Hello,<div><br></div><div>I'm trying to get stonith via SBD working on Debian Squeeze. All of the components seem to be there, and I've very carefully followed the guide at <a href="http://www.linux-ha.org/wiki/SBD_Fencing">http://www.linux-ha.org/wiki/SBD_Fencing</a> . Where things seem to fall down is that there's nothing at all in corosync's init script to start SBD, and according to the SBD page it should be started by the cluster's init script. What I end up with is Pacemaker showing that my stonith-SBD primitive is alive and running, when in reality there is no sbd process on any node at all.</div>
<div><br></div><div><div>============</div><div>Last updated: Fri Oct 21 11:17:16 2011</div><div>Stack: openais</div><div>Current DC: xen2 - partition with quorum</div><div>Version: 1.0.9-74392a28b7f31d7ddc86689598bd23114f58978b</div>
<div>2 Nodes configured, 2 expected votes</div><div>3 Resources configured.</div><div>============</div><div><br></div><div>Online: [ xen2 xen1 ]</div><div><br></div><div> historyslave3<span class="Apple-tab-span" style="white-space:pre"> </span>(ocf::heartbeat:Xen):<span class="Apple-tab-span" style="white-space:pre"> </span>Started xen2</div>
<div> historydb<span class="Apple-tab-span" style="white-space:pre"> </span>(ocf::heartbeat:Xen):<span class="Apple-tab-span" style="white-space:pre"> </span>Started xen1</div><div> stonith-SBD<span class="Apple-tab-span" style="white-space:pre"> </span>(stonith:external/sbd):<span class="Apple-tab-span" style="white-space:pre"> </span>Started xen2</div>
</div><div><br></div><div><br></div><div>Has anybody else encountered this and solved it already? I like the idea of corosync's script handling it so the cluster stack doesn't even start if the SBD device is unavailable. However, since the debian init script doesn't know of SBD, is it safe to simply start it manually on boot? If so, I suppose it needs an init script rather than rc.local, so it gets started before corosync?</div>
<div><br></div><div>Thanks,</div><div>Mark</div>