[Oisf-devel] Mem leaks

[Martin Holste]

Ok, thats what I thought for workers.

Rev 58 seems stable, though the mem use is larger than a few revs ago.

On Thu, Oct 6, 2011 at 1:00 PM, Eric Leblond <eric at regit.org> wrote:
> Hello,
>
> On Thu, 2011-10-06 at 18:55 +0100, Chris Wakelin wrote:
>> I've tried running the latest git master with PF_RING autofp (I didn't
>> hit Martin's problem as I'd specified interface and cluster_id on the
>> command line) on both machines (students and campus) and it seems pretty
>> OK.
>>
>> Memory on both reached 9.8GB and stayed there, and killing suricata on
>> the students machine (after 40 minutes) didn't segfault, I think, though
>> I had core dumping disabled.
>>
>> I'm now trying "workers" with 6 threads. It seems broadly similar to
>> Will's "single" mode. Is that correct? What is the difference between
>> these and "single" in the git master?
>
> None :) but "single" is mono thread and "workers" is a set of "single"
> threads.
>
> BR,
>
>>
>> Best Wishes,
>> Chris
>>
>>
>> On 06/10/11 15:46, Chris Wakelin wrote:
>> > I was wondering whether Anoop's ac fixes (b7b580) might be the issue,
>> > but I'm not at all sure. It might be that my cores sizes are bigger
>> > anyway because I switched from b2g to ac. Getting them every time I
>> > killed suricata certainly brought their size to my attention :(
>> >
>> > I'll try the latest git master for a bit! Might even try runmode=worker :)
>> >
>> > Best Wishes,
>> > Chris
>> >
>> > On 06/10/11 15:32, Victor Julien wrote:
>> >> On 10/06/2011 04:30 PM, Martin Holste wrote:
>> >>> Yeah, maybe, but the baseline is almost double.  Do you want me to
>> >>> switch back to master for awhile?
>> >>
>> >> If you can, please. The bigger baseline is certainly unintentional, so
>> >> we'll look into that as well.
>> >>
>> >> Btw, I think Eric fixed the pfring shutdown thing as well in the updated
>> >> master.
>> >>
>> >> Cheers,
>> >> Victor
>> >>
>> >>> On Thu, Oct 6, 2011 at 9:25 AM, Victor Julien <victor at inliniac.net> wrote:
>> >>>> So could it be that the master just has a higher baseline but stops
>> >>>> growing at some point as well? The higher baseline might be explained by
>> >>>> some AC pattern matcher updates.
>> >>>>
>> >>>> On 10/06/2011 02:59 AM, Martin Holste wrote:
>> >>>>> bc5, still no leaks.
>> >>>>>
>> >>>>> On Wed, Oct 5, 2011 at 7:54 PM, Victor Julien <victor at inliniac.net> wrote:
>> >>>>>> On 10/05/2011 11:41 PM, Martin Holste wrote:
>> >>>>>>> Seems to be no leaks thus far after 4 hours.
>> >>>>>>
>> >>>>>> So what git ref is it you are running now exactly? Lost track.
>> >>>>>>
>> >
>>
>>
>
> --
> Eric Leblond
> Blog: http://home.regit.org/
>
> _______________________________________________
> Oisf-devel mailing list
> Oisf-devel at openinfosecfoundation.org
> http://lists.openinfosecfoundation.org/mailman/listinfo/oisf-devel
>
>