Hi,
We got a corrupted database server, and I need some advice on what else
can I do.
Specs:
- Postgres 8.3.3 @ Linux dbserver4 2.6.24-etchnhalf.1-686-bigmem
- 4 GB RAM
- PGDATA with 700GB (70GB used), software RAID-0 with 3 sata-2
(blame it on me, ugh...)
So, the steps till now:
1 - Vacuum analyze fails this night:
2008-10-17 02:36:30 BRST: PANIC: corrupted item pointer: 44
2008-10-17 02:36:30 BRST: STATEMENT: VACUUM ANALYZE;
2 - We found the corrupted table, but the data was ok:
# vacuumdb -vz matriz -t total
PANIC: corrupted item pointer: 44
# psql matriz
matriz# select * from total;
OK
3. Tried postgres single
# pg_ctl stop
# postgres --single -D /postgres matriz
matriz# vacuum analyze
PANIC: corrupted item pointer: 44
4. reindex
# postgres --single -D /postgres matriz
matriz# reindexdb
OK
5. Again, vacuum single
# postgres --single -D /postgres matriz
matriz# vacuum analyze
PANIC: corrupted item pointer: 44
6. Tried to connect again, but...
# postgres --single -D /postgres matriz
I got that message when Postgres cant create the shared memory segment,
as if the OS has not enough shmmax. Weird.
# pg_ctl start
Again, same message complaining aboyt shared memory.
7. Reboot
8. Postgres is up, but...
# psql matriz
psql: FATAL: could not read block 0 of relation 1664/0/2676: read only
0 of 8192 bytes
9. Reset xlog
# pg_resetxlog
OK
10. Same error
# psql matriz
psql: FATAL: could not read block 0 of relation 1664/0/2676: read only
0 of 8192 bytes
Right now this is the message I got every time I try to connect, even in
single mode.
We tried to fsck the raid partition, but with no luck.
Hope there is something else I can do...
--
[]´s, ACV
