Excerpts from Magnus Hagander's message of jue ago 16 09:46:28 -0400 2012:
> On Thu, Aug 16, 2012 at 3:29 PM, Alvaro Herrera
> <alvherre@2ndquadrant.com> wrote:
> >> This is hte message with messageid being delivered into majordomo.
> >> Unfortunately, majordomo has no logging at all that I know of. But
> >> Alvaro has sometimes been able to track down where messages are as
> >> they go through there :) Alvaro? (our local session id on the
> >> majordomo server is 1T1kSS-0000A1-Fh)
> >
> > What do you mean majordomo has no logging? It sure does. In fact, we
>
> I didn't say that. I said "that I know of".
>
> I've looked *everywhere* for a logfile.
>
> Silly me thinking a unix perlscript would have an actual logfile. Per
> what you write here, it seems to be in one of the databases...
Well :-) It's lists/postgresql.org/GLOBAL/_log, so no wonder you didn't
find it. It is almost a plain-text file; the fields in there are
separated by ^A. You can actually see it and grep in it, though
mj_shell seems a reasonable interface to decode it.
> > So we know it *failed*, though we don't know *why* ... it does seem that
> > Majordomo does not log this info.
>
> Hah. that would be the most important reason for it to log it of course..
No doubt.
> Is there a way to find out how big it was? IIRC, depesz wrote
> something about cutting the size down for the second message - maybe
> it got rejected because it was too big?
The maximum allowed message length seems to be 40000 bytes (except for
lists that have raised it explicitely such as pgsql-hackers). Should we
raise that limit?
--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services