Hi,

I wonder whether heartbeat can utilize a shared disk via FC to increase the 
redundancy of its heartbeat. Actually, I was almost convinced that this is
implemented already.

Consider the following setup: 
- two datacenters hosting a server and a SAN each
- a (highly available) network connection between the datacenters/servers
- no serial connection
- EVMS and heartbeat v2 to manage the disks
- (stonith via another virtual network and HP-iLO)

A LUN on each SAN is visible (via multipath) to both servers (in both 
datacenters) and so it will prevent a split brain in case the (highly
available) network does not deserve its attribute at some point in time.
Especially, because the iLO access might be faulty as well if more than
the virtual network is lost.

I'd like to use an additional shared EVMS volume or some (remote-) HBA-Feature
for heartbeat communication between the servers - actually, just using
the FC connection for communication as well. Of course, I don't mean
or even activate the local HBA-heartbeat since failover between HBAs is
managed by multipath.

The HBAs are two Emulex 9802 (targets are mapped with lpfc) connected
to LUNs at a local and a remote IBM DS8300 SAN.

If heartbeat over disk is possible (like diskhb with HACMP for AIX)
 I would be greatful to get some hints.

Regards,
Andreas Huck
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems

Reply via email to