Open Stack

On Thu, Sep 3, 2015 at 8:43 AM, Andrey Pavlov <andrey.mp at gmail.com> wrote:

> Hello,
>
> We have rally job with fake virt driver. And we run it periodically.
> This job runs 200 servers and measures 'show' operations.
>
> On 18.08 it was run well[1]. But on 21.08 it was failed by timeout[2].
> I tried to understand what happens.
> I tried to check this job with 20 servers only[3]. It passed but I see
> that
> operations with neutron take more time now (list subnets, list network
> interfaces).
> and as result start and show instances take more time also.
>
> Maybe anyone knows what happens?
>

Looking at the merged Neutron patches between the 18th and 21st, there's a
lot of
candidates, including QoS and work around quotas.

I think the best way to find out would be to run a profiler against Neutron
from the 18th,
and Neutron from the 21st while running the Rally tests, and finding out if
the major
bottlenecks moved. Last time I profiled Neutron I used GeventProfiler:
https://pypi.python.org/pypi/GreenletProfiler

Ironically I was having issued with the profiler that comes with Eventlet.


>
>
> [1]
> http://logs.openstack.org/13/211613/6/experimental/ec2-api-rally-dsvm-fakevirt/fac263e/
> [2]
> http://logs.openstack.org/74/213074/7/experimental/ec2-api-rally-dsvm-fakevirt/91d0675/
> [3]
> http://logs.openstack.org/46/219846/1/experimental/ec2-api-rally-dsvm-fakevirt/dad98f0/
>
> --
> Kind regards,
> Andrey Pavlov.
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150903/cfdcd275/attachment.html>