- 10月 15, 2022
-
-
由 Michael Royster 创作于
use JDK 17 node label if for pull request
-
由 Mathew Hogan 创作于
Reviewer: @alok123t
- 10月 14, 2022
-
-
由 Stanislav Kozlovski 创作于
This PR merges in the 4 changes that introduce the ProduceInboundCapacityGoal to SBC: - 70bdcebf1f3689d3306b7ae178c02660d5a85e81 - the PRODUCE_IN resource This patch adds the Resource.PRODUCE_IN resource which will be a subset of Resource.NW_IN but only applying to the producer's bytes in in the broker It also adds two new configs: - `producer.inbound.capacity.threshold` - `producer.in.max.bytes.per.second` to help configure the maximum produce in bytes a broker can receive - b79752d410b945038fc4c5f27067dc729e0aa326 - extends the ComputeEvenClusterLoadPlan API to support the resource This patch extends the ComputeEvenClusterLoadPlan Admin API to expose a goal's PRODUCE_IN resource stats - f4129f2c41d5c3e5ed1bb7ed6627d6bf040354e0 - adds the kafka metric definition This patch adds a new KafkaMetricDef associated with Resource.PRODUCE_IN -> PRODUCER_BYTES_IN. Both it and the old associated-with Resource.NW_IN metric LEADER_BYTES_IN are mapped to the RawMetricType.TOPIC_BYTES_IN through a small-refactor introducing a RawMetricTypeMapper class instead of the traditional TYPE_TO_DEF map. We additionally add tests for the proper population of this new resource/metric in SamplingUtils/RawMetricsHolderTest/LoadTest, as well as populate the resource in the RandomCluster tests. - 5107a0c82629bb6db9c4cb9a97eba898bbb7f929 adds the goal and tests This patch adds the ProducerInboundCapacity goal, and configures is as part of the default goals. It additionally commits a property-based test for said goal too. There are a fair amount of changes in the PBT generation logic as was required to ensure leaders get more properly distributed, especially for excluded topics.
-
由 Honshu Priyadarshi 创作于
* KDATA-318: Implement the cloud API for DA offset map. Wrote get/put/delete api for offset-map file.
-
由 Alyssa Huang 创作于
-
由 Jason Gustafson 创作于
(cherry-pick of https://github.com/apache/kafka/pull/12741 for ce-kafka/v0.4105.x.) When the `BrokerServer` starts its shutting down process, it transitions to `SHUTTING_DOWN` and sets `isShuttingDown` to `true`. With this state change, the follower state changes are short-cutted. This means that a broker which was serving as leader would remain acting as a leader until controlled shutdown completes. Instead, we want the leader and ISR state to be updated so that requests will return NOT_LEADER and the client can find the new leader. We missed this case while implementing https://github.com/apache/kafka/pull/12187. This patch fixes the issue and updates an existing test to ensure that `isShuttingDown` has not effect. We should consider adding integration tests for this as well. We can do this separately. CONFLUENT CHERRY-PICK NOTES: Conflicts: core/src/main/scala/kafka/server/ReplicaManager.scala core/src/test/scala/unit/kafka/server/ReplicaManagerTest.scala In ce-kafka, there is no `shuttingDown` field, so we modified the patch to use the `BrokerState` instead. Reviewers: David Jacot <djacot@confluent.io>
-
由 Jason Gustafson 创作于
Similar to https://github.com/apache/kafka/commit/a8fcbcc08f9d187dcf4c630fe3e639c291545756 upstream, change `displayGranularity` to 0 for property tests as well. This ensures that the gradle output shows the executor that was used for each test, which helps when debugging unexpected system exits during the gradle run. Reviewers: Niket <niket-goel@users.noreply.github.com>
-
由 Milo Simpson 创作于
- Remove old KsqlDeveloperWrite/KsqlResourceOwner roles - KsqlAdmin can navigate UI and has AccessWithToken - update Scope class to support Ksql
-
由 Jason Gustafson 创作于
