[ClusterLabs] Resources restart

Digimer lists at alteeve.ca
Sun Sep 11 23:34:39 UTC 2016


On 12/09/16 06:24 AM, Gienek Nowacki wrote:
> 
> 
> 2016-09-11 21:48 GMT+02:00 Andreas Kurz <andreas.kurz at gmail.com
> <mailto:andreas.kurz at gmail.com>>:
> 
>     Hi,
> 
>     On Sun, Sep 11, 2016 at 6:28 PM, Gienek Nowacki <nowackig at gmail.com
>     <mailto:nowackig at gmail.com>> wrote:
> 
>         Hi,
> 
>         I have a problem with my cluster.
>         There are two nodes: wirt1v and wirt2v with pacemaker, corosync,
>         dlm,
>         drbd (/dev/drbd2) and filesystem mounted as /virtfs2 with gfs2.
>         Each node contains LVM partition on wihich is DRBD.
> 
>         The situation is as follow:
> 
>         pcs cluster standby wirt2v
>          ...all is OK, there is possible to use /virtfs2 on wirt1v node
> 
>         pcs cluster unstandby wirt2v
>          This situation causes restarting/remounting Drbd2/Virtfs2 on
>         the wirt1v node.
> 
> 
>     Looks like you are missing the meta-attribute "interleave=true" for
>     your filesystem clone:
> 
>     http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-single/Pacemaker_Explained/index.html#_clone_options
>     <http://clusterlabs.org/doc/en-US/Pacemaker/1.1-pcs/html-single/Pacemaker_Explained/index.html#_clone_options>
> 
>     Regards,
>     Andreas
>       
> 
> 
> Hi,
> 
> Yes it's right.  There was needed interleave=true
> 
> Thanks for yor help,
> Gienek

Separately, it looks like you haven't configured fencing in DRBD (but
you do have stonith is pacemaker). Please set 'fencing
resource-and-stonith;' and configure the 'crm-{un,}fence-peer.sh'
{un,}fence-handlers.

This way, if a split-brain were about to occur, DRBD will block and wait
for confirmation that pacemaker fenced the peer.

-- 
Digimer
Papers and Projects: https://alteeve.ca/w/
What if the cure for cancer is trapped in the mind of a person without
access to education?




More information about the Users mailing list