Help with ORA-00305 error

My group is experiencing a very strange problem with our redo logs, involving error ORA-00305. Hopefully someone out there can shed a little light on what is going on.

We are using Oracle v6.037 on an AT&T GIS 3555, a 4 Pentium multiprocessing machine running UNIX System V Release 4. We use raw slices for all of our tablespaces, as well as our redo logs. We have 3 redo log slices. We also use link files for each raw slice. A link file is a special character file to (in this case) a raw device. This way, we can create a link file named
'/dev/redo1', which is much easier to remember and type than something like
'/dev/rdsk/c100t2d0s0'.

As I said, we have 3 redo logs, /dev/redo1, /dev/redo2 and /dev/redo3. Recently, our hard disk containing these three slices was repartitioned. All three slices were destroyed and recreated. Within a day following the repartitioning, the database crashed. I did not see the error which caused it to fail, but the programmer who did said that he thought that it indicated that LGWR died. A ps -ef revealed that none of the Oracle processes were running. I went into SQLDBA and tried to do a startup. I received an error, saying that I had to shutdown an instance before starting it. I tried to do a shutdown, and received the error ORA-00474, "SMON process terminated with error". I had to do a "shutdown abort" to close the database. I then tried to to a startup, and received the error ORA-00305, "database name '' is wrong in log '/dev/redo3'". At this point, I called Oracle Support. They advised me to do a manual database recovery, and then to drop logfile /dev/redo3. This worked, I was able to bring up the database.

Yesterday, I recreated logfile /dev/redo3. While doing some database maintenance (in NOARCHIVELOG mode), I received the following errors:

CREATE UNIQUE INDEX IDX_IC_ACTUALS ON IC_ACTUALS( SERVICE_DATE, TFA, SFA,

                                      *

ORA-00604: error occurred at recursive SQL level 1
ORA-00449: background process 'LGWR' unexpectedly terminated with error 305
ORA-00305: database name '' is wrong in log ''
ORA-00604: error occurred at recursive SQL level 1
ORA-00449: background process 'LGWR' unexpectedly terminated with error 305
ORA-00305: database name '' is wrong in log ''
ORA-00471: DBWR process terminated with error

ERROR at line 1:
ORA-00471: DBWR process terminated with error

ERROR at line 1:
ORA-00471: DBWR process terminated with error

ERROR at line 1:
ORA-00471: DBWR process terminated with error

Segmentation Fault - core dumped

        The database processes died at this point. Again I went into SQLDBA, tried to start the database, was told to shut it down first, received ORA-00474 when I tried a normal shutdown. Shutdown abort succeeded, startup failed with ORA-00305 "database name '' is wrong in log '/dev/redo3'" (in the prior messages, /dev/redo3 was not specified, but here it was). So I did a startup mount, and since I was NOARCHIVELOG, I did an ALTER DATABASE, dropping logifle /dev/redo3. ALTER DATABASE OPEN was successful at this point. The database is up and running again. Shutdowns and startups work normally.

        To myself, it seems that there is a problem on the UNIX/device side. The errors seem to indicate that the database cannot access the contents of /dev/redo3. On the other hand, it lets me create a logfile using /dev/redo3, so it must be able to access it at the time of creation (and deletion). On the UNIX side, nothing is apparently wrong with the raw slices and files. We've had no problems with the other two redo log files.

        As I said at the beginning of this long post, if anyone can share any knowledge related to this situation, it would be greatly appreciated. At this point, I am thinking of having /dev/redo3 created on another disk, and/or renaming /dev/redo3 to something else.

Thank you,

Lou Avrami ( attmail!lavrami )

ORA-00305 "database name '' is wrong in log '/dev/redo3'". Received on Thu Dec 01 1994 - 21:58:34 CET