the future. #5093
• [FEATURE] Added zstd as an option for grpc compression #5092
• [FEATURE] Ring: Add new kv store option dynamodb. #5026
• [FEATURE] Cache: Support redis as backend for caching bucket and index cache. #5057
• [FEATURE] Querier/Store-Gateway: Added -blocks-storage.bucket-store.ignore-blocks-within allowing to filter out the recently created blocks from being synced by queriers and store-gateways. #5166
• [FEATURE] AlertManager/Ruler: Added support for keep_firing_for on alerting rulers.
• [FEATURE] Alertmanager: Add support for time_intervals. #5102
• [FEATURE] Added snappy-block as an option for grpc compression #5215
• [FEATURE] Enable experimental out-of-order samples support. Added 2 new configs ingester.out_of_order_time_window and blocks-storage.tsdb.out_of_order_cap_max. #4964
• [ENHANCEMENT] Querier: limit series query to only ingesters if start param is not specified. #4976
• [ENHANCEMENT] Query-frontend/scheduler: add a new limit frontend.max-outstanding-requests-per-tenant for configuring queue size per tenant. Started deprecating two flags -query-scheduler.max-outstanding-requests-per-tenant and -querier.max-outstanding-requests-per-tenant, and change their value default to 0. Now if both the old flag and new flag are specified, the old flag's queue size will be picked. #4991
• [ENHANCEMENT] Query-tee: Add /api/v1/query_exemplars API endpoint support. #5010
• [ENHANCEMENT] Let blocks_cleaner delete blocks concurrently(default 16 goroutines). #5028
• [ENHANCEMENT] Query Frontend/Query Scheduler: Increase upper bound to 60s for queue duration histogram metric. #5029
• [ENHANCEMENT] Query Frontend: Log Vertical sharding information when query_stats_enabled is enabled. #5037
• [ENHANCEMENT] Ingester: The metadata APIs should honour querier.query-ingesters-within when querier.query-store-for-labels-enabled is true. #5027
• [ENHANCEMENT] Query Frontend: Skip instant query roundtripper if sharding is not applicable. #5062
• [ENHANCEMENT] Push reduce one hash operation of Labels. #4945 #5114
• [ENHANCEMENT] Alertmanager: Added -alertmanager.enabled-tenants and -alertmanager.disabled-tenants to explicitly enable or disable alertmanager for specific tenants. #5116
• [ENHANCEMENT] Upgraded Docker base images to alpine:3.17. #5132
• [ENHANCEMENT] Add retry logic to S3 bucket client. #5135
• [ENHANCEMENT] Update Go version to 1.20.1. #5159
• [ENHANCEMENT] Distributor: Reuse byte slices when serializing requests from distributors to ingesters. #5193
• [ENHANCEMENT] Query Frontend: Add number of chunks and samples fetched in query stats. #5198
• [ENHANCEMENT] Implement grpc.Compressor.DecompressedSize for snappy to optimize memory allocations. #5213
• [ENHANCEMENT] Querier: Batch Iterator optimization to prevent transversing it multiple times query ranges steps does not overlap. #5237
• [BUGFIX] Updated golang.org/x/net dependency to fix CVE-2022-27664. #5008
• [BUGFIX] Fix panic when otel and xray tracing is enabled. #5044
• [BUGFIX] Fixed no compact block got grouped in shuffle sharding grouper. #5055
• [BUGFIX] Fixed ingesters with less tokens stuck in LEAVING. #5061
• [BUGFIX] Tracing: Fix missing object storage span instrumentation. #5074
• [BUGFIX] Ingester: Fix Ingesters returning empty response for metadata APIs. #5081
• [BUGFIX] Ingester: Fix panic when querying metadata from blocks that are being deleted. #5119
• [BUGFIX] Ring: Fix case when dynamodb kv reaches the limit of 25 actions per batch call. #5136
• [BUGFIX] Query-frontend: Fix shardable instant queries do not produce sorted results for sort, sort_desc, topk, bottomk functions. #5148, #5170
• [BUGFIX] Querier: Fix /api/v1/series returning 5XX instead of 4XX when limits are hit. #5169
• [BUGFIX] Compactor: Fix issue that shuffle sharding planner return error if block is under visit by other compactor. #5188
• [BUGFIX] Fix S3 BucketWithRetries upload empty content issue #5217
• [BUGFIX] Query Frontend: Disable absent, absent_over_time and scalar for vertical sharding. #5221
• [BUGFIX] Catch context error in the s3 bucket client. #5240
• [BUGFIX] Fix query frontend remote read empty body. #5257
• [BUGFIX] Fix query frontend incorrect error response format at SplitByQuery middleware. #5260