joi, 12 martie 2015

When CRS-4535 Cannot communicate with Cluster Ready Services

Usually you can workaround this situation by crsctl stop and crsctl start, but let's assume that it won't work and:

tail -f -n 50 $ORACLE_BASE/grid/log/rac2/cssd/ocssd.log

you will notice something like "has a disk hb but no network hb" in it.


Well, from now, all you should know is that you've got a network problem.
Ping all the nodes through all the networks (vip, priv, etc), check if it is a problem then:

ifdown ethX
ifupe thX
 
if the problem is still there:

ifconfig -a

mind the IPs, maybe one of the interfaces has lost the settings, usually the problem is on the priv network...

Repair the network and you will notice the crsctl start crs won't bother you again.
Enjoy!