On Wed, 1 Apr 2015 04:33:07 -0700 (MST)
TonyS <tony@exquisiteimages.com> wrote:
> On Wed, April 1, 2015 12:18 am, Tom Lane-2 [via PostgreSQL] wrote:
> >
> > TonyS <tony@exquisiteimages.com> writes:
> >
> >> Running "analyze verbose;" and watching top, the system starts out
> >> using no swap data and about 4GB of cached memory and about 1GB of used
> >> memory. As it runs, the amount of used RAM climbs, and eventually the
> >> used swap memory increases to 100% and after being at that level for a
> >> couple of minutes, the analyze function crashes and indicates "server
> >> closed the connection unexpectedly."
> >
> > ANALYZE is supposed to work in a constrained amount of memory, though
> > that amount depends quite a bit on what the data is and what you've got the
> > statistics target set to.
> >
> > We've heard reports that there are memory leaks in (some versions of?)
> > PostGIS's analyze support. Whether that would apply to you would depend
> > on whether you're using PostGIS.
> >
> > Hard to say more without a lot more concrete info about what your
> > data is, what PG version you're using, etc.
> >
> > regards, tom lane
> >
>
> Thanks for the response Tom.
>
> I am not using PostGIS. The data in my system is mostly along the lines of
> what you would see in an accounts payable, accounts receivable, and
> billing type situation. Names and addresses of individuals, information
> about billing, payments received, payments sent etc.
>
> All of my indexes are b-tree indexes.
>
> Currently, the largest individual table is 1.8GB.
>
> select version() returns:
> PostgreSQL 9.3.6 on x86_64-unknown-linux-gnu, compiled by gcc (Ubuntu
> 4.8.2-19ubuntu1) 4.8.2, 64-bit
>
> OS: Ubuntu 14.04.1 LTS
>
> Physical memory: 8GB
>
> The postgresql log has these entries at the crash point:
> 2015-04-01 06:24:37 EDT LOG: server process (PID 1384) was terminated by
> signal 9: Killed
> 2015-04-01 06:24:38 EDT DETAIL: Failed process was running: analyze verbose;
> 2015-04-01 06:24:38 EDT LOG: terminating any other active server processes
>
> I started this process at 11PM, so it ran for about 7.5 hours before
> crashing.
>
> Is there anything else that would be helpful?
Don't know if I'm on the right track with this, but what is
maintenance_work_mem set to on this system?
--
Bill Moran