Re: [HACKERS] SIGSEGV on CREATE FUNCTION with plpgsql

> Wayne Piekarski <wayne(at)senet(dot)com(dot)au> writes:
> > So can someone give me some advice? Do you want me to provide a stack dump
> > or something?
>
> A stack trace might help --- I'm not sure why you are seeing this
> problem if there are no functions approaching 4k of text.

Ok, I've got a test postgres set up which I used for my profiling so I'll
have a play and get a stack dump and see if I can work out whats causing
this. Been busy lately so haven't had a chance ... more later on this.

> > Right now I'm reluctant to play with my plpgsql functions because I'm
> > scared its going to die and I wont be able to reload them back in.
>
> You can play with them in a playpen installation... I wouldn't do that
> sort of testing on a production installation either. A playpen is
> always a good thing to have. Note you can put multiple playpens on
> one machine --- all you need is a separate data directory and socket
> number for each one.

I've got multiple testing databases, but due to the random nature of the
problem, the functions will reload normally, but when the pg_dump output
reloads them, it does it in a different order and dies. So it is dependent
on order and a bunch of other things, and so even with testing, I still
can't be sure it won't break the real production database.

I'll have a look through the code and see if I can spot something obvious.
My functions are all <= 1k and so I'm miles away from 2k or 4k problems
with btree indices.

thanks,
Wayne

------------------------------------------------------------------------------
Wayne Piekarski Tel: (08) 8221 5221
Research & Development Manager Fax: (08) 8221 5220
SE Network Access Pty Ltd Mob: 0407 395 889
222 Grote Street Email: wayne(at)senet(dot)com(dot)au
Adelaide SA 5000 WWW: http://www.senet.com.au