We were faced with a show stopper Ora-600
opening the database and best we could do was a partial recovery of
Luckily for me, I had tried this in a test environment and I can live
with the dataloss.
I thought I would share this experiance with you so that it would be
beneficial to someone
We are running on 126.96.36.199 on Solaris 9. We were in the process of
tuning a heavy batch run for a an application. being a test
environment, we were on noarchivelog mode.
We had set LOG_PARALLELISM to true and bounced the database and were
pursuing the tuning operation and then for a maintenance we had to
bounce this database and when we started the databse, the database
failed with Ora-600
I had raised a TAR with oracle and Oracle response was this is a bug
that is fixed in 10.2 and hence they were asking me whether should
they backport this to 9.2.06/188.8.131.52 but would lake its own sweet
Instead of waiting for the backport, I had asked them if there any
workaround possible to open the database so that I can continue with
my work. Thier reply was
set _allow_resetlogs_corruption=true, recover the database and then do
a open resetlogs. I had followed the same but again faced with the
same error. after some desparate trail and error, was able to open the
database but with a minimal dataloss.
Lesson I learnt from this exercise is that always use log_parallelism
with caution and would also suggest folks not to try this directly in
a critcal environment.