[Openais] Cluster not started after init 6

[Steven Dake]

On Wed, 2010-03-31 at 10:18 +0200, Haussecker, Armin wrote:
> Hello,
>  
> in our cluster consisting of two nodes we can see the following
> effect:
>  
>       * on both nodes openais is stopped
>       * on one node, init 6 is called to reboot the system
>       * during boot processing, openais is started, but the concerned
>         node remains in state OFFLINE; we can see the following
>         message in /var/log/messages:
>         crmd: [17541]: info: ais_dispatch: Membership 1844: quorum
>         still lost
>         (about every three minutes)
>       * stopping openais by command /etc/init.d/openais stop does not
>         work, stop processing not ends correctly, it must be
>         interrupted by CTRL+C:
>          /etc/init.d/openais stop
>         Stopping OpenAIS daemon
>         (aisexec): .................................................................^C
>       * to stop openais forced, we got pid of ais process (ps ax |
>         grep ais), and killed this process
>       * afterwards we started openais by command /etc/init.d/openais
>         start, and everything worked well !!
>         
> What can be the reason for this behavior ??
>  
> our Linux kernel version: 2.6.27.45-0.1-xen
> our openais version: openais-0.80.3-26.8.1
>  
> Thank you very much for your help.     
>  
> Armin Haußecker
>  
>  
Armin,

Could just be bugs in openais shutdown related to Pacemaker.  At this
point openais is widely deployed and I am not interested in risking
those deployments to ensure Pacemaker has an effective shutdown.  The
reason is the amount of work that went into making Corosync shutdown
properly with Pacemaker was significant, and I don't want to destabilize
current openais users who don't use Pacemaker.

I highly recommend using Corosync in the case that you intend to use
Pacemaker.  We have spent significant engineering effort making shutdown
work well for the Corosync+Pacemaker use case.

For more info on why corosync was created check out our faq:

http://www.corosync.org/doku.php?id=faq:why

To get the software:

http://www.corosync.org

Andrew provides prebuilt RPMS for a variety of distros at
www.cluterlabs.org


Regards
-steve


>  
>  
>  
>  
> _______________________________________________
> Openais mailing list
> Openais at lists.linux-foundation.org
> https://lists.linux-foundation.org/mailman/listinfo/openais