[<prev] [next>] [day] [month] [year] [list]
Message-ID: <20110920174852.GA2866@nik-comp.lan>
Date: Tue, 20 Sep 2011 19:48:52 +0200
From: Nikola Ciprich <nikola.ciprich@...uxbox.cz>
To: linux-kernel@...r.kernel.org, stable@...nel.org
Cc: nikola.ciprich@...uxbox.cz
Subject: 2.6.32.42 - fsnotify_event slabs increasing constantly
Hello,
I'd like to ask for help with finding out what's constantly eating memory on one
of our servers. Today, I noticed that amount of free memory seems decreasing, and
digging a bit into it revealed that one of the slab caches is increasing constantly:
9959348 9959348 100% 0.12K 292922 34 1171688K fsnotify_event
and still slowly growing. The box is running pacemaker/corosync cluster and a lot of asterisk
SW PBX processes. Asterisks might be the culprits here, as they keep scanning spool directory each, but
to my knowledge, 1.4.x versions we're using do not support *notify yet...
I will disable this functionality anyways, because we don't need it, but I can't restart all asterisks right
now...
So my questions are:
Can I somehow find out who's using all this slabs?
Can I free somehow free them without rebooting the whole box?
I'd love to (even temporarily) solve this somehow now, and plan some maintanance window in near future, but I definitely
need to get rid of those slabs somehow, because it seems to cause quite heavy load on MM and the whole box seems to be
getting overloaded...
The box is quad core machine with 4GB of memory, running x86_64 centos 5, with 2.6.32.42.
Could somebody please to help me hunt this problem down?
thanks a lot in advance
nik
--
-------------------------------------
Ing. Nikola CIPRICH
LinuxBox.cz, s.r.o.
28. rijna 168, 709 01 Ostrava
tel.: +420 596 603 142
fax: +420 596 621 273
mobil: +420 777 093 799
www.linuxbox.cz
mobil servis: +420 737 238 656
email servis: servis@...uxbox.cz
-------------------------------------
Content of type "application/pgp-signature" skipped
Powered by blists - more mailing lists