[
https://issues.apache.org/jira/browse/HDFS-2769?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13182369#comment-13182369
]
Aaron T. Myers commented on HDFS-2769:
--------------------------------------
bq. We need to prevent failover if the active gets ahead of the standby,
This is the crux of the issue. How will the standby (or failover controller)
know that the shared edits dir hasn't been receiving updates, and therefore
isn't up-to-date? We need to ensure that FS updates hit the shared storage, or
else on death of the active the standby may come up as the active, but be
missing recent edits.
Once HA supports multiple shared edits dirs, we should change things to support
a pool of shared edits dirs which are collectively treated as a required
resource, but are individually treated as redundant.
> HA: When HA is enabled with a shared edits dir, that dir should be marked
> required
> ----------------------------------------------------------------------------------
>
> Key: HDFS-2769
> URL: https://issues.apache.org/jira/browse/HDFS-2769
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: ha, name-node
> Affects Versions: HA branch (HDFS-1623)
> Reporter: Aaron T. Myers
> Assignee: Aaron T. Myers
>
> HDFS-2430 introduced the concept of "required" edits/name dirs, which are not
> treated as redundant when considering NN storage failure. When a shared edits
> dir is used, this should automatically be marked as required.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
