> Hi, > > On Fri, Feb 29, 2008 at 12:34:28AM +0100, [EMAIL PROTECTED] wrote: >> > It will reboot the other node if you have a STONITH device configured. >> > Do you have one? >> >> No, I don't. >> Yes I know, I need a stonith device but I don't have a real one yet. >> >> Anyway even without should the node which is not able to stop its >> resources not reboot voluntary?? I mean instead of ignoring this fault?! >> >> Is there a way to do this? > > No. The quote is true, but only for older v1 style clusters where > the resource manager would reboot the node. With CRM/v2 you need > a stonith device.
Thanks for making that clear. It could be quite handy to give the crm the capabilities to reboot the machine, cause in my mind the stonith device is the last resort. So if the crm detects it can not stop a resource it definitaly should shoot itself. Furthermore I found at http://www.linux-ha.org/CIB/Idioms: Causing a node to reboot when a resource stop fails <op id="dummy-resource-stop-id" name="stop" on_fail="fence"/> This is a operation for crm. How is this supposed to work? If I would have a stonith device, would that node do suicide, or ask others to shoot? I tried it and the node doesn't reboot either. Actually I have the following scenario on Node A and B (2-node-cluster): A has Filesystem | B has nothing /* Filesystem is a AoE Network Filesystem */ A disconnected from network | B is still connected -> split brain condition forced A wants to shoot B | B wants to shoot A /* they are trying to shoot each other at the same time!?! */ /* lets say A was faster than B /* A checks pingd | B is dead /* Filesystem needs pingd connectivity */ A tries to stop Filesystem /* A cant stop Filesystem, cause its a Network Filesystem */ A ignores this error instead of rebooting itself -> Now resource is off and both nodes are useless Someone has an idea how to solve this? Your help is really appreciated! Cheers, Holger > > Thanks, > > Dejan > >> Cheers, >> Holger >> >> > >> > On Thu, Feb 28, 2008 at 12:16 PM, <[EMAIL PROTECTED]> wrote: >> >> Hi, >> >> >> >> I read at http://www.linux-ha.org/HeartbeatResourceAgent that >> >> "After repeated stop failures, we will do a status on the resource. >> If >> >> the >> >> status reports that the resource is still running, then we will >> reboot >> >> the >> >> machine to make sure things are really stopped." >> >> >> >> I have a 2 node cluster and the nodes are connected per >> >> Ata-over-Ethernet >> >> to some Filesystems. Each resource depends on pingd and will be >> stopped >> >> if >> >> there is no connectivity. >> >> >> >> So when I disconnect node mrbrown from the network I'll get a >> >> split-brain >> >> condition. Because of the pingd constraint mrbrown tries to stop the >> >> Filesystem-resource. This will timeout or produce errors because the >> >> Filesystem is not reachable. >> >> But mrbrown doesn't reboot as it should. It case of ignores that >> >> error?!? >> >> Did I made a mistake? >> >> >> >> Btw. there is no real stonith device at the moment, just ssh and >> this >> >> will >> >> not work without network. You can find some output of mrbrown in >> close. >> >> >> >> Any help is really appreciated.. >> >> Cheers, >> >> Holger >> >> >> >> _______________________________________________ >> >> Linux-HA mailing list >> >> [email protected] >> >> http://lists.linux-ha.org/mailman/listinfo/linux-ha >> >> See also: http://linux-ha.org/ReportingProblems >> >> >> > >> > >> > >> > -- >> > Serge Dubrouski. >> > _______________________________________________ >> > 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 > _______________________________________________ > 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
