recovery after interrupt in the middle of a previous recovery

Or Kroyzer <orkroyzer@gmail.com> writes:

I am using postgres 8.3.1,

... you really ought to be using 8.3.something-recent ...

and have implemented warm standby very much like
the one described in the high availability documentation on this site.
It seems to work well except for this problem: I've had a case where the
postgresql server was interrupted while in recovery (I think it was a user
interrupt, the log sais:

. LOG: received fast shutdown request
LOG: archive recovery complete
FATAL: terminating connection due to administrator command
LOG: startup process (PID 6033) exited with exit code 1
LOG: aborting startup due to startup process failure

And after that, pg doesn't go through the recovery script provided in
recovery.conf, and doesn't manage to come up. The log sais:

LOG: database system was interrupted while in recovery at log time
2010-05-26 02:00:03 IDT
HINT: If this has occurred more than once some data might be corrupted and
you might need to choose an earlier recovery target.
LOG: could not open file "pg_xlog/000000CA0000000A0000006D" (log file 10,
segment 109): No such file or directory
LOG: invalid primary checkpoint record
LOG: could not open file "pg_xlog/000000CA0000000A0000006D" (log file 10,
segment 109): No such file or directory
LOG: invalid secondary checkpoint record
PANIC: could not locate a valid checkpoint record
LOG: startup process (PID 8081) was terminated by signal 6: Aborted
LOG: aborting startup due to startup process failure

Hmm. Try putting back your recovery.conf file --- it will have been
renamed at the point where "archive recovery complete" was printed.
This example suggests that we might be doing that too early.

regards, tom lane

Thanks.

2010/5/26 Tom Lane <tgl@sss.pgh.pa.us>