Hello everyone. I am setting up a heartbeat v1 cluster, and I have
some questions about configuring stonith. I've set up heartbeat a few
different times, but never with shared storage so stonith was not a
big issue.
1. Under what conditions is the stonith device invoked? I think a
node will be stonith'ed (if that's a word) if a resource's stop action
exits abnormally, but are there other situations which will trigger a
stonith action? I want to design a test plan that verifies the
stonith configuration is performing as expected.
2. Who 'does' the stonith-ing? Is it always node A triggering stonith
on node B, and vice versa, or are there some situations where a node
will commit 'suicide' and stonith itself? I've wondered if SMITH
(shoot myself in the head) ever occurs.
3. I am planning to use the stonith_host directive in ha.cf, so as to
keep all heartbeat configuration in that single file. In the
documentation for the 'hostfrom' stonith_host parameter, the phrase
"or * to mean it is accessible from any host" is confusing. Are we to
configure which node the stonith device is attached to, or which nodes
the device is accessible from? It seems that the former makes the
most sense, but I'm unsure.
If I have nodes calvin and hobbes, and both have HP ilo2 devices, is
this the correct configuration?
stonith_host calvin external/riloe {{riloe paramters for ilo2
device attached to calvin here}}
stonith_host hobbes external/riloe {{riloe paramters for ilo2
device attached to hobbes here}}
4. In searching around, I found an email thread from August 2008
mentioning some problems with the riloe stonith plugin. Where can I
go to see if these problem is still a current concern?
http://lists.community.tummy.com/pipermail/linux-ha/2008-August/034323.html
5. ilo2 is integrated with its host machine, but has a separate
ethernet port. In the case of a network failure, where my nodes are
only able to communicate over a serial connection, will one still be
able to stonith the other? (Can the ilo attached to node A be
triggered without an active ethernet connection?)
Thanks for any help you can provide.
regards,
alex dean
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
