[
https://issues.apache.org/jira/browse/YARN-7939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16454270#comment-16454270
]
Shane Kumpf commented on YARN-7939:
-----------------------------------
Thanks for the patch, [~csingh]! Overall, this looks good to me. I just wanted
to clarify a couple of points regarding LCE and the latest patch.
{quote}I think this is blocked by the relaunch logic introduced in YARN-7973.
{quote}
I did some testing here, and that patch is not related. This upgrade patch is
leveraging the Reinitialize features, not Relaunch. Reinit is broken with LCE,
which is an existing issue unrelated to this patch or YARN-7973. The problem
that occurs because the launch_container.sh and tokens file that is copied to
the work directory of the container during the initial launch. These files are
not cleaned up prior to launching the upgraded container and container-executor
errors out because the files already exist in the work dir.
The reason we don't see this for Relaunch is that {{ContainerRelaunch}}
explicitly handles cleaning up those files. YARN-8194 likely needs to do the
same.
{code}
/**
* Clean up container's previous files for container relaunch.
*/
private void cleanupPreviousContainerFiles(Path containerWorkDir) {
// delete ContainerScriptPath
deleteAsUser(new Path(containerWorkDir, CONTAINER_SCRIPT));
// delete TokensPath
deleteAsUser(new Path(containerWorkDir, FINAL_CONTAINER_TOKENS_FILE));
}
{code}
I don't think YARN-8194 should block this patch since it is an existing issue,
but note that the upgrade feature can't be tested on a single node cluster
running LCE until YARN-8194 is resolved. The single NM will crash and there
isn't another NM to launch the new container.
> Yarn Service Upgrade: add support to upgrade a component instance
> ------------------------------------------------------------------
>
> Key: YARN-7939
> URL: https://issues.apache.org/jira/browse/YARN-7939
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Chandni Singh
> Assignee: Chandni Singh
> Priority: Major
> Attachments: YARN-7939.001.patch, YARN-7939.002.patch,
> YARN-7939.003.patch, YARN-7939.004.patch, YARN-7939.005.patch,
> YARN-7939.006.patch, YARN-7939.007.patch, YARN-7939.008.patch,
> YARN-7939.009.patch, YARN-7939.010.patch, YARN-7939.011.patch,
> YARN-7939.012.patch, serviceam.log, upgrade_logs.tgz
>
>
> Yarn core supports in-place upgrade of containers. A yarn service can
> leverage that to provide in-place upgrade of component instances. Please see
> YARN-7512 for details.
> Will add support to upgrade a single component instance first and then
> iteratively add other APIs and features.
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
