Kris Jurka <books@ejurka.com> writes:
> This also produces a second core file in the directory the postmaster was
> started in which contains:
> #0 0x402a5721 in kill () from /lib/libc.so.6
> #1 0x402a54c5 in raise () from /lib/libc.so.6
> #2 0x402a69e8 in abort () from /lib/libc.so.6
> #3 0x08231994 in errfinish (dummy=0) at elog.c:473
> #4 0x080b1e2c in ReadControlFile () at xlog.c:3130
> #5 0x080b2257 in XLOGShmemInit () at xlog.c:3381
This appears to be here:
ereport(PANIC,
(errcode_for_file_access(),
errmsg("could not open control file \"%s\": %m",
ControlFilePath)));
but it's difficult to go any further without knowing the errno.
I certainly can't think of any good reason why the recovery process
should be unable to open pg_control. Do you have the postmaster log?
regards, tom lane