[jira] [Commented] (BOOKKEEPER-960) Re-replicator: bookie checking should handle flapping bookie registration

["Ivan Kelly (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/BOOKKEEPER-960?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16208382#comment-16208382 ] 

Ivan Kelly commented on BOOKKEEPER-960:
---------------------------------------

Migrated to github https://github.com/apache/bookkeeper/issues/587

> Re-replicator: bookie checking should handle flapping bookie registration
> -------------------------------------------------------------------------
>
>                 Key: BOOKKEEPER-960
>                 URL: https://issues.apache.org/jira/browse/BOOKKEEPER-960
>             Project: Bookkeeper
>          Issue Type: Improvement
>          Components: bookkeeper-server
>            Reporter: Yiming Zang
>            Assignee: Yiming Zang
>              Labels: ghmigrated, iktriage0
>
> Problem:
> currently re-replicator only uses the view of available/readonly bookies right at the time doing bookie checking. it would accidentally treat a bookie disappeared from zookeeper (e.g. zookeeper session expired, bookie restarted, flapping bookie registration due to network/gc) as lost bookies, which introduce unnecessary re-replication.
> Solution:
> introduce 'auditorStaleBookieInterval', if a bookie never register in the given interval, it would be marked as 'stale' bookies and re-replicate all ledgers belongs to that bookie. the default value is set 30 minutes.
> Fixes:
> - refactor bookie watcher to allow notifying bookie list thru BookiesListener
> - introduce 'auditorStaleBookieInterval' to be able to mark bookies as 'stale' if bookies aren't registered themselves to zookeeper
> - add more info logging about critical steps on re-replication logic
> - misc changes



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)