[
https://issues.apache.org/jira/browse/YARN-9184?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16737234#comment-16737234
]
Eric Badger commented on YARN-9184:
-----------------------------------
I think this would be a nice option to have, but should definitely be
configured as off by default. There are many different problems that can occur
on a cluster if the nodes all decide to pull the image every time it goes to
run a container.
- We have to hit the docker registry for every container run on the cluster,
which would be massive on a large cluster
- The images could be significantly different if an image was pushed halfway
through a job. Then some containers would run with the old image and the rest
would run with the newer image.
- {{docker pull}} can take a very long time to complete. It could either
increase the runtime of the job by a bunch or cause a container timeout
If the admin knows that these problems won't exist, then it could be a nice
feature to have
> Docker run doesn't pull down latest image if the image exists locally
> ----------------------------------------------------------------------
>
> Key: YARN-9184
> URL: https://issues.apache.org/jira/browse/YARN-9184
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: nodemanager
> Affects Versions: 3.1.0, 3.0.3
> Reporter: Zhaohui Xin
> Assignee: Zhaohui Xin
> Priority: Major
>
> See [docker run doesn't pull down latest image if the image exists
> locally|https://github.com/moby/moby/issues/13331].
> So, I think we should pull image before run to make image always latest.
--
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
