I am not understanding why there are secondary joins defined in the models. I suspect this might break other things, but maybe you can test that this at least makes the scheduling faster: http://paste.openstack.org/show/32534/ That seems to generate a much more acceptable query. - Chris On Feb 25, 2013, at 9:40 PM, Sam Morrison <sorrison at gmail.com> wrote: > > On 26/02/2013, at 4:31 PM, Chris Behrens <cbehrens at codestud.com> wrote: > >> After thinking more, it does seem like we're doing something wrong if the query itself is returning 300k rows. :) I can take a better look at it in front of the computer later if no one beats me to it. > > Yeah I think it's more than a missing index :-) > > The query does 2 INNER JOINS on aggregate_hosts then 2 INNER JOINS on aggregate_metadata then does a further 2 LEFT OUTER JOINS on aggregate_hosts. > Thanks for the help, > Sam -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130226/6e6e2b3d/attachment.html>