You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@tez.apache.org by "Siddharth Seth (JIRA)" <ji...@apache.org> on 2017/03/02 03:19:45 UTC
[jira] [Updated] (TEZ-3644) Cleanup container list stored in AMNode
[ https://issues.apache.org/jira/browse/TEZ-3644?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Siddharth Seth updated TEZ-3644:
--------------------------------
Attachment: TEZ-3464.01.patch
Completed containers are removed from AMNode whenever a dag completes. They cannot be removed while a dag is running - since a failure of a node may require a node to re-run.
Containers which have not completed cannot be removed after dag completion to allow for re-use.
Fairly straight-forward implementation.
An event is sent out from AMContainer whenever it is entering a final state. AMNode tracks these. Removes completed containers when a dag completes.
[~rajesh.balamohan] - please review.
> Cleanup container list stored in AMNode
> ---------------------------------------
>
> Key: TEZ-3644
> URL: https://issues.apache.org/jira/browse/TEZ-3644
> Project: Apache Tez
> Issue Type: Bug
> Reporter: Siddharth Seth
> Assignee: Siddharth Seth
> Attachments: TEZ-3464.01.patch
>
>
> A node tracks containers that it has run, to send out notifications in case of node failure. This list is only cleaned up when the node moves to an unhealthy state / moves back into a healthy state.
> These containers need to be cleaned up at some point.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)