crash with latest git HEAD

Started by Jon Nelsonalmost 12 years ago2 messagesbugs
Jump to latest
#1Jon Nelson
jnelson+pgsql@jamponi.net

The following query results in a crash, ultimately due to the process
running out of memory:

create table blah as select md5(generate_series(1,50000000)::text) as a;

I'm running 602b27ab8e45fbb07cf5b275b0593b38190232e4.

I haven't yet done any git bisecting.

--
Jon

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs

#2Tom Lane
tgl@sss.pgh.pa.us
In reply to: Jon Nelson (#1)
Re: crash with latest git HEAD

Jon Nelson <jnelson+pgsql@jamponi.net> writes:

The following query results in a crash, ultimately due to the process
running out of memory:
create table blah as select md5(generate_series(1,50000000)::text) as a;

The behavior looks the same to me in either 9.3 or HEAD: it does consume a
lotta memory (about 7GB) but eventually finishes. If it's crashing for
you, you probably need to look at your system-level configuration,
particularly the per-process memory ulimit level vs. available RAM/swap
and whether the dreaded OOM killer is enabled.

The reason it's eating so much memory is you've got a profligate SRF in
the SELECT's targetlist. Try it as, eg,

create table blah as select md5(g::text) as a from generate_series(1,50000000) g;

While it's be nice to not have SRFs in the targetlist be a memory problem,
it's not an easy thing to do, and I doubt anyone is going to look very
hard for a fix for what is basically a deprecated feature.

regards, tom lane

--
Sent via pgsql-bugs mailing list (pgsql-bugs@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-bugs