1996fanrui commented on code in PR #758: URL: https://github.com/apache/flink-web/pull/758#discussion_r1803996694
########## docs/content/posts/2024-10-30-release-kubernetes-operator-1.10.0.md: ########## @@ -0,0 +1,75 @@ +--- +title: "Apache Flink Kubernetes Operator 1.10.0 Release Announcement" +date: "2024-10-30T18:00:00.000Z" +authors: +- mateczagany: + name: "Mate Czagany" +aliases: +- /news/2024/10/30/release-kubernetes-operator-1.10.0.html +--- + +The Apache Flink community is excited to announce the release of Flink Kubernetes Operator 1.10.0! + +The release includes several improvements to the autoscaler, and introduces a new Kubernetes custom resource called FlinkStateSnapshot to manage job snapshots. +The process of job upgrades has also been enhanced which makes it possible to now use the last-state upgrade mode with session jobs. + +We encourage you to [download the release](https://flink.apache.org/downloads.html) and share your experience with the +community through the Flink [mailing lists](https://flink.apache.org/community.html#mailing-lists) or +[JIRA](https://issues.apache.org/jira/browse/flink)! We're looking forward to your feedback! + +## Highlights + +### FlinkStateSnapshot + +With this version comes also a new custom resource called FlinkStateSnapshot. +This is used to describe savepoint or checkpoint for a Flink job. +The savepoint/checkpoint fields found in FlinkDeployment and FlinkSessionJob status are therefore deprecated, and the operator will create new FlinkStateSnapshot resources for periodic, update and manual savepoints/checkpoints. + +Users can also create new FlinkStateSnapshot resources, which will instruct the operator to trigger new checkpoint/savepoint. + +This new feature is enabled by default, unless disabled by setting `kubernetes.operator.snapshot.resource.enabled` to false or if the FlinkStateSnapshot CRD was not found on the Kubernetes cluster. + +You can read more about this feature [here](https://nightlies.apache.org/flink/flink-kubernetes-operator-docs-release-1.10/docs/custom-resource/snapshots/). + + +### Last-State Upgrade Mode + +For deployments using last-state upgrade mode, the operator - instead of forcefully deleting the JobManager pods to trigger a restart - will cancel the job via REST API and extract the last checkpoint info after cancellation if the job is healthy. +This change makes it possible to finally use the last-state upgrade mode for session jobs as well. + + +### Autoscaler Delayed Scale Down + +With the introduction of the configuration option `job.autoscaler.scale-down.interval`, the operator can now optimize multiple scale-down operations to a single one to prevent too many unnecessary downscales, thus improving job availability. +Please note that `job.autoscaler.scale-up.grace-period` has been removed with this change. + + +### Other Autoscaler Improvements +- Optimized cases where partitions or key groups cannot be evenly distributed to subtasks in case of Kafka and Pulsar Review Comment: Sorry, this optimization is applied for source task and downstream task with hash link(key group). In the beginning, FLINK-36192 proposes to optimize source task(Kafka and Pulsar), we found the downstream task with hash link(key group) could be optimized together during review. I updated the JIRA title and PR title, unfortunately the commit message is not accurate, sorry for that. ```suggestion - Optimized cases where partitions or key groups cannot be evenly distributed to subtasks ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@flink.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org