Root.sh fails while installing 11.2.0.3 Grid Infrastructure

Had this issue with root.sh failing on node2 while building a multi-node Linux 11.2.0.3 Grid Infrastructure installation recently and received ORA-15018, ORA-15072 and ORA-15080 which are ASM related.

However the scandisks and listdisks were working fine on all the nodes and was listing the ASM disks correctly.

Further investigation revealed that on node2 where the root.sh was failing with the above errors, the /usr/sbin/oracleasm querydisk -d disk_name shows the device as '/dev/sdc1' whereas all other nodes were showing the correct multipath device names starting with dm-xx.

Changing the ORACLEASM_SCANORDER & ORACLEASM_SCANEXCLUDE parameters as below followed by a restart of asmlib too didnt work.

ORACLEASM_SCANORDER="dm"
ORACLEASM_SCANEXCLUDE="sd"

This has been documented in MOS note 1059847.1 which is not updated in the standard installation doc for Linux yet.

So as a final resort, I've deconfigured the clusterware on node2 using rootcrs.pl -deconfigure with force option followed by a restart of asmlib and that picked up the right multipath device for the ASM disk.

#$GRID_HOME/crs/install/rootcrs.pl -deconfig -force

After this step, root.sh worked perfectly fine on this node and all other remaining nodes.