Invalid page header in block 592826 of 319584981

[gkhangelani]

Hi

I am junior Database Administrator, I would like to know how to handle an error like below , please explain in simple terms . This error appears very often in our client's database (Postgres 7.3.4) running on Linux - CentOS :



[root@e014400 bin]# cat /tmp/postgres.log
LOG: database system was interrupted being in recovery at 2010-03-05 16:39:17 SAST
This probably means that some data blocks are corrupted
and you will have to use the last backup for recovery.
LOG: checkpoint record is at 34B/16BBE754
LOG: redo record is at 34B/16B3552C; undo record is at 0/0; shutdown FALSE
LOG: next transaction id: 398339069; next oid: 320074847
LOG: database system was not properly shut down; automatic recovery in progress
LOG: redo starts at 34B/16B3552C
PANIC: Invalid page header in block 592826 of 319584981
LOG: startup process (pid 2983) was terminated by signal 6
LOG: aborting startup due to startup process failure

[rski]

LOG: database system was not properly shut down; automatic recovery in progress

Does the client have any problems with closing database server.

This error appears very often in our client's database

Does it mean that these errors appears when database is working (not when starting up).

[gkhangelani]

Once the error is there the user can't use the server to trade because the software disallows them.

[rski]

Yes, but the error appears when database is working not when starting up. Am I right?

[gkhangelani]

Yes, you are right.. it's when the database is doing different things

[rski]

I'm afraid it is also problem with hd. Can you do a test and try to zip a directory with databases. As I wrote earlier try to run fsck (you don't have to fix bad blocks if there are any, just check a disk)
I think it is the similar to problem you wrote int another post.

[gkhangelani]

Many Thanks, I think you are right. I will try that.

[gkhangelani]

Yesterday we had a similar problem in one of the DB servers, a developer said he REINDEXed the affected table and the problem was fixed ...