(cherry-pick of https://github.com/apache/kafka/pull/12741 for ce-kafka/master.) When the `BrokerServer` starts its shutting down process, it transitions to `SHUTTING_DOWN` and sets `isShuttingDown` to `true`. With this state change, the follower state changes are short-cutted. This means that a broker which was serving as leader would remain acting as a leader until controlled shutdown completes. Instead, we want the leader and ISR state to be updated so that requests will return NOT_LEADER and the client can find the new leader. We missed this case while implementing https://github.com/apache/kafka/pull/12187. This patch fixes the issue and updates an existing test to ensure that `isShuttingDown` has not effect. We should consider adding integration tests for this as well. We can do this separately. CONFLUENT CHERRY-PICK NOTES: Conflicts: core/src/main/scala/kafka/server/ReplicaManager.scala core/src/test/scala/unit/kafka/server/ReplicaManagerTest.scala In ce-kafka, there is no `shuttingDown` field, so we modified the patch to use the `BrokerState` instead. Reviewers: David Jacot <djacot@confluent.io>
-
由 vasusharma343 创作于
* Config changes to support different hashing logic for offset map * ignore the hash type in KafkaConfigTest * Valid strings for hashing * renamed hash config param
-
由 Jason Gustafson 创作于
Currently the describe feature tool shows the wrong version string for `confluent.metadata.version`: ``` root@kafka-0:/# /opt/confluent/bin/kafka-features.sh --bootstrap-server localhost:9071 describe Feature: confluent.metadata.version SupportedMinVersion: 3.0-IV1 SupportedMaxVersion: 3.3-IV3 FinalizedVersionLevel: 3.3-IV1 Epoch: 2321447 Feature: metadata.version SupportedMinVersion: 3.0-IV1 SupportedMaxVersion: 3.3-IV3 FinalizedVersionLevel: UNKNOWN [0] Epoch: 2321447 ``` We should use the confluent release instead of the apache release. Reviewers: Niket <niket-goel@users.noreply.github.com>
-
由 byang 创作于
* add customplug rbac for connect-plugin-service * add cumstomplug for rbac admin * update to CustomConnectorPlugin * update UT Co-authored-by: Bill Yang <38335771+gmamebill@users.noreply.github.com>
-
由 Ismael Juma 创作于
When we use Java 11 or newer, we also enable TLS 1.3 which uses different cipher names. The test was dynamically updating the cipher suites to only include a TLS 1.2 cipher suite even though the cluster was configured to support TLS 1.3 and TLS 1.2. When the client connected, it would prefer TLS 1.3 even though no valid cipher was available for that. To fix the test, I updated the code to use TLS 1.3 ciphers when using Java 11 or newer. As a separate change, I suggest we do not allow cipher suite configurations that don’t include at least one valid cipher for each enabled TLS protocol version. The behavior is quite odd otherwise. I filed https://confluentinc.atlassian.net/browse/KSECURITY-791 for that. Reviewers: Jeff Kim <kimkb2011@gmail.com>
- 10月 13, 2022
-
-
由 andymg3 创作于
-
由 Michael Li 创作于
There have been some concerns regarding the semantics around the activeFilters. E.g. see [here](https://confluent.slack.com/archives/CE95YL7B5/p1665435492716619) and [here](https://github.com/confluentinc/schroedinger/pull/3144/files#r978168109). In order to provide better backwards/forwards compatibility for the RemoteConfiguration API and better UX, an empty activeFilter array should designate that telemetry be disabled. A missing activeFilter array designates to use the _default include list. We do this at the expense of losing the ability to enforce all fields to ensure we are always working with a valid RemoteConfiguration object and forcing users to be explicit in their configurations. https://confluentinc.atlassian.net/browse/METRICS-4649
-
由 Xavier Léauté 创作于
* refactored telemetry parsing using a thin abstraction to avoid code duplication across formats * added unit tests covering both formats
-
由 Raman Verma 创作于
- 10月 12, 2022
-
-
由 Arvind Thirunarayanan 创作于
-
由 Sushant Mahajan 创作于
* TRUST -> POOL * POLICY -> FILTER * SUBJECT_CLAIM -> IDENTITY_CLAIM
-
由 Manikumar Reddy 创作于
Currently ConfluentAuditLogProvider instance is not getting started in Kraft controller. This is due to error in authorizer startup. We are not configuring httpServerBinder, due to this we are getting NPE during authorizer startup. Due to this we are not emitting the audit events to audit event pipeline.