You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
v0.30 brings many important fixes & optimizations to compaction, store gateway, receive replication and querying. Make sure to try the new PromQL engine which is more & more efficient every week.
NOTE: Querier's query.promql-engine flag enabling the new PromQL engine is now unhidden. We encourage users to use new experimental PromQL engine for efficiency reasons.
Furthermore, we recommend you use Redis as a caching client (if you use store GW or query frontend caching) and Ketama algorithm as receiver hashing algorithm ( --receive.hashrings-algorithm=ketama - introducing consistent hashing to receiver).
Changes
Fixed
#5716 DNS: Fix miekgdns resolver LookupSRV to work with CNAME records.
#5844 Query Frontend: Fixes @ modifier time range when splitting queries by interval.
#5854 Query Frontend: lookback_delta param is now handled in query frontend.
#5860 Query: Fixed bug of not showing query warnings in Thanos UI.
#5856 Store: Fixed handling of debug logging flag.
#5230 Rule: Stateless ruler support restoring for state from query API servers. The query API servers should be able to access the remote write storage.
#5880 Query Frontend: Fixes some edge cases of query sharding analysis.
#5893 Cache: Fixed redis client not respecting SetMultiBatchSize config value.
#5966 Query: Stop relying on non-existent hints for mint and maxt when selecting series for the api/v1/series HTTP endpoint.
#5910: Receive: Fixed ketama quorum bug that was could cause success response for failed replication. This also optimize heavily receiver CPU use.
Added
#5814 Store: Added metric thanos_bucket_store_postings_size_bytes that shows the distribution of how many postings (in bytes) were needed for each Series() call in Thanos Store. Useful for determining limits.
#5703 StoreAPI: Added hash field to series' chunks. Store gateway and receive implements that field and proxy leverage that for quicker deduplication.
#5801 Store: Added a new flag --store.grpc.downloaded-bytes-limit that limits the number of bytes downloaded in each Series/LabelNames/LabelValues call. Use thanos_bucket_store_postings_size_bytes for determining the limits.
#5836 Receive: Added hidden flag tsdb.memory-snapshot-on-shutdown to enable experimental TSDB feature to snapshot on shutdown. This is intended to speed up receiver restart.
#5839 Receive: Added parameter --tsdb.out-of-order.time-window to set time window for experimental out-of-order samples ingestion. Disabled by default (set to 0s). Please note if you enable this option and you use compactor, make sure you set the --enable-vertical-compaction flag, otherwise you might risk compactor halt.
#5889 Query Frontend: Added support for vertical sharding label_replace and label_join functions.
#5819 Store: Added a few objectives for Store's data summaries (touched/fetched amount and sizes). They are: 50, 95, and 99 quantiles.
#5837 Store: Added streaming retrival of series from object storage.
#5940 Objstore: Support for authenticating to Swift using application credentials.
#5945 Tools: Added new no-downsample marker to skip blocks when downsampling via thanos tools bucket mark --marker=no-downsample-mark.json. This will skip downsampling for blocks with the new marker.
#5977 Tools: Added remove flag on bucket mark command to remove deletion, no-downsample or no-compact markers on the block
Changed
#5785 Query: thanos_store_nodes_grpc_connections now trimms external_labels label name longer than 1000 character. It also allows customizations in what labels to preserve using query.conn-metric.label flag.
#5542 Mixin: Added query concurrency panel to Querier dashboard.
#5909 Receive: Compact tenant head after no appends have happened for 1.5 tsdb.max-block-size.
#5593 Cache: Switched Redis client to Rueidis. Rueidis is faster and provides client-side caching. It is highly recommended to use it so that repeated requests for the same key would not be needed.
#5896 *: Upgraded Prometheus to v0.40.7 without implementing native histogram support. Querying native histograms will fail with Error executing query: invalid chunk encoding "<unknown>" and native histograms in write requests are ignored.
#5838 Mixin: Added data touched type to Store dashboard.
#5922 Compact: Retry on clean, partial marked errors when possible.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
v0.30 brings many important fixes & optimizations to compaction, store gateway, receive replication and querying. Make sure to try the new PromQL engine which is more & more efficient every week.
Furthermore, we recommend you use Redis as a caching client (if you use store GW or query frontend caching) and Ketama algorithm as receiver hashing algorithm (
--receive.hashrings-algorithm=ketama- introducing consistent hashing to receiver).Changes
Fixed
lookback_deltaparam is now handled in query frontend.forstate from query API servers. The query API servers should be able to access the remote write storage.SetMultiBatchSizeconfig value.api/v1/seriesHTTP endpoint.chunks_fetched_durationwrong calculation.Added
thanos_bucket_store_postings_size_bytesthat shows the distribution of how many postings (in bytes) were needed for each Series() call in Thanos Store. Useful for determining limits.hashfield to series' chunks. Store gateway and receive implements that field and proxy leverage that for quicker deduplication.--store.grpc.downloaded-bytes-limitthat limits the number of bytes downloaded in each Series/LabelNames/LabelValues call. Usethanos_bucket_store_postings_size_bytesfor determining the limits.tsdb.memory-snapshot-on-shutdownto enable experimental TSDB feature to snapshot on shutdown. This is intended to speed up receiver restart.--tsdb.out-of-order.time-windowto set time window for experimental out-of-order samples ingestion. Disabled by default (set to 0s). Please note if you enable this option and you use compactor, make sure you set the--enable-vertical-compactionflag, otherwise you might risk compactor halt.label_replaceandlabel_joinfunctions.no-downsamplemarker to skip blocks when downsampling viathanos tools bucket mark --marker=no-downsample-mark.json. This will skip downsampling for blocks with the new marker.Changed
thanos_store_nodes_grpc_connectionsnow trimmsexternal_labelslabel name longer than 1000 character. It also allows customizations in what labels to preserve usingquery.conn-metric.labelflag.tsdb.max-block-size.Error executing query: invalid chunk encoding "<unknown>"and native histograms in write requests are ignored.Removed
ThanosReceiveTrafficBelowThresholdalert.New Contributors
Commits
query.promql-engineflag; added info in docs. by @bwplotka in querier: Unhidequery.promql-engineflag; added info in docs. #5913efficientgo/core/testutilfor tests by @saswatamcode in Useefficientgo/core/testutilfor tests #5933Full Changelog: v0.29.0...v0.30.0 🎉
This discussion was created from the release v0.30.0.
Beta Was this translation helpful? Give feedback.
All reactions