RE: RAC - Powering on One Node Panics all other nodes
Date: Tue, 26 Jul 2011 17:49:48 +0100
We had an issue where Solaris servers that are using IPMP for the network interfaces to make them "high available" picks an IP address to connect to to check that the network is working ok. It turns out that nodes 1 and 2 of the 3 node cluster were using server 3's IP so when server 3 was rebooted it paniced nodes 1 and 2! This was for the interconnect network. The resolution was to setup a gateway on the switches and tell IPMP to use this to check it's connectivity rather than a random selected IP.
You could ask your Solaris admins to check this out perhaps?
Date: Mon, 25 Jul 2011 07:04:11 -0400
Subject: RAC - Powering on One Node Panics all other nodes. From: Ram Srinivasan <srinivasanram2004_at_gmail.com>
We have a 4-node RAC cluster with about 4-5 databases running. These are M4000 Solaris 10 - 4CPUs - 64 GB - 8 processors Last time when the sysadmins rebooted the machines, they complained that powering up and rebooting the node 4 panics all the other nodes, and the crs is going down in all the other nodes. They can't give any other details. So, they have kept the node 4 powered off to keep the other 3 nodes running, and they suggest that node 4 be removed from the cluster. Even the removal of the node they want to do it but they want us to teach them how to do it, as we do not have "sudo /bin/bash" priv.
databases are at 10.2.0.4.4
crs is at 10.2.0.4.0.
- Did anyone have this kind of problems, where 'powering on one node' panics all the other nodes with regard to crs?
- Does the "Addition/Removal of nodes from the clusters fall under DBAs' activities, or sysadmins' activities?
http://www.freelists.org/webpage/oracle-l Received on Tue Jul 26 2011 - 11:49:48 CDT