[Linux-HA] when Filesystem script fails because fs is mounted, heartbeat still starts

[Alan Robertson]

Michael Will wrote:
> two servers with one shared storage /dev/sda1 mounted at /home:
> 
> haresource looks like this:
> 
> primary 10.1.1.10 httpd Filesystem::/dev/sda1::/home::xfs mysqld applicationd mon
> 
> A user error unexpectedly damaged the filesystem:
> 
> /home was mounted by hand, resource.d/Filesystem fails, and heartbeat complains
> about it but still starts. At failover time, it does not unmount the filesystem,
> probably because it had failed at start-time, and the secondary mounts the
> already mounted filesystem - just like in a split brain situation that we
> carefully avoided by having a serial and an ethernet heartbeat.
> 
> Is there a way to tell heartbeat to not startup if one of the services fails, 
> or do I need to hack resource.d/Filesystem to prevent this error by having it go
> through the stop-code before the start-code if it finds the filesystem to be
> already mounted?

If the fileystem is already mounted by hand, then Filesystem should succeed 
(it might say something though).  If it doesn't then that's a bug...

If the unmount fails with 1.2.1 and later, then the machine will reboot 
precisely to avoid this particular situation.


-- 
     Alan Robertson <alanr at unix.sh>

"Openness is the foundation and preservative of friendship...  Let me claim 
from you at all times your undisguised opinions." - William Wilberforce