Recovering from Failure in Oracle root.sh

By Scott Jesse, Bill Burton, Bryan Vongray on September 26, 2013


When root.sh fails, it is possible to clean up and rerun it once the problem has been resolved. The command to deconfigure CRS on the local node is

   [root]# /u01/app/11.2.0/grid/crs/install/rootcrs.pl -deconfig

NOTE

The –force flag may be required if the CRS stack is not started.

This will stop any processes running and then clean up the rc.* and init.d directories, but it will not clear any data from devices that may have been written to for voting and OCR files, which you would of course do only on the last node in the cluster. The rootcrs.pl has a -lastnode flag that is used specifically for clearing these files:

   [root]# /u01/app/11.2.0/grid/crs/install/rootcrs.pl –deconfig -lastnode

If all else fails and you still need to clear the voting and OCR files, then after ensuring that clusterware is down on all nodes, you can either remove them if they are on a shared file system or write zeros to the start of the device for block devices in ASM:

 [root]# dd if=/dev/zero of=/dev/xvdb1 count=100 bs=1M

NOTE

Take great care that you have the right device. You need to clear the disk header, because this operation is not easily reversible. Also, if this device is in ASMLib, after doing dd, you will have to re-create the disk using oracleasm.

Once the deconfigure is complete and problems are resolved, the root.sh can be rerun on this node.

Related Posts

Leave a Reply