Re: CRS problems

  • From: "David Ballester" <ballester.david@xxxxxxxxx>
  • To: development@xxxxxxxxxxxxxxxxx
  • Date: Wed, 17 Sep 2008 11:03:52 +0200

2008/9/16 Martin Bach <development@xxxxxxxxxxxxxxxxx>

> Dear all,
> I've got a playground RAC Linux 32bit installation running on
> RHEL5.0 over Oracle VM.
> This worked nicely when I initially installed it for 2 nodes. I then wanted
> to play around some more and add a third node to the cluster.
> worked fine for clusterware, it copied the sofwtare and prompted me to run
> the various scripts as root and so far so good. However, it failed to start
> the ONS resource on the 3rd (= new) node.
> 3 minutes later the whole 3rd node came tumbling down on me with core dumps
> filling up all of my partition containing $CRS_HOME up to 100%, a point
> where each daemon (crsd, cssd, evmd, ons) simply dies. I recon there is
> something wrong with the ONS configuration (onsctl ping only lists 2 nodes,
> no mention of the 3rd).
> It's not a problem caused by either iptables or selinux, both are disabled.
> Has anybody tried adding in new members to the cluster yet in 11g?
> Thanks for all input,
> Martin
> --

To be sure, you can extract the installed packages from one of the running
nodes and from the failing one

rpm -qa | sort >  /tmp/installed.rpms  ( for example )

and compare both files to see if you have the same software. Core dumps in
this case smells a failing libraries  ( the new node is the same
architecture ? )

only an idea



Other related posts: