From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Lonni J Friedman <netllama(at)gmail(dot)com> |
Cc: | pgsql-general <pgsql-general(at)postgresql(dot)org> |
Subject: | Re: pg_restore with -j fails (works without -j option) |
Date: | 2011-08-26 02:41:45 |
Message-ID: | 27785.1314326505@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Lonni J Friedman <netllama(at)gmail(dot)com> writes:
> [ this doesn't work: ]
> $ cat 2011-08-25-1314280801-nightly.out | pg_restore -j2 -U lfriedman -v -d nightly
It's basically impossible for that to work. -j implies spawning
multiple processes that will be wanting to read concurrently from
different places in the input file. That cannot happen when the source
is a pipe. In this particular example, the pipe construct isn't even
doing anything useful; you could perfectly well write this instead:
$ pg_restore -j2 -U lfriedman -v -d nightly 2011-08-25-1314280801-nightly.out
> pg_restore: [custom archiver] cannot reopen stdin
> pg_restore: *** aborted because of error
I agree though that that's not a terribly friendly error message.
I thought we had some code in there to complain about non-seekable
input files, but it looks like we missed a case.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Merlin Moncure | 2011-08-26 03:04:19 | Re: passing cursors from one PL function to another |
Previous Message | Martín Marqués | 2011-08-26 00:49:59 | Re: passing cursors from one PL function to another |