Ask Your Question

Automatic instance evacuation after node failure

asked 2014-09-17 09:22:39 -0500

caddo gravatar image

Hi all,

in the blueprint for Havana there were referrals for a possible implementation of something that would trigger Evacuate APIs in case of a node failure, sort of a vmware-ha like functionality.

Some people with legacy tenant applications are considering Openstack + vSphere just for this and sounds so lame to me...

Anyone with an update about this?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2014-10-04 22:21:29 -0500

jtaguinerd gravatar image


Just to share, I've sort of made a work-around for the automatic failover of the instance, using a third party monitoring tool like Nagios. In our case our instances are all boot from volume, so it is a bit easier. When Nagios detects that a compute node is down, it triggers a script to update the host field of the instance in mysql and then do a hard-reboot to recreate the xml file.

edit flag offensive delete link more


I like this idea. Are you able to share the specifics of your implementation?

f1gless gravatar imagef1gless ( 2014-10-06 02:12:36 -0500 )edit

Hi, 2 Controllers on active-passive setup , 8 Compute node (7 active-1passive), 3 ceph node. When you do nova service-list only 7 of the compute nodes are up, so that nova-scheduler will only use these nodes when launching the instance. The 8th node will only start when any of the compute node dies.

jtaguinerd gravatar imagejtaguinerd ( 2014-10-07 02:33:08 -0500 )edit

Get to know Ask OpenStack

Resources for moderators

Question Tools



Asked: 2014-09-17 09:21:54 -0500

Seen: 2,044 times

Last updated: Oct 04 '14