Ask Your Question
0

high availability for controller and compute in openstack

asked 2019-03-13 12:34:17 -0500

sanjay khuntia gravatar image

Hello All, Is there any latest tools/service available in openstack releases which can provide high availability with respect to control and compute services?

Please let me know how this can be achieved ?

thanks in advance for your replies.

regards,

edit retag flag offensive close merge delete

2 answers

Sort by ยป oldest newest most voted
0

answered 2019-03-13 16:43:17 -0500

OpenStack has no built-in high-availability. Deployment tools add high-availability using solutions like Galera or Percona for the database, and Pacemaker, HA-proxy, Keepalived etc for the OpenStack services. The stateless nature of OpenStack services simplifies this task. RabbitMQ has HA built-in.

Examples for HA implementations: SUSE OpenStack Cloud uses a combination of HAproxy and Keepalived, so does Kolla, whereas Red Hat OpenStack cloud prefers Pacemaker with Corosync.

I believe that compute nodes are typically not configured for HA, since the expectation is that instances can be recreated elsewhere in case a compute node goes down (cattle vs pets).

edit flag offensive delete link more

Comments

to add, there's a Masakari Service for Instance HA, I haven't tested it yet though https://www.openstack.org/software/re...

vblando gravatar imagevblando ( 2019-03-20 01:54:33 -0500 )edit
0

answered 2019-03-14 17:34:24 -0500

Here's how to deploy high availability with OpenStack-Ansible: https://docs.openstack.org/project-deploy-guide/openstack-ansible/draft/app-config-prod.html (https://docs.openstack.org/project-de...)

For compute nodes high availability you would need:

  • instances to use volume storage (Cinder), instead of local storage. And high availability in your storage solution
  • high availability on network/connection between compute nodes and storage

In an instance that a compute nodes dies, instance is booted on another compute node using the same disk from Cinder.

edit flag offensive delete link more

Comments

Thanks Adrian for your reply. Whether instance level HA is handled automatically by the openstack environment or do I have to do something specifically inside compute? Is there openstack service/tools available to achieve instance high availability inside a compute?

rgds,

sanjay khuntia gravatar imagesanjay khuntia ( 2019-03-15 04:27:56 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2019-03-13 12:34:17 -0500

Seen: 82 times

Last updated: Mar 14