Hi folks, first of all thanks for your answers/hints.
Sunil Mushran wrote:
No, one does not have to have 3 nodes when one only wants 2 nodes.If I have to speculate I would think that node1 fenced because it could not complete the disk hb io within the hbtimeout. If all you severed was the commlink, then that should not effect the disk hb traffic. That is unless the commlink isalso being used by drbd (via linux ha)....
Yes you're dramatically right.It seems that weird issue was due to the fact I used eth1, a.k.a. private link used by
DRBD, for the ocfs2 cluster config. Here's what I have now.One machine, maq1, node0 ocfs2, eth0 for public link to the network, eth1 crossover cable to maq2 for DRBD. Another machine, maq2, node1 ocfs2, eth0 for public link, eth1 crossover cable to maq1 for DRBD.
So I set up into cluster.conf the public IPs of the nodes for ocfs2 cluster.When I ifconfig down eth0 on maq2, both machines say they lost comm with other node, I/O are blocked so no split-brain config, and after 28 seconds maq2 self-fences. Maq1 doesn't, as being the lower node number of the cluster, and all is really fine.
But in mirror case, as mentioned by Gavin SCOTT on http://oss.oracle.com/pipermail/ocfs2-users/2008-January/002374.html ,if I ifconfig down eth0 on maq1 (simulating a network card failure, or even a power supply crash, ...) then both machines signal comm lost, I/O are blocked,
and after 28 seconds maq2 self-fences.I think I've understood this is the wanted behavior, maq1 being the lower node number, but the result is unfortunately my entire cluster is down... The good node is fenced (even if rebooting, it doesn't want to remount my ocfs2 data disk because not joining its
peer), and my bad node is still up with a deceased network card... Where is the trick for getting my cluster working ? The only thing I think of is :We do not use heartbeat2 from linux HA, we have written our own cluster manager.
We could check at regular intervals (<10s of ocfs2 timeout, let's say every 5 seconds for example) if the network comm between the 2 nodes is up. If not, on maq2, if network comm is still OK (checking ifconfig status, or pinging a third party such as a router), then maq2 is OK, and comm is lost between the 2 nodes because of maq1.
So on maq2, stop the ocfs2 heartbeat for avoiding self-fence, by usingocfs2_hb_ctl -K -d /dev/drbd0 (please tell me if I misunderstood this command) and remote fence maq1 (if not a power supply failure, but a network card one for example,
we power off the bad node).So our cluster will still continue to work in degraded mode, until we repair and power
up maq1, and restart o2cb and ocfs2 on both nodes.So do you think doing that could be efficient for having a strong cluster or do you have
a better idea ?Sorry for being long, but I really do not understand how could I set up a real HA 2-node cluster... and I can't really add a third node as it would need to have a shared ocfs2 device with the 2 first nodes, and drbd can only handle 2 devices...
Once more many thanks for your help and have a good week-end ! Best regards, Laurent.
begin:vcard fn:Laurent Neiger n:Neiger;Laurent org;quoted-printable:CNRS Grenoble;Centre R=C3=A9seau & Informatique Commun adr:B.P. 166;;25, avenue des Martyrs;Grenoble;;38042;France email;internet:[EMAIL PROTECTED] title;quoted-printable:Administrateur Syst=C3=A8mes & R=C3=A9seaux tel;work:(0033) (0)4 76 88 79 91 tel;fax:(0033) (0)4 76 88 12 95 note:Certificats : http://igc.services.cnrs.fr/Doc/General/trust.html x-mozilla-html:TRUE url:http://cric.grenoble.cnrs.fr version:2.1 end:vcard
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ Ocfs2-users mailing list [email protected] http://oss.oracle.com/mailman/listinfo/ocfs2-users
