Yes, that's true, and that's what we're doing really. Our order of upgrade is q-srv -> q-ovs-agent -> q-dhcp and rest. If that causes issues, please submit bug for it. We should address that in our playbooks.
On 8 August 2016 at 17:02, Ihar Hrachyshka <ihrac...@redhat.com> wrote: > (Adding back ops@) > > Michał Jastrzębski <inc...@gmail.com> wrote: > >> Yes, but it's a bit different. Difference being if you're doing it >> right, you can keep nova running during upgrade. Can't say the same >> for neutron. > > > There is ongoing work to make it possible. The way it looks now, it will > take Ocata the least. Till that happy time, you should indeed all > controllers before upgrading database. > > That said, you can still run your agents while you upgrade your controller > nodes. Once you are done with controllers, you can upgrade one remaining > node running agents at a time. > > Ihar _______________________________________________ OpenStack-operators mailing list OpenStack-operators@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators