Re: BUG #6785: Memory Leak in plpgsql

On Mon, Jul 30, 2012 at 05:56:22PM +0200, Andres Freund wrote:
> Hi,
>
> On Monday, July 30, 2012 05:38:07 PM Anderson Valadares wrote:
> > I understand, but the memory should not be returned after the execution of
> > the function?
> Well, that depends on how memory was allocated by the libc. When it used brk()
> to allocate memory its rather likely that the memory cannot directly be
> returned because some block of memory in the new memory is still used by some
> permanent memory context.
>
> > Below is the result of running with more than 55,000 calls.
> >
> > PID USER PR NI VIRT SWAP RES CODE DATA SHR S P %CPU %MEM TIME+
> > COMMAND
> > 618 postgres 15 0 1687m 1.2g 507m 4684 505m 3796 S 1 0.0 4.2 0:04.90
> > postgres: postgres test [local] idle
> Interesting. I just let the thing run - by accident - for 30+ minutes and it
> still hovered at 96MB.

FYI, I did a blog entry that mentions when memory is returned:

http://momjian.us/main/blogs/pgblog/2012.html#February_1_2012

Specifically, only single memory allocations greater than MMAP_THRESHOLD
are returned to the operating system.

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ It's impossible for everything to be true. +