<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
</head>
<body bgcolor="#ffffff" text="#000000">
Hi,<br>
<br>
On RHEL6 :<br>
I'm configuring OCFS2 under Pacemaker, and it requires to <br>
use the <b>Stack pcmk</b> instead of the Stack o2cb. <br>
Once I successfully start the clone-dlm and clone-o2cb under Pacemaker
on both nodes,<br>
I launched some test loops of concurrent access writing on a shared
LUN, <br>
just a cp -R of files to a LUN, but same files and same LUN on<br>
both nodes. <br>
The result is :<br>
On node1 , it successfully writes the files and loops a few minutes<br>
but on node2, even the first cp -R remains stalled ...<br>
until I stop the loop on node1 and just after the loop on node 2 <br>
works fine ... <br>
<br>
And when I do the same test but without interrupting the loop on<br>
node1, so the loop on node2 remains stalled at first iteration, but<br>
after a few minutes, the loop on node1 gets also stalled ! And so<br>
the write are definitively stalled on both nodes ...<br>
so it seems there is a lock management problem ?<br>
<br>
Note that exactly the same tests with ocfs2 started in a "standard way"<br>
(with Stack o2cb) are working fine with each node writing in parallel<br>
on same shared LUN.<br>
<br>
Is this an already known problem ?<br>
<br>
<b>By the way,</b> <b>is there a documentation somewhere</b> where I
could<br>
get information about <b>pcmk</b> <b>stack versus o2cb stack</b> ?
the "why"<br>
of both stacks, and why it is mandatory to use pcmk stack when<br>
configuring ocfs2 under Pacemaker ?<br>
<br>
Thanks <br>
Regards<br>
Alain<br>
<br>
</body>
</html>