[
https://issues.apache.org/jira/browse/FLINK-26316?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Yang Wang updated FLINK-26316:
------------------------------
Description:
Reported by user in this [ML
thread|[https://lists.apache.org/thread/9nf4j2b45zxhtwksgcjsvxwgcjvx6t83].]
If you use the following command to start a Flink k8s application, the
savepoint restore settings does not take effect.
{code:java}
$FLINK_HOME/bin/flink run-application -t kubernetes-application \
-Dkubernetes.cluster-id=$CLUSTER_ID \
-Dkubernetes.namespace=$NAMESPACE \
-Dkubernetes.container.image=$IMAGE \
-Dexecution.savepoint.ignore-unclaimed-state=true \
-Dexecution.savepoint.path=oss://flink-test/flink-ha \
local:///opt/flink/examples/streaming/StateMachineExample.jar
{code}
Not only the K8s application, it also does not work for yarn application and
yarn per-job mode.
The root cause is that {{SavepointRestoreSettings#none()}} will override the
effective configuration wrongly in {{{}ProgramOptions#applyToConfiguration{}}}.
was:
If you use the following command to start a Flink k8s application, the
savepoint restore settings does not take effect.
{code:java}
$FLINK_HOME/bin/flink run-application -t kubernetes-application \
-Dkubernetes.cluster-id=$CLUSTER_ID \
-Dkubernetes.namespace=$NAMESPACE \
-Dkubernetes.container.image=$IMAGE \
-Dexecution.savepoint.ignore-unclaimed-state=true \
-Dexecution.savepoint.path=oss://flink-test/flink-ha \
local:///opt/flink/examples/streaming/StateMachineExample.jar
{code}
Not only the K8s application, it also does not work for yarn application and
yarn per-job mode.
The root cause is that {{SavepointRestoreSettings#none()}} will override the
effective configuration wrongly in {{{}ProgramOptions#applyToConfiguration{}}}.
> Savepoint restore setting does not take effect when configured via dynamic
> config options
> -----------------------------------------------------------------------------------------
>
> Key: FLINK-26316
> URL: https://issues.apache.org/jira/browse/FLINK-26316
> Project: Flink
> Issue Type: Bug
> Components: Command Line Client
> Reporter: Yang Wang
> Priority: Major
>
> Reported by user in this [ML
> thread|[https://lists.apache.org/thread/9nf4j2b45zxhtwksgcjsvxwgcjvx6t83].]
>
> If you use the following command to start a Flink k8s application, the
> savepoint restore settings does not take effect.
> {code:java}
> $FLINK_HOME/bin/flink run-application -t kubernetes-application \
> -Dkubernetes.cluster-id=$CLUSTER_ID \
> -Dkubernetes.namespace=$NAMESPACE \
> -Dkubernetes.container.image=$IMAGE \
> -Dexecution.savepoint.ignore-unclaimed-state=true \
> -Dexecution.savepoint.path=oss://flink-test/flink-ha \
> local:///opt/flink/examples/streaming/StateMachineExample.jar
> {code}
> Not only the K8s application, it also does not work for yarn application and
> yarn per-job mode.
>
> The root cause is that {{SavepointRestoreSettings#none()}} will override the
> effective configuration wrongly in
> {{{}ProgramOptions#applyToConfiguration{}}}.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
