03.10.2014 18:05, Daniel Dehennin wrote: I'd recommend to make sure that: 1. clvmd runs in 'corosync' mode, not 'openais' (controlled by -I command-line switch), because otherwise it uses buggy LCK AIS service instead of well tested CPG+DLM, 2. You have recent enough version of lvm2. 2.02.102 should be ok, you need git commit 431eda6 (https://git.fedorahosted.org/cgit/lvm2.git/commit/?id=431eda63cc0ebff7c62dacb313cabcffbda6573a), introduced somewhere between 2.02.99 and 2.02.102. I didn't test that commit with corosync-1, but that should work for it. Hope this helps, Vladislav > Digimer <lists@xxxxxxxxxx> writes: > >> Can you paste your full pacemaker config and the logs from the other >> nodes starting just before the lost node went away? > > Sorry, I forgot to attach it: > > > > > Here are the logs on the 3 hypervisors, note that pacemaker does not start at bootime: > > > > > > -- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster
