Oracle crashes immediately about the start

Question

Hi,
we stopped our database and tried to start it. For some odd reason it keeps crashing with following errors:
Recovery of Online Redo Log: Thread 1 Group 4 Seq 32013 Reading mem 0
Mem# 0 errs 0: /var/opt/oracle/oradata/DB_SID/redo4.log
Block recovery completed at rba 32013.208.16, scn 1.144749304
Mon Feb 23 12:15:14 2009
SMON: slave died unexpectedly, downgrading to serial recovery
Mon Feb 23 12:15:14 2009
Errors in file /var/opt/oracle/admin/DB_SID/bdump/DB_SID_smon_1344.trc:
ORA-07445: exception encountered: core dump [kdr9ir2rst0()+443] [SIGSEGV] [Addre
ss not mapped to object] [0x7FBDDFFFF8] [] []
Mon Feb 23 12:15:17 2009
Errors in file /var/opt/oracle/admin/DB_SID/bdump/DB_SID_pmon_1332.trc:
ORA-00474: SMON process terminated with error
Mon Feb 23 12:15:17 2009
PMON: terminating instance due to error 474
Instance terminated by PMON, pid = 1332
Last time when it happened I managed to start the database succesfully by turning off parallel recover, but now it doesn't help. Any idea how to workaround that? This is oracle 10.2.0.1 on Red Hat EL4. I know that on metalink there are several bugs covering that and suggestions that the bug is fixed in newer versions, but I would like to get the database to runnable state, because the client doesn't have valid backups.
Regards,
Jarek

Answer 1

HI,
You most probably have to contact Oracle Support for this since this seems to be related to Xsaction recovery.
Anyway, try the following before requesting oracle support...
1. Start your instance
2. Check the bdump for any trace files
3. Check your UNDO tablespace
4. Reboot your machine if possible.
try these steps and if failed..contact oracle support