Thread: FATAL: bogus freespace amount

FATAL: bogus freespace amount

From
"Carlos H. Reimer"
Date:
Hi,
 
We are facing the following problem during the PostgreSQL 8.2.4 startup in a Windows XP/SP2 box:
 

2008-03-31 13:35:12  1: LOG:  database system was interrupted at 2008-03-31 09:45:44
2008-03-31 13:35:12  2: LOG:  checkpoint record is at 0/7588FFD8
2008-03-31 13:35:12  3: LOG:  redo record is at 0/7588FFD8; undo record is at 0/0; shutdown TRUE
2008-03-31 13:35:12  4: LOG:  next transaction ID: 0/100487; next OID: 909454
2008-03-31 13:35:12  5: LOG:  next MultiXactId: 1; next MultiXactOffset: 0
2008-03-31 13:35:12  6: LOG:  database system was not properly shut down; automatic recovery in progress
2008-03-31 13:35:12  7: LOG:  record with zero length at 0/75890038
2008-03-31 13:35:12  8: LOG:  redo is not required
2008-03-31 13:35:12  9: LOG:  database system is ready
2008-03-31 13:35:12  10: FATAL:  bogus freespace amount
2008-03-31 13:35:12  1: LOG:  startup process (PID 856) exited with exit code 1
2008-03-31 13:35:12  2: LOG:  aborting startup due to startup process failure
2008-03-31 13:35:13  1: LOG:  logger shutting down

What can be done to bring the system up again?
 
Thank you in advance!
 
Reimer
 

Re: FATAL: bogus freespace amount

From
Tom Lane
Date:
"Carlos H. Reimer" <carlos.reimer@opendb.com.br> writes:
> 2008-03-31 13:35:12  10: FATAL:  bogus freespace amount

This probably means $PGDATA/global/pg_fsm.cache has gotten trashed.
Fortunately that's just a cache --- you should be able to delete the
file and start up.

            regards, tom lane