Question - Cause of DATA BLOCK Corruption?

From: James Yorton <jjy_at_typhoon.xnet.com>
Date: 1998/07/14
Message-ID: <6ogak1$mss$1@flood.xnet.com>#1/1

We just recovered from a corrupted .dbf file which held a rollback segment. Users reported a problem, which got worse quickly. We were unable to even startup and ended up starting by bypassing the affected rollback segment using an undocumented parameter in the init.ora file. The system was deployed in 1994 and has run smoothly. This was our first corruption problem.

After startup (bypassing the corruption) ... The database was exported (no errors). Everything was dropped/removed and recreated and the data was imported (no errors).

The sysadmins tell me there was no errors during the reboot of the machine. The disks were not reformatted. And there was nothing suspicious in the /var/adm/messages file.

The scary thing is we cannot pinpoint WHY this happened in the first place. Therefore we are uncertain if this will happen again.

Some specifics about our environment:

Oracle Version 7.1.4.1.0
running on Sparc 10, Sun OS 4.1.3
Oracle's directories are NFS mounted
ran in no archive log mode (archive enabled now)
part of the DB's files resided on a disk partition that I'm told is specifically reserved for the OS (that has been fixed now.)

In the alert log:

...

Errors in file /usr/vendor/Oracle/rdbms/log/ora_16412.trc: ORA-00600: internal error code, arguments: [4414], [0], [0], [12221], [4], [], [
], []

ORA-01092: ORACLE instance terminated. Disconnection forced
ORA-06508: PL/SQL: could not find program unit being called
ORA-06512: at line 154
ORA-01092: ORACLE instance terminated. Disconnection forced

...

In that trace file:

Mon Jun 29 11:22:44 1998
Mon Jun 29 11:22:45 1998
*** SESSION ID:(31.120)
ksedmp: internal or fatal error
ORA-00600: internal error code, arguments: [4414], [0], [0], [12221], [4], [], [
], []

ORA-01092: ORACLE instance terminated. Disconnection forced
ORA-06508: PL/SQL: could not find program unit being called
ORA-06512: at line 154
ORA-01092: ORACLE instance terminated. Disconnection forced

Current SQL statement for this session:

...

Another part of the alert log:

Tue Jun 30 15:36:41 1998
Errors in file /usr/vendor/Oracle/rdbms/log/ora_5669.trc: ORA-00600: internal error code, arguments: [4194], [22], [17], [], [], [], [], [
]

ORA-00600: internal error code, arguments: [4194], [22], [17], [], [], [], [], [
]

Tue Jun 30 15:36:45 1998
Errors in file /usr/vendor/Oracle/rdbms/log/ora_5669.trc:

ORA-01578: ORACLE data block corrupted (file # 23, block # 13158)
ORA-01110: data file 23: '/usr3/burro/oracle/dbs/rbs1.dbf'
ORA-00600: internal error code, arguments: [4194], [22], [17], [], [], [], [], [

]

ORA-00600: internal error code, arguments: [4194], [22], [17], [], [], [], [], [
]

Tue Jun 30 15:36:45 1998
Errors in file /usr/vendor/Oracle/rdbms/log/pmon_5483.trc: ORA-00600: internal error code, arguments: [4194], [22], [17], [], [], [], [], [
]

Tue Jun 30 15:36:46 1998
Errors in file /usr/vendor/Oracle/rdbms/log/pmon_5483.trc:

ORA-01578: ORACLE data block corrupted (file # 23, block # 13158)
ORA-01110: data file 23: '/usr3/burro/oracle/dbs/rbs1.dbf'
ORA-00600: internal error code, arguments: [4194], [22], [17], [], [], [], [], [

]

I've opened a TAR with Oracle. But since the problem has been corrected I will be closing it. Oracle Support has helped us recover, but has no idea why it occurred. I saw nothing specific in Oracle's MetaLink support pages either.

Can anyone provide input on what we can do to prevent this from happening again? Any ideas why it happened in the first place?

Thanks.

Jim Received on Tue Jul 14 1998 - 00:00:00 CDT