Zimbra Heartbeat failover problem

[lassaad]

Hi all,

I just put up a cluster Zimbra (just Zimbra-LDAP) with DRBD/heartbeat, synchronisation data is perfect (drbd), but the problem is that the control of the high availability is partial
so :

- If the master machine fails : takeover
- If I restart heartbeat : takeover

but if I do "/etc/init.d /zimbra stop" on the master node : nothing happened.

I tried to modify the script in zimbra in init.d, but nothing new, I doubt very much about the version of heartbeat, I use version 2.

Thank you for your support.

Best regards,

Lassaad.

[lassaad]

I forgot to mention that I am new to the forum and I do not master English well so I am sorry for the wrong phrases.

[uxbod]

Just stopping ZCS is not going to be seen by the cluster as a failure. What services are you monitoring ?

[lassaad]

thank you for your reply,

I supervise the Zimbra service, in fact my haresources below:


zimbrambox11 drbddisk::r0 Filesystem::/dev/drbd0::/opt::ext3 IPaddr::192.168.1.181 zimbra


thanks

[lassaad]

no ideas ?

[nzahar]

Hi ,
As you may know Zimbra consists of many different daemons. You should try to monitor them as well. This can be tricky though as you will have t put them in a kind of "group", that is if one fails, all should be relocated. Try doing something easier. Why don't you install a vm inside drbd. Then only monitor this vm and relocate it.
Vm's are really well documented for drbd.

Nikos