[Openstack-operators] Neutron nf_conntrack performance

John Edrington jle at cosmicfire.net
Sat Aug 16 13:52:42 UTC 2014


Hi Stuart, 

I had a similar sounding problem on CentOS running icehouse (ksoftirqd/0 at 100%, tcpdump showed GRE router advertisement flood). The workaround I found was to disable ipv6 by editing /etc/sysctl.conf.

John

Sent from my mobile device.

----- Reply message -----
From: "Stuart Fox" <stuart at demonware.net>
To: "openstack-operators at lists.openstack.org" <openstack-operators at lists.openstack.org>
Subject: [Openstack-operators] Neutron nf_conntrack performance
Date: Sat, Aug 16, 2014 2:44 AM

Hey other ops!
Im having a serious problem with my neutron router getting spin locked in nf_conntrack_tuple_taken.
Has anybody else experienced it?
As the incoming request rate goes up, so nf_conntrack_tuple_taken runs very hot on CPU0 causing ksoftirqd/0 to run at 100%. At that point internal pings on the GRE network go sky high and its game over.


Ubuntu 14.04/Icehouse 2014.1.1 on an ibm x3550 with 4 10G intel nics.
eth0 - Mgt
eth1 - GRE
eth2 - Public
eth3 - unused 

Any help very much appreciated!


BR,
Stuart
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20140816/90ad70cd/attachment.html>


More information about the OpenStack-operators mailing list