does openstack has any project with support fault tolerance of intance with zero downtime

asked 2020-06-11 06:02:39 -0500

Maaz gravatar image

We are running instance on SSD host, which uses localssd as cinder backend, but the problem is whenever the Raid Controller fails ( Disks are in Raid 5)we lost the instance data till the time we fix the Hardware problem. And we need to provision the replacement VMs and setup restore all the data from backup.

I was thinking if we have something similar to FT in vmware which can be implemented in existing setup to have a duplicate running copy of instance (Block data as well as RAM content ) on other SSD host and at the time of failure it can automatically marks secondary copy of instance as active one without any downtime.

Or I am expecting too much from opensource project . :P

edit retag flag offensive close merge delete

Comments

There have been discussions around this subject, see https://specs.openstack.org/openstack.... There is a project named Masakari: https://docs.openstack.org/masakari and https://wiki.openstack.org/wiki/Masakari, sponored by NTT I think.

Bernd Bausch gravatar imageBernd Bausch ( 2020-06-12 20:42:35 -0500 )edit

Tripleo seems to offer some kind of instance HA: https://docs.openstack.org/project-de....

Being open-source is not an issue. However, just like for closed-source products, there must be an incentive for creating a solution.

Bernd Bausch gravatar imageBernd Bausch ( 2020-06-12 20:45:13 -0500 )edit

Thank you Bernd, much appreciated. Doesn't seem to fit our requirement, but got to learn about new stuff.

Maaz gravatar imageMaaz ( 2020-06-15 01:03:28 -0500 )edit