Releases: redpanda-data/redpanda
Releases · redpanda-data/redpanda
v21.5.6
release v21.5.6 - arm[v8] support added to release pipelines - schema registry preliminary release added - trasaction bugfixes landed - rpk container start fixes for multiple nodes on rpk container - support for `rpk` to have `-1` parameters as args for kafka api - fixed metrics double registration for archival/s3 storage - added HoneyBadger testing
v21.5.5
release v21.5.5 Bugfix release - This release will have slighly higher boot time for heavy users of compacted topics since it will discard previous indices and rebuild them using int64 logic instead of int32 logic, that caused previously integer overflow causing an infinite loop for big partitions > 4GB files - Fix stale readers in the cache - Adds an advaned technique for fetching partition data that materializes topic metdata on a core-local basis for making a parallel plan of execution. We borrowed this technique from databases in the requqest-parse-plan-execute model for reading remote core streaming partitions. This has a 10-100x lower latency impact and improves throughput just a littlebit.
v21.5.4
release v21.5.4 Minor release - add `rpk container start --set "..arbitrary flags"` which eliminates the need for a configuration file - fixes for transaction fencing - added major compaction-controller rework to achieve steady state of compaction without affecting foreground work in a constant learning loop. Compaction automatically adjusts to the foreground work and otherwise learns to keep up with the compacted segment work This should eliminate connection timeouts on large CPU and IO hogs for workloads with thousands of compacted partitions.
v21.5.3
release v21.5.3 Minor release: - add s3 telemetry to client code - fix rpk start command
v21.5.2
release v21.5.2 - v0 produce support for cockroachdb cdc - rpk start --set for inline config txforms - Wasm stability fixes - additional performance improvements in GCS storage - ipv6 support for GCS - acls support - optional prom setting for k8s operator - archival telemetry now exposed via prometheus - initial transactions support in beta. GA in 2 weeks or so - move to autogenerated fetch as the last hand written method. now all is codegen correctly
v21.5.2-beta3
release v21.5.2-beta3 - fix GCS ipv6 resolution - fix GCS stalls on compacted topics - add infrastructure for embeded schema registry
v21.5.1
release v21.5.1 Features - enable http proxy in k8s with mTLS - enable large batch reads with 10x latency reduction - rpk acls support Bugs: - fix append entries gate closed - improve read latency for large batch latency - connection errors with SASL TLS and http proxy - sassl handshake version 0 support for Kafka Python client - wasm deregistration of topics Non features: - major progress in transaction support, missing 1 pr - rpk cloud is now connecting to vectorized cloud - fetch support is now codegen as well - last feature
v21.4.15
release v21.4.15 - overlay kafka::offset index so mirror-maker2 "just works" - add queue-depth backpressure algorithm loosely based on the PARDA latency-based queue control https://www.usenix.org/legacy/event/fast09/tech/full_papers/gulati/gulati_html/index.html - add external connectivity to GKE - controlled upgrade procedure on kubernetes for redpanda containers
v21.4.14
release v21.4.14 - exposes auto create topics in k8s - fixes Wasm script management bugs - fixes telemetry for raft leadership dashboard
v21.4.13
release v21.4.13 - Initial Wasm support - HTTP JSON support in addition to kafka binary JSON