[Linux-cluster] bladecenter fencing problem

[Lode Vermeiren]

Hi all,

we are trying to set up a cluster between two IBM BladeCenter servers.
I have read the documentation numerous times, and searched the list
archives, but I have not yet found a clear explanation of all the
concepts of the clustering solution.

At the moment it almost works.. except the most important part, the
actual takeover.

What works:

* nodes A and B are online
* service is running on node A
* cluster services on node A are stopped (eg. clean shutdown of node A)
* service is restarted on node B

What doesn't work:
* nodes A and B are online
* service is running on node A
* node A fails (eg. powercut)
* service is not restarted on node B.. Node B tries to fence node A, but
this stalls. The service hangs at "stopping" in system-config-cluster.

This seems to be the same problem as Theodorus (theo at tkd.co.id) was
experiencing (see post <TKDNET9mK6PYzDabigP000000e6 at tkdnet.tkd.co.id> on
March 21 2006), but I haven't found follow-ups to that message.

Our environment:
2 x Xeon Blade, RHEL 4.3 x86_64, Bladecenter fencing

Any pointers to a solution or to what I'm doing wrong are greatly
appreciated!

Best,
Lode