Work is underway in cinder to allow active/active HA (or active/active/active/... for the truly paranoid) volume managers for a single backend, for remote backends that can support it. This is likely to take most of the Kilo cycle to finish.
You can use pacemaker or other H/A technology (I suspect some sort of monitored VM is the VMWare equivalent) to improve the availability of cinder-volume in the current release. On 8 October 2014 13:58, LiuJeremy <jeremy-...@outlook.com> wrote: > > Hi, > > I have an issue about single point of failure in OpenStack+VMware > scenario. > If one nova-compute or cinder-volume crush. The VCenter which connect to > the nova-compute or cinder-volume will unavailable. > > I think we can deploy the nova-compute/cinder-volume to a VM which have the HA > (High Available) capacity. > > What do you think about the above solution? And could you provide another > solution, > especially in physical deployment scenario (deploy the > nova-compute/cinder-volume > in a physical host)? > > Thanks, > > Jeremy Liu > > _______________________________________________ > OpenStack-dev mailing list > OpenStack-dev@lists.openstack.org > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- Duncan Thomas _______________________________________________ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
