[ 
https://issues.apache.org/jira/browse/YARN-8665?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16618286#comment-16618286
 ] 

Chandni Singh edited comment on YARN-8665 at 9/17/18 11:52 PM:
---------------------------------------------------------------

Patch 1 includes support for cancel upgrade.
To test the patch :
1. yarn app -launch test1 sleeper
2. yarn app -upgrade test1 -initiate sleeper_v2.json
3. yarn app -upgrade test1 -instances sleeper-0
4. yarn app -upgrade test1 -cancel

[~eyang] [~gsaha] Could you please review?

Please note that with my changes, the ServiceAM will only send a second reinit 
request for the same container until it has received a response from the NM for 
previous reinit request.

I see an intermittent issue with NM when there are subsequent reinitialization 
requests made for the same container.  NM sends premature response back to 
ServiceAM that container is running but the state of the container in NM is 
still {{REINITIALIZING_AWAITING_KILL}}. This is hard to reproduce manually. 
Since this is a NM side bug, I am going to open a new ticket to fix it. 


was (Author: csingh):
Patch 1 includes support for cancel upgrade.
To test the patch :
1. yarn app -launch test1 sleeper
2. yarn app -upgrade test1 -initiate sleeper_v2.json
3. yarn app -upgrade test1 -instances sleeper-0
4. yarn app -upgrade test1 -cancel

[~eyang] [~gsaha] Could you please review?

Please note that with my changes, the ServiceAM will only send a second reinit 
request for the same container until it has received a response from the NM for 
previous reinit request.

I see an intermittent issue with NM when there are subsequent reinitialization 
requests made for same container.  NM sends premature response back to 
ServiceAM that container is running but the state of the container in NM is 
still {{REINITIALIZING_AWAITING_KILL}}. This is hard to reproduce manually. 
Since this is a NM side bug, I am going to open a new ticket to fix it. 

> Yarn Service Upgrade:  Support cancelling upgrade
> -------------------------------------------------
>
>                 Key: YARN-8665
>                 URL: https://issues.apache.org/jira/browse/YARN-8665
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Chandni Singh
>            Assignee: Chandni Singh
>            Priority: Major
>         Attachments: YARN-8665.001.patch
>
>
> When a service is upgraded without auto-finalization or express upgrade, then 
> the upgrade can be cancelled. This provides the user ability to test upgrade 
> of a single instance and if that doesn't go well, they get a chance to cancel 
> it.



--
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

Reply via email to