The following is an error I got on my first RAC install. The solutions were simple, but I found it very frustrating figuring out the solution. For one I got a tip from another Oracle user and the second I just got lucky figuring out what was wrong since I was sure I’d checked it. In the second case it is clearly something Oracle could do a better job of reporting, and would be very simple to do. I can’t say for sure without testing each one independently if both caused this, but I did solution #1 first and then solution #2 and it resolved it. So, try these if you’re experiencing the same problem and are sure you have connectivity including ssh.
FYI – In this particular situation it was RedHat 5.3 and Oracle 220.127.116.11.0, but I’ve now heard the same stories with 11.2.0.x.0.
Here is the error:
PRVF-7616 : Node connectivity failed for subnet "18.104.22.168" between "flux-rac-node-wcdp-02 - bond0 : 22.214.171.124" and "flux-rac-node-wcdp-01 - bond0 : 126.96.36.199"
Turn off the ssh banner. This is a relatively simple task – as root on each node,
vi /etc/ssh/sshd_config #Banner /etc/issue.net:wqservice sshd stop service sshd start
Make sure you can “ping” each server from each server as the oracle user. As it happens, I had verified “ping” worked from every server to every other server, but as the “root” user – oops. This was a simple fix:
chmod o+rx /bin/ping
Some might not like this option, but it works and is simple, and you’d have to make it in through a lot of security, so I’m guessing ping is the least of the worries.