[Linux-cluster] clustat stuck

frederic randriamora frederic at ovsg.univ-ag.fr
Fri Oct 29 23:30:26 UTC 2010


I have a 4 node cluster, with multipathed qdisk on a san. The nodes are 
running redhat 5.4.

After a minor change made in cluster.conf on node3 properly propagated 
by ccs_tool update, clustat is no longer correctly responding in the 
other 3 nodes.
node3 is neither nodeid 1 nor qdisk master.

clustat on node3 runs fine

clustat on the other nodes

either hangs with
connect(8, {sa_family=AF_FILE, path="/var/run/cluster/rgmanager.sk"...}, 110
from strace

or times out with
Timed out waiting for a response from Resource Group Manager
without displaying the still running services

cman_tool services et al. are just fine everywhere,

Although all the services are running fine, I cannot move/stop them 
anymore with clusvcadm.

How to get out of that situation?

The change I made was adding "use_virsh=0" to cluster.conf for some vm 
definitions in cluster.conf

Thanks a lot

More information about the Linux-cluster mailing list