<p dir="ltr"><br>
On Dec 10, 2013 7:38 PM, "Joshua Harlow" <<a href="mailto:harlowja@yahoo-inc.com">harlowja@yahoo-inc.com</a>> wrote:<br>
><br>
> Yup, IMHO this is where the rally[1] project, and others that attempt to<br>
> push the boundaries are really helping openstack become much better around<br>
> these types of issues. At yahoo! in a short-time it will not be that<br>
> uncommon to boot 100+ VMs concurrently (traffic spikes, other random QA<br>
> testing that requires more than 10 instances...). So the earlier these<br>
> issues get solved then the better it will be for everyone (and they likely<br>
> are not simple issues to solve, sadly).</p>
<p dir="ltr">I don't think this issue in particular is very hard to fix for nova-networking but I don't know enough about neutron to speak for it as well. I included a possible route to fixing this in the bug itself.</p>
<p dir="ltr">><br>
> [1] <a href="https://wiki.openstack.org/wiki/Rally">https://wiki.openstack.org/wiki/Rally</a><br>
><br>
> On 12/10/13 10:17 AM, "Edgar Magana" <<a href="mailto:emagana@plumgrid.com">emagana@plumgrid.com</a>> wrote:<br>
><br>
> >FYI. Concurrent calls are a problem all around in OpenStack!<br>
> ><br>
> >Edgar<br>
> ><br>
> >On 12/10/13 8:47 AM, "Abhishek Chanda" <<a href="mailto:abhishek@cloudscaling.com">abhishek@cloudscaling.com</a>> wrote:<br>
> ><br>
> >>** Changed in: nova<br>
> >> Status: New => Triaged<br>
> >><br>
> >>--<br>
> >>You received this bug notification because you are subscribed to<br>
> >>neutron.<br>
> >><a href="https://bugs.launchpad.net/bugs/1257788">https://bugs.launchpad.net/bugs/1257788</a><br>
> >><br>
> >>Title:<br>
> >> nova-network doesn't handle large amount of concurrent booting<br>
> >> instances well<br>
> >><br>
> >>Status in OpenStack Neutron (virtual network service):<br>
> >> Triaged<br>
> >>Status in OpenStack Compute (Nova):<br>
> >> Triaged<br>
> >><br>
> >>Bug description:<br>
> >> As seen by the large-ops test nova-network has trouble booting 150 VMs<br>
> >> concurrently. Part of the problem is that setting up the network for<br>
> >> each instance is relatively expensive with many locks and root-wrapped<br>
> >> calls. Both of which significantly hurt performance. One possible fix<br>
> >> would be to do more bulk network operations when possible, so instead<br>
> >> of 100 separate calls to a root-wrapped tool call it once.<br>
> >><br>
> >>To manage notifications about this bug go to:<br>
> >><a href="https://bugs.launchpad.net/neutron/+bug/1257788/+subscriptions">https://bugs.launchpad.net/neutron/+bug/1257788/+subscriptions</a><br>
> ><br>
> ><br>
> ><br>
> >_______________________________________________<br>
> >OpenStack-dev mailing list<br>
> ><a href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a><br>
> ><a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
><br>
><br>
> _______________________________________________<br>
> OpenStack-dev mailing list<br>
> <a href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a><br>
> <a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
</p>