Yesterday, I came across a small problem with an Oracle RAC installation using ASM. After a restart of one of the three (Windows Server 2003) nodes, three of the five database instances did not start automatically…
Running ‘crs_stat’ showed status OFFLINE and target ONLINE for these instances, but strangely there was no entry in the alert log telling these instances did not start because of ’something’ was wrong. It even looked like that they did not got any signal to start at all! OK, so I checked the CRS log for any strange behavior and it showed ASM finishing after the database instances gave up starting.
A quick search on the ‘new and improved’ My Oracle Support (old Metalink), resulted in running ‘crs_stat -p’ and check for [REQUIRED_RESOURCES=] for each instance. It showed that for the particular node, ASM was not required for these three database instances. Modifying this with ’srvctl modify instance -d db1 -i node1 -s +ASM1′ (for example), did the trick.
I found this solution on Metalink: Doc ID: NOTE:276208.1 “CRS is not starting first the ASM instance”, although I can not imagine hitting this bug on 10.2.0.4.