Postgres Pro
Downloads
Home   >   mailing lists

Thread: Postgres does not want to start

Postgres does not want to start

From
Jason Lee
Date:
When I checked my system this morning, I could not connect to any of my databases.  After some digging, I found that it
wasreturning this error: 

DEBUG:  database system was shut down at 2002-09-14 11:53:40 CDT
DEBUG:  ReadRecord: unexpected pageaddr 0/1564000 in log file 0, segment 1, offset 5783552
DEBUG:  invalid primary checkpoint record
DEBUG:  ReadRecord: unexpected pageaddr 0/1564000 in log file 0, segment 1, offset 5783552
DEBUG:  invalid secondary checkpoint record
FATAL 2:  unable to locate a valid checkpoint record
DEBUG:  proc_exit(2)
DEBUG:  shmem_exit(2)
DEBUG:  exit(2)
DEBUG:  reaping dead processes
DEBUG:  startup process (pid 3961) exited with exit code 2
DEBUG:  aborting startup due to startup process failure
DEBUG:  proc_exit(1)
DEBUG:  shmem_exit(1)
DEBUG:  exit(1)

Any thoughts on what might cause that and how I can fix it?  It seems that it's having problems with the checkpoint
and/ortransaction logs, but I'm not sure.  Thanks! 

--
jason
Sin is real, it doesn't feel, It always, only always, steals.
Run to the cross the only joy that's real. -- downhere

Re: Postgres does not want to start

From
Tom Lane
Date:
Jason Lee <jason@theleehouse.net> writes:
> When I checked my system this morning, I could not connect to any of my databases.  After some digging, I found that
itwas returning this error: 
> DEBUG:  database system was shut down at 2002-09-14 11:53:40 CDT
> DEBUG:  ReadRecord: unexpected pageaddr 0/1564000 in log file 0, segment 1, offset 5783552
> DEBUG:  invalid primary checkpoint record
> DEBUG:  ReadRecord: unexpected pageaddr 0/1564000 in log file 0, segment 1, offset 5783552
> DEBUG:  invalid secondary checkpoint record
> FATAL 2:  unable to locate a valid checkpoint record

The page address in the xlog page seems to be 20000 hex less than it
ought to be.  This might be a dropped bit, or something worse.  Do you
have postmaster log entries from just before this startup failure began
happening?  Was there a crash, or was it an intended shutdown?  It would
be useful to run contrib/pg_controldata and see what it says.

I think you will need to run pg_resetxlog to get back to a usable state,
but it's important first to try to understand what happened, so that you
know how much damage is likely to have occurred.

            regards, tom lane