NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
Go modules · #237 by repository stars
Last release 4 days ago
28 Sep 2026
Ships on a steady schedule
a new release about every 9 days
Nearly every release is documented
notes for 59 of the last 60 stable releases
Nothing withdrawn
no release was ever pulled
9 years old
1521 releases · first in 2018
Nothing published for this version
Nothing published for this version
Nothing published for this version
One column per quarter.
- #3700 Compact/Web: Make old bucket viewer UI work with vanilla Prometheus blocks. - #3657 \*: It's now possible to configure HTTP transport options
--block-meta-fetch-concurrency allowing to configure number of go routines for block metadata synchronization.--query-range.request-downsampled flag enabling additional queries for downsampled data in case of empty or incomplete response to range request.--tsdb.allow-overlapping-blocks flag to allow overlapping tsdb blocks and enable vertical compaction.--query.default-step flag to set default step. Useful when your tenant scrape interval is stable and far from default UI's 1s.--hash-func. If some function has been specified, writers calculate hashes using that function of each file in a block before uploading them. If those hashes exist in the meta.json file then Compact does not download the files if they already exist on disk and with the same hash. This also means that the data directory passed to Thanos Compact is only cleared once at boot or if everything succeeds. So, if you, for example, use persistent volumes on k8s and your Thanos Compact crashes or fails to make an iteration properly then the last downloaded files are not wiped from the disk. The directories that were created the last time are only wiped again after a successful iteration or if the previously picked up blocks have disappeared.--store.enable-index-header-lazy-reader is enabled.--store.enable-index-header-lazy-reader is enabled.Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
### Added - #3380 Mixin: Add block deletion panels for compactor dashboards. - #3568 Store: Optimized inject label stage of index lookup. - #3566 Stor
hints field to LabelNamesRequest and LabelValuesRequest. Hints are an opaque data structure that can be used to carry additional information from the store and its content is implementation-specific.thanos tools bucket rewrite command allowing to delete series from given block.--receive.hashrings alternative to receive.hashrings-file flag (lower priority). The flag expects the literal hashring configuration in JSON format.thanos_objstore_bucket_operation_failures_total alert.no such host error when using dnssrv+ or dnssrvnoa+.segments/ if the file is bigger than 1GB. To change the defaults see the docs.meta.json file doesn't cause block cleanup anymore. This has a potential to generate corrupted blocks under specific conditions. Partial block is left in bucket for later cleanup.Nothing published for this version
### Fixed - #3532 compact: do not cleanup blocks on boot. Reverts the behavior change introduced in #3115 as in some very bad cases the boot of Thanos
### Fixed - #3480 Query Frontend: Fixed regression. - #3734 pkg/rules/proxy: fix hotlooping when receiving client errors ### Changed - #3498 Enabled d
This, generally, reduces baseline memory usage of store when inactive, as well as a total number of mapped files (which is limited to 64k in some syst
--compact.cleanup-interval time period. By default it is 5 minutes.--query.http-method is set GET.query-frontend.org-id-header flag to specify HTTP header(s) to populate slow query log (e.g. X-Grafana-User).--store.enable-index-header-lazy-reader flag, the store-gateway will load into memory an index-header only once it's required at query time. Index-header will be automatically released after --store.index-header-lazy-reader-idle-timeout of inactivity.
hints field to LabelNamesResponse and LabelValuesResponse. Hints in an opaque data structure that can be used to carry additional information from the store and its content is implementation specific.
thanos tools bucket mark command that allows to mark given block for deletion or for no-compactexperimental.enable-index-cache-postings-compression flag.thanos_compactor_blocks_marked_for_deletion_total to thanos_compactor_blocks_marked_total with marker label. Compactor will now automatically disable compaction for blocks with large index that would output blocks after compaction larger than specified value (by default: 64GB). This automatically handles the Promethus format limit.thanos_replicate_origin_.* metrics.
thanos_replicate_origin_meta_loads_total can be replaced by blocks_meta_synced{state="loaded"}.thanos_replicate_origin_partial_meta_reads_total can be replaced by blocks_meta_synced{state="failed"}.thanos_compactor to thanos_compact, thanos_querier to thanos_query and thanos_ruler to thanos_rule.Nothing published for this version
New Thanos component, Query Frontend has more options and supports shared cache (currently: Memcached).
