[Spacewalk-list] Apache2::SizeLimit too big

[Jesus M. Rodriguez]

On Fri, Jan 16, 2009 at 7:33 PM, Brian Kosick <bkosick at mxlogic.com> wrote:
> Hi All,
>
> I just upgraded to 0.4 from 0.3, I was seeing these errors under 0.3 as
> well.
>
> This is on a server that runs spacewalk and a mercurial repo.  The Specs
> are
>
> Dual Dual Core Xeons, 4GB of RAM, 8GB Swap and a 3TB array.
>
> It mostly occurs when I try to delete software packages... anywhere from
> the default page display of 50 (-500)  Though I can be doing other
> things as well like trying to manage/deploy configs to more than a few
> servers at a time.
>
>
> [Fri Jan 16 17:25:13 2009] (29909) Apache2::SizeLimit httpd process too
> big, exiting at SIZE=444864 KB  SHARE=12964 KB  REQUESTS=2 LIFETIME=0
> seconds
> [Fri Jan 16 17:25:13 2009] (29912) Apache2::SizeLimit httpd process too
> big, exiting at SIZE=444996 KB  SHARE=12964 KB  REQUESTS=2 LIFETIME=0
> seconds
> [Fri Jan 16 17:25:14 2009] (29910) Apache2::SizeLimit httpd process too
> big, exiting at SIZE=444952 KB  SHARE=12800 KB  REQUESTS=2 LIFETIME=0
> seconds
> [Fri Jan 16 17:25:14 2009[Fri Jan 16 17:25:14 2009] (29911)
> Apache2::SizeLimit httpd process too big, exiting at SIZE=445084 KB
> SHARE=12892 KB  REQUESTS=2 LIFETIME=0 seconds
> [Fri Jan 16 17:25:14 2009] (29922) Apache2::SizeLimit httpd process too
> big, exiting at SIZE=445084 KB  SHARE=12896 KB  REQUESTS=2 LIFETIME=0
> seconds
> [Fri Jan 16 17:25:14 2009] (29913) Apache2::SizeLimit httpd process too
> big, exiting at SIZE=445084 KB  SHARE=12896 KB  REQUESTS=2 LIFETIME=0
> seconds
> ] (29914) Apache2::SizeLimit httpd process too big, exiting at
> SIZE=445084 KB  SHARE=12896 KB  REQUESTS=2 LIFETIME=0 seconds
>
> Any suggestions?
>

Sadly that's a bug we've been chasing for quite some time. Most recently
reported here: https://bugzilla.redhat.com/show_bug.cgi?id=465796

I think it's a memory leak because if you increase the SizeLimit it
will continue
to grow, I haven't been able to find a value where it stabilizes. If
you have the
memory, you might want to try setting it to 1_024_000 to see if it helps.

It seems to also only affect the perl stack, as the python and java seem to
do well with memory (at least they seem to level off at some maximum
threshold).

Another tidbit is this ONLY affects 64-bit Spacewalk. We have not been able
to get this to happen on a 32-bit version EVER.  So any data you can give us
would be helpful, feel free to add your comments and observations to the but
mentioned above.

Sincerely,
jesus