On Jan 26, 2008, at 5:55 PM, Atanas Dyulgerov wrote:
Hi all,
I'm running Heartbeat v2.1.1. I have two Heartbeat nodes with drbd
resource running in master/slave. I use Pingd to ping third machine
in my network let's call it pingnode.
In the initial state my first node DRBD is in master state and my
second in slave.
1. In case one the first node loses connectivity to the pingnode and
according to my constraints and timeout parameters (deadping ~ 5s)
the node goes in Slave state for about 10s
and is stopped after that?
and the second node DRBD turns to Master. Both Heartbeats remain
connected during that time.
Everything works perfect for me.
2. In case two I unplug the network cable of the first node. The
node loses connectivity to the pingnode and to the second Heartbeat
node. What I want is the first node DRBD to go in Slave state again
for 10s as in case one. However this does not happen. After a minute
Heartbeat detects that the other peer is dead and then brings down
the DRBD.(initdead and deadtime = 50s)
I want Heartbeat to put DRBD in Slave state in 10s as it does in
case one. How can I do that?
I need first node DRBD to go in Slave state before second node goes
in Master. Now for a short period of time both DRBDs appear in
Master state at the same time.
Without that network connection there is no way for the two nodes to
communicate - and certainly no way to co-ordinate the actions of your
DRBD resource.
This is why STONITH is so highly encouraged - because without it you
end up with this sort of situation and a large probability of trashing
your data.
Any help or suggestion is greatly appreciated.
Regards,
Atanas
_______________________________________________
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
