[Linux-HA] mixed storage use (nfs and hw)
dejanmm at fastmail.fm
Wed Feb 8 09:12:22 MST 2006
It's hard to imagine that nobody faced this kind of situation
before, but somehow I was not able to find anything about it in
the list archives. I'm not even sure if the Subject fits. In a
way, it is similar to the standard SAP setup.
We should implement a 3-node cluster which is connected to a f/o
shared storage (IBM DS4300). User applications (apache/tomcat)
should have access to the same data (the same disk). In a normal
state, one of the nodes mounts the data directly over the f/o and
exports it to the other two over NFS and a virtual IP address.
Now, if the "primary" node fails, one of the other nodes should
mount the disk over the f/o, take over the virtual IP from the
failed node, and export it to the remaining node. Before that it
has to stop all resources (applications) which depend on NFS.
Basically, we have to stop and start again the good and running
resources on the node which takes over resources from the failing
node. It is not clear to me how can one do this with CRM.
Can we achieve that with the Heartbeat v2? Has anybody already
solved this using Heartbeat?
P.S. Sorry if this appears twice, the first message still didn't
get to the list after two hours. Sigh!
More information about the Linux-HA