[PATCH]: Inconsistent recmaster during election.

[Kenny Dinh]

[PATCH] updated.

I have a more targeted fix in the new version of my patch.  Please
disregard my previous patch.  In the new patch, we should let the recovery
master win the election if the recmaster sent the election message and we
agreed that it is the recovery master.

Please review and push if it looks good.

Thank you!

On Mon, Dec 28, 2015 at 1:04 PM, Kenny Dinh <kdinh at peaxy.net> wrote:

> Hello,
>
> I have a cluster with 2 CTDB nodes.  My test does the following:
>
> * Restart ctdb service on RecMaster node (Node 1).
> * Restart ctdb service on NonRecMaster node (Node 0).
> * When ctdb service on NonRecMaster(0) is started, it triggered an
> election.
> * During the election process, NonRecMaster(0) node responded to
> "CTDB_CONTROL_GET_RECMASTER" from RecMaster(1) with Node 1 as the
> recmaster.  However, when processing the election message, NonRecMaster(0)
> node indicated that it won the election.  The inconsistency caused the
> RecMaster(1) to trigger new election.  The cycle is repeated 4 times until
> RecMaster(1) decided to ban the culprit( Node 0).
>
> Attached are the ctdb log from RecMaster and NonRecMaster, and a proposed
> patch.  Essentially, NonRecMaster should not win the election if it
> responded that a different node is the recovery master.
>
> Please verify if my patch is correct.
>
> Thanks,
> Kenny
>
>
>
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: let-recmaster-win-election.patch
Type: application/octet-stream
Size: 1195 bytes
Desc: not available
URL: <http://lists.samba.org/pipermail/samba-technical/attachments/20151229/b5b7cd1e/let-recmaster-win-election.obj>