On Jan 30, 2008, at 7:59 AM, Adrian Overbury wrote:
Hi,
I've got a two-node NFS cluster running heartbeat 2.1.2, with drbd
0.7, nfs, a floating IP address for nfs and a filesystem resource.
The last piece of our puzzle for this cluster is the LDAP database.
The rest of our system uses OpenLDAP for authentication and
authorisation. Right now the LDAP master is stuck on fs1 (the
cluster hostnames are fs1 and fs2. Original, no?), because it's not
managed by Heartbeat at all.
What I want to do with this resource agent is be able to specify two
config files. One is used when the resource is in a master state,
the other when it's a slave. The only difference between the two is
that the master file has entries for all the replica destinations,
and the slave one has an entry telling it to expect replication data
from the master (which is the cluster's floating IP).
Now, I've never written an OCF agent before, so I'm looking for any
tips, traps or pitfalls that anyone knows or thinks I might come
across. A good guide on how an OCF agent or master/slave agent is
constructed would be good too, as well as some way to test the agent
--
have you seen http://linux-ha.org/v2/Concepts/MultiState and/or the
ocf-tester script?
there is also an RA called "Stateful" that might be a good template
preferably without having it actually fail over my LDAP servers.
that part probably isn't possible - I'd suggest creating some sort of
sandbox to experiment in
Regards,
Adrian
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
