Primary DB Configuration: a 2 node RAC cluster
Solaris: 5.9 Generic_118558-25 sun4u sparc SUNW,Netra-T12
Oracle: 10.2.0.3 RAC
Veritas Cluster: Symantec/Veritas Storage Foundation for Oracle RAC
4.1 MP1 (VERITAS-4.1_p3.1:2005-10-24)
Databases: MDB and RPMP
This past Sunday we upgraded this RAC system from 10.2.0.3 to 10.2.0.3
and am now experiencing a world of hurt. Here is list of the major
problems we are experiencing:
1. DBMS_SCHEDULER jobs not running to schedule. Frequent jobs (i.e.
repeat every 20 seconds to every 5/10/60 minutes) sometime run but
mostly don't. I cannot find anything (docs, Metalink notes, google)
that talks about debugging scheduler problems. Could this be a RAC
2. Getting frequent (i.e. every couple of seconds) core dumps.ORA-07445
: exception encountered: core dump [opidsa()+480] [SIGSEGV]
[Address not mapped to object] [0x000000000]  . I believe that
this Oracle bug 5745817 with no fix available.
3. SQLNet connectivity issues. Connections are getting ORA-03113
errors. This may correlate to Step 2 core dumps but not certain.
4. For RPMP db, trying to upgrade single instance/database physical
standby to 10.2.0.3 (install new Oracle 10.2.0.3 binaries and then
apply archive logs) the recovery fails. Recover is asking for a
archive log from October 3, 2006 (when standby db built) but problem
SCN is from 2/26/2007 01:45 (from scn_to_timestamp function). I am not
sure why I need such an old archive log. My fix was to rebuild the
physical standby database.
We have a Sev 1 Metalink service request but are not making much
progress. So I am wondering if anyone else has experienced these
problems after upgrading to Oracle 10.2.0.3 on Solaris?
And as an aside we did test the 10.2.0.3 upgrade in our integration
environment and had no problems/issues.
Thanks for any help