- 11月 19, 2022
-
-
由 Joel Hamill 创作于
-
由 Alex Cheng 创作于
* DGS-5318 load kafka policy for SDS for Topic ReadCatalog operations * DGS-5318 updated sr rbac role definitions to include Topic Read/WriteCatalog operations
-
由 Ian McDonald 创作于
* add extended timeout * ssh timeout * remove ssh timeout
-
由 Ning Shan 创作于
* Set strategy IGNORE for demoted broker in SBC
-
由 Brian Bushree 创作于
-
由 Mathew Hogan 创作于
Create KRaftNodeStartUpEvent and implement serialization/deserialization Reviewer: @alok123t
-
由 Lingnan Liu 创作于
* MINOR: Enable GoalOptimizationLogger * Refactor constructor of GoalOptimizationLogger
-
由 Justine Olshan 创作于
For now we plan to update this to be a percentage while we are waiting on the sbc centralized component to limit demotions. We will either demote one broker or more if the configured percentage of brokers in the cluster is greater than 1.
-
由 Alok Thatikunta 创作于
This PR introduces the snapshot capability for the internal Tier topic. This is based on https://confluentinc.atlassian.net/wiki/spaces/KSTORAGE/pages/2549122180/Tier+Topic+Snapshots - Adds TierTopicSnapshotCoordinator, startup/shutdown routine as part of BrokerServer, KafkaServer. A single instance exists for each cluster, which runs on a leader broker for internal tier topic partition 1 (_confluent-tier-state-1) - Adds TierTopicSnapshotManager which is responsible for taking continuous snapshots. This is started as part of coordinator - More new Kafka config for max records per snapshot, look up retention hours - Adds EpochAndSeqNumber and SequencedObject as cloud utils for uniquely identifying a snapshot - Adds LinkedCloudObject as a cloud util for ordering snapshot objects - Support for dynamic config - Helper utility TierTopicSnapshotObject to encode/decode snapshot naming - Helper utility TierTopicSnapshotObjectStoreUtils for snapshot related object store calls, this implements an exponential backoff Updates to restore - Introduces TierTopicSnapshotMetadataComparator for sorting snapshots based on epoch, seqNumber, endTimestamp and startTimestamp Minor - Updates TierTopicPartitionSnapshot to fix previous offsets - Adds a util class to read the entire InputStream to a byte array `readFullyToArray`, refactors TierStateFetcher, CompactionTask to use this method - Refactors TierTopicSnapshotMetadata to use TierTopicSnapshotObject Tests - Adds TierTopicSnapshotCoordinatorTest for TierTopicSnapshotCoordinator - Adds TierTopicSnapshotManagerTest for TierTopicSnapshotManager - Adds LinkedCloudObjectTest for LinkedCloudObject - Minor additions to TierTopicPartitionSnapshotTest reviewers: @junrao @rohitshekhar29 @mathewdhogan @zhengyd2014
- 11月 18, 2022
-
-
由 Sanjana Kaundinya 创作于
-
由 Stanislav Kozlovski 创作于
abb7c286 introduced a new ProposalGenerator class used by KafkaCruiseControl to generate proposals. The GoalViolationsIntegrationTest wasn't updated to mock the return value from that class
-
由 Lingnan Liu 创作于
* Integrate GoalOptmizationHistory with E2E balancing workflow * Implement movementReason populating * Added unit tests * Decouple self-healing and goal optimization history tracking * Don't suspend offline replicas * Address comments * Add suspended topic partitions in OptimizerResult logging * Handle merge conflicts
-
由 Lingnan Liu 创作于
* Add GoalOptimizationLogger * Add info logging on suspended topic partition, convert full movement history to debug logging * Code refactor to address comments * Added logging on remaining duration of suspension
-
由 Jason Gustafson 创作于
(Cherry-pick of apache/kafka#12819 for confluent/master. No conflicts.) For tests which use the console consumer service, we are currently enabling TRACE logging by default. I have seen some system tests where this produces GBs of logging. A better default is probably DEBUG. Reviewers: José Armando García Sancio <jsancio@apache.org>
-
由 Georgi 创作于
Skip soft goal execution when incremental balancing is enabled and the hard goals prior to the soft goal generated partition movements.
-
由 Justine Olshan 创作于
This is the first step to enabling network health manager. It will run and collect metrics but mitigation is not allowed to be on for kraft clusters. Added a test to validate that mitigation can not be enabled.
-
由 Patrick Sullivan 创作于
* add CellLoadResolver interface and cell dto * stub admin client api * add default CellLoadResolver and unit testing * checkstyle * comment and add api for metric-granularity loads * checkstyle * make Metric public for build
- 11月 17, 2022
-
-
由 Honshu Priyadarshi 创作于
Limit the CPU usage and throttle the temporary disk write throughput at 10mb/s. This change removes the sustained throughput of 10mb/s of temp disk write and help in case of long running DA audit in high risk clusters. Also, it include the fix for issue discovered in prod which causes false durability alert. The fix is for DA use stream cloud object in-memory and filter the keys instead of writing object to disk
-
由 Lingnan Liu 创作于
-
由 Lingnan Liu 创作于
MINOR: Add log when suspended topic partition has a higher history epoch than goal optimization history (#8082)
-
由 Ning Shan 创作于
* MovementExclusionGoal should honor eligible source or destination
-
由 Patrick Sullivan 创作于
-
由 Sanjana Kaundinya 创作于
- 11月 16, 2022