tillrohrmann commented on a change in pull request #14254: URL: https://github.com/apache/flink/pull/14254#discussion_r532455932
########## File path: docs/deployment/ha/kubernetes_ha.md ########## @@ -23,77 +23,50 @@ specific language governing permissions and limitations under the License. --> -## Kubernetes Cluster High Availability -Kubernetes high availability service could support both [standalone Flink on Kubernetes]({% link deployment/resource-providers/standalone/kubernetes.md %}) and [native Kubernetes integration]({% link deployment/resource-providers/native_kubernetes.md %}). +Flink's Kubernetes HA services use [Kubernetes](https://kubernetes.io/) for high availability services. -When running Flink JobManager as a Kubernetes deployment, the replica count should be configured to 1 or greater. -* The value `1` means that a new JobManager will be launched to take over leadership if the current one terminates exceptionally. -* The value `N` (greater than 1) means that multiple JobManagers will be launched simultaneously while one is active and others are standby. Starting more than one JobManager will make the recovery faster. +* Toc +{:toc} -### Configuration -{% highlight yaml %} -kubernetes.cluster-id: <ClusterId> -high-availability: org.apache.flink.kubernetes.highavailability.KubernetesHaServicesFactory -high-availability.storageDir: hdfs:///flink/recovery -{% endhighlight %} +Kubernetes high availability services can only be used when deploying to Kubernetes. +Consequently, they can be configured when using [standalone Flink on Kubernetes]({% link deployment/resource-providers/standalone/kubernetes.md %}) or the [native Kubernetes integration]({% link deployment/resource-providers/native_kubernetes.md %}) -#### Example: Highly Available Standalone Flink Cluster on Kubernetes -Both session and job/application clusters support using the Kubernetes high availability service. Users just need to add the following Flink config options to [flink-configuration-configmap.yaml]({% link deployment/resource-providers/standalone/kubernetes.md %}#common-cluster-resource-definitions). All other yamls do not need to be updated. - -<span class="label label-info">Note</span> The filesystem which corresponds to the scheme of your configured HA storage directory must be available to the runtime. Refer to [custom Flink image]({% link deployment/resource-providers/standalone/docker.md %}#customize-flink-image) and [enable plugins]({% link deployment/resource-providers/standalone/docker.md %}#using-plugins) for more information. - -{% highlight yaml %} -apiVersion: v1 -kind: ConfigMap -metadata: - name: flink-config - labels: - app: flink -data: - flink-conf.yaml: |+ - ... - kubernetes.cluster-id: <ClusterId> - high-availability: org.apache.flink.kubernetes.highavailability.KubernetesHaServicesFactory - high-availability.storageDir: hdfs:///flink/recovery - restart-strategy: fixed-delay - restart-strategy.fixed-delay.attempts: 10 - ... -{% endhighlight %} +## Configuration -#### Example: Highly Available Native Kubernetes Cluster -Using the following command to start a native Flink application cluster on Kubernetes with high availability configured. -{% highlight bash %} -$ ./bin/flink run-application -p 8 -t kubernetes-application \ - -Dkubernetes.cluster-id=<ClusterId> \ - -Dtaskmanager.memory.process.size=4096m \ - -Dkubernetes.taskmanager.cpu=2 \ - -Dtaskmanager.numberOfTaskSlots=4 \ - -Dkubernetes.container.image=<CustomImageName> \ - -Dhigh-availability=org.apache.flink.kubernetes.highavailability.KubernetesHaServicesFactory \ - -Dhigh-availability.storageDir=s3://flink/flink-ha \ - -Drestart-strategy=fixed-delay -Drestart-strategy.fixed-delay.attempts=10 \ - -Dcontainerized.master.env.ENABLE_BUILT_IN_PLUGINS=flink-s3-fs-hadoop-{{site.version}}.jar \ - -Dcontainerized.taskmanager.env.ENABLE_BUILT_IN_PLUGINS=flink-s3-fs-hadoop-{{site.version}}.jar \ - local:///opt/flink/examples/streaming/StateMachineExample.jar -{% endhighlight %} +In order to start an HA-cluster you have to configure the following configuration keys: -### High Availability Data Clean Up -Currently, when a Flink job reached the terminal state (`FAILED`, `CANCELED`, `FINISHED`), all the HA data, including metadata in Kubernetes ConfigMap and HA state on DFS, will be cleaned up. +- **high-availability mode** (required): +The [`high-availability`]({% link deployment/config.md %}#high-availability-1) option has to be set to `KubernetesHaServicesFactory`. -So the following command will only shut down the Flink session cluster and leave all the HA related ConfigMaps, state untouched. -{% highlight bash %} -$ echo 'stop' | ./bin/kubernetes-session.sh -Dkubernetes.cluster-id=<ClusterId> -Dexecution.attached=true -{% endhighlight %} + <pre>high-availability: org.apache.flink.kubernetes.highavailability.KubernetesHaServicesFactory</pre> + +- **Storage directory** (required): +JobManager metadata is persisted in the file system [`high-availability.storageDir`]({% link deployment/config.md %}#high-availability-storagedir) and only a pointer to this state is stored in ZooKeeper. Review comment: Good catch :-) ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org