Standalone Grid Infrastructure is not upgraded properly after 11201 to 11202 upgrade
I had to upgrade Oracle 184.108.40.206 Database to Oracle 220.127.116.11 running on Linux x86-64. The datafiles and FRA are on ASM so the steps for upgrade are first to upgrade Grid Infrastructure (GI) and then the database. I will not go into details for the whole upgrade but I will mention the steps roughly for completeness:
1)6880880 Opatch 11.2 in $GI_HOME
2)9655006 to the 18.104.22.168 GI
3) Oracle GI to 22.214.171.124 patch 10098816(p10098816_112020_Linux-x86-64_3of7.zip)
4) Database software for 126.96.36.199(p10098816_112020_Linux-x86-64_1of7.zip, p10098816_112020_Linux-x86-64_2of7.zip)
The main subject of this post is in fact the thing which I will remember after the successful upgrade and it will be on the top of my head for some time. Guess what it is?
All went well, GI and database upgraded to 188.8.131.52 successfully on Red Hat Enterprise Linux 64 bit. But guess what, after subsequent reboot standalone Grid Infrastructure is not starting. Moreover it tried to start old grid.
After some investigation I tried to remove old grid, but the deinstall is not working. Metalink points some manual procedure for reinstalling but at the time of the installation it was not available and posted in Metalink. So I detached and then removed old grid home and database home with rm –rf . This cleaned the environment but it did not help.
The new Grid Infrastructure is not starting. I noticed that “crsctl start has” is hanging.
After some investigation it turned out that the upgrade scripts did not do the upgrade clean. In case of standalone GI root upgrade script finished successfully but missed to copy ohasd scripts to the right place. It is reported as a Bug 10167269 STANDALONE GI INSTALL: INIT.OHASD/OHASD NOT UPDATED AFTER 11201 TO 11202 UPGRADE.
The workaround to manually copy ohasd script worked out for me. This is what I did :
cp $GRID_HOME/crs/init/init.ohasd /etc/init.d/init.ohasd
cp $GRID_HOME/crs/init/ohasd /etc/init.d/ohasd
kill –HUP 1
crsctl start has
In case you happen to know this before the upgrade you can try to fix the script in 184.108.40.206 $GRID_HOME/crs/install/roothas.pl in advance.
I am glad finally all went well. Otherwise, the alternative is to start fresh 220.127.116.11 installation from the p10098816_112020 patch.