[
https://issues.apache.org/jira/browse/FLINK-9777?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16541842#comment-16541842
]
ASF GitHub Bot commented on FLINK-9777:
---------------------------------------
Github user dawidwys commented on a diff in the pull request:
https://github.com/apache/flink/pull/6297#discussion_r202085772
--- Diff:
flink-yarn/src/main/java/org/apache/flink/yarn/cli/FlinkYarnSessionCli.java ---
@@ -386,10 +386,10 @@ private ClusterSpecification
createClusterSpecification(Configuration configurat
}
// JobManager Memory
- final int jobManagerMemoryMB =
MemorySize.parse(configuration.getString(JobManagerOptions.JOB_MANAGER_HEAP_MEMORY)).getMebiBytes();
+ final int jobManagerMemoryMB =
MemorySize.parse(configuration.getString(JobManagerOptions.JOB_MANAGER_HEAP_MEMORY),
MemorySize.MemoryUnit.MEGA_BYTES).getMebiBytes();
--- End diff --
I think we cannot do this as users might not use the `config.sh`. At the
same time we think we really should keep backwards compatibility, as otherwise
there might be complaints. We've discussed this with @GJL offline and we think
the way to go right now would be to check everywhere we use the new option to
check for the old one as well if the new one was not set.
That is:
*
`org.apache.flink.client.deployment.ClusterSpecification#fromConfiguration`
* `org.apache.flink.yarn.YarnResourceManager#YarnResourceManager`
* `org.apache.flink.yarn.cli.FlinkYarnSessionCli#createClusterSpecification`
Probably we should wrap it in some utility method.
Also for the yarn command line we should keep the old behaviour. This means
we should add "m" suffix in
`org.apache.flink.yarn.cli.FlinkYarnSessionCli#applyCommandLineOptionsToConfiguration`
> YARN: JM and TM Memory must be specified with Units
> ----------------------------------------------------
>
> Key: FLINK-9777
> URL: https://issues.apache.org/jira/browse/FLINK-9777
> Project: Flink
> Issue Type: Bug
> Components: Documentation, YARN
> Affects Versions: 1.6.0
> Environment: commit 9f736d1927c62d220a82931c4f5ffa4955910f27
> Reporter: Gary Yao
> Assignee: vinoyang
> Priority: Blocker
> Labels: pull-request-available
> Fix For: 1.6.0
>
>
> FLINK-6469 breaks backwards compatibility because the JobManager and
> TaskManager memory must be specified with units (otherwise bytes are
> assumed). The command to start a YARN session as documented
> ([https://github.com/apache/flink/blob/9f736d1927c62d220a82931c4f5ffa4955910f27/docs/ops/deployment/yarn_setup.md)
>
> |https://github.com/apache/flink/blob/9f736d1927c62d220a82931c4f5ffa4955910f27/docs/ops/deployment/yarn_setup.md]
> would not work because 1024 bytes and 4096 bytes are not enough for the heap
> size. The command finishes with the following exception:
> {noformat}
> java.lang.reflect.UndeclaredThrowableException
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1854)
> at
> org.apache.flink.runtime.security.HadoopSecurityContext.runSecured(HadoopSecurityContext.java:41)
> at
> org.apache.flink.yarn.cli.FlinkYarnSessionCli.main(FlinkYarnSessionCli.java:802)
> Caused by: org.apache.flink.client.deployment.ClusterDeploymentException:
> Couldn't deploy Yarn session cluster
> at
> org.apache.flink.yarn.AbstractYarnClusterDescriptor.deploySessionCluster(AbstractYarnClusterDescriptor.java:420)
> at
> org.apache.flink.yarn.cli.FlinkYarnSessionCli.run(FlinkYarnSessionCli.java:599)
> at
> org.apache.flink.yarn.cli.FlinkYarnSessionCli.lambda$main$2(FlinkYarnSessionCli.java:802)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1836)
> ... 2 more
> Caused by: org.apache.flink.util.FlinkException: Cannot fulfill the minimum
> memory requirements with the provided cluster specification. Please increase
> the memory of the cluster.
> at
> org.apache.flink.yarn.AbstractYarnClusterDescriptor.validateClusterSpecification(AbstractYarnClusterDescriptor.java:453)
> at
> org.apache.flink.yarn.AbstractYarnClusterDescriptor.deployInternal(AbstractYarnClusterDescriptor.java:475)
> at
> org.apache.flink.yarn.AbstractYarnClusterDescriptor.deploySessionCluster(AbstractYarnClusterDescriptor.java:413)
> ... 7 more
> Caused by: java.lang.IllegalArgumentException
> at
> org.apache.flink.util.Preconditions.checkArgument(Preconditions.java:123)
> at
> org.apache.flink.runtime.clusterframework.ContaineredTaskManagerParameters.calculateCutoffMB(ContaineredTaskManagerParameters.java:115)
> at
> org.apache.flink.yarn.AbstractYarnClusterDescriptor.validateClusterSpecification(AbstractYarnClusterDescriptor.java:450)
> ... 9 more
> {noformat}
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