Highlights:
storeMatch[] arguments with __address__ matchers.--web.prefix-header is used.web.external-prefix for Compactor UI.log_queries_longer_than to log-queries-longer-than.storeMatch[] parameter to queries.storeMatch[] to Labels APIs; Time range metadata based store filtering is supported on Labels APIs.thanos_sidecar_prometheus_store_received_frames histogram metric.query.metadata.default-time-range flag to specify the default metadata time range duration for retrieving labels through Labels and Series API when the range parameters are not specified. The zero value means range covers the time since the beginning.cache-compression-type flag to use compression in the query frontend cache./debug/fgprof endpoint on HTTP port allowing to get off-CPU profiles as well.Cache-Control HTTP response header which controls caching behaviour. So far no-store value is supported and it makes the response skip cache.tools bucket cleanup CLI tool that deletes all blocks marked to be deleted.thanos_sidecar_reloader_config_apply_operations_total and rename metric thanos_sidecar_reloader_config_apply_errors_total to thanos_sidecar_reloader_config_apply_operations_failed_total.thanos_query_gate_queries_max. Remove metric thanos_query_concurrent_selects_gate_queries_in_flight.thanos_bucket_store_queries_concurrent_max to thanos_bucket_store_series_gate_queries_max.thanos_objstore_bucket_operation_failures_total.thanos_sidecar_reloader_config_apply_operations_total and rename metric thanos_sidecar_reloader_config_apply_errors_total to thanos_sidecar_reloader_config_apply_operations_failed_total.Nothing published for this version
Nothing published for this version
Added new Thanos component: Query Frontend responsible for response caching, query scheduling and parallelization (based on Cortex Query Frontend).
Highlights:
--web.external-prefix and --web.route-prefix.--receive.local-endpoint.thanos_compact_downsample_total metric for downsample of 5m resolution blocks.--store.grpc.series-sample-limit implementation. The limit is now applied to the sum of all samples fetched across all queried blocks via a single Series call, instead of applying it individually to each block./api/v1/rules is called and the time zone is not UTC; rules field is an empty array now if no rules have been defined in a rule group. Thanos Rule's /api/v1/rules endpoint no longer returns the old, deprecated partial_response_strategy. The old, deprecated value has been fixed to WARN for quite some time. Please use partialResponseStrategy.maxSourceResolution when auto downsampling is enabled.--query.lookback-delta flag to override the default lookback delta in PromQL. The flag should be lookback delta should be set to at least 2 times of the slowest scrape interval. If unset it will use the PromQL default of 5m./api/v1/status/flags, /api/v1/status/runtimeinfo and /api/v1/status/buildinfo) are now available on all components with a HTTP API.start and end fields to Store API LabelNamesRequest and LabelValuesRequest.reloader_config_apply_errors_total metric. Add new flags --reloader.watch-interval, and --reloader.retry-interval.thanos_bucket_store_cached_series_fetch_duration_seconds and thanos_bucket_store_cached_postings_fetch_duration_seconds.storeMatch[] to select matching stores when debugging the querier. See documentationthanos_bucket_store_cached_postings_compression_time_seconds to thanos_bucket_store_cached_postings_compression_time_seconds_total.--tsdb.no-lockfile) to override behavior.operation label value getrange changed to get_range for thanos_store_bucket_cache_operation_requests_total and thanos_store_bucket_cache_operation_hits_total to be consistent with bucket operation metrics.encrypt_sse: true option is now deprecated in favour of sse_config. If you used encrypt_sse, the migration strategy is to set up the following block:sse_config:
type: SSE-S3
Nothing published for this version
Nothing published for this version
TSDB now does memory-mapping of Head chunks and reduces memory usage.
tsdb.MultiError./api/v1/stores now guarantees to return a string in the lastError field.index.cache.json. The hidden flag --store.disable-index-header was removed.chunk_object_size_ttl to chunk_object_attrs_ttl.--index.generate-missing-cache-file and the metric thanos_compact_generated_index_total were removed.--resolution is now in Go duration format.bucket tool bucket verify --id-whitelist flag to --id.--compaction and --resolution flags.--block-viewer.global.sync-block-interval to configure metadata sync interval for the bucket UI.Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Nothing published for this version
Your coding agent can read these notes before it upgrades. Set up the MCP server →