Ask Your Question
0

additional controller node(s)

asked 2020-01-22 18:22:40 -0500

happy75 gravatar image

Hi Openstack experts,

I have some questions with regards to adding additional controller nodes:

(1) when you install and configure an additional controller node (with a different ip address of course) in the openstack environment, does the additional controller node automatically get connected to the same openstack environment and be able to talk to the old controller nodes which already existed in the openstack environment?

(2) if you have 3 controller nodes in the same openstack environment, do these 3 controller nodes take on different roles or do they basically perform the same tasks? do the load automatically get distributed among the 3 controller nodes? If one of the controller nodes go down, does that mean the other 2 nodes will take on what that one was doing so no operation would be disrupted?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2020-01-22 19:02:37 -0500

does the additional controller node automatically get connected to the same openstack environment

No. You have to install it so that it gets connected to the environment. For example, you will have to create a database and message queue configuration to which both controllers are connected.

do these 3 controller nodes take on different roles or do they basically perform the same tasks?

You can configure them either way.

You will benefit from studying one of the commercial OpenStack architectures. While SUSE OpenStack cloud has been discontinued, the documentation is still available, and I happen to have some familiarity with it, so here is a picture of a three-controller setup: https://documentation.suse.com/soc/8/.... And here a description (sadly, no picture) of the so-called mid-range deployment, which features several controllers with different roles such as core, metering, database/messaging, and networking: https://documentation.suse.com/soc/8/....

do the load automatically get distributed among the 3 controller nodes? If one of the controller nodes go down, does that mean the other 2 nodes will take on what that one was doing so no operation would be disrupted?

OpenStack doesn't have high-availability built-in; you need to add something like HA-Proxy. Luckily, due to the stateless nature of most OpenStack services, this is not too hard. The picture above will help you.

edit flag offensive delete link more

Comments

thanks. So to clarify, for the configuration of the additional controller, do i just use the same set of instructions as the existing node (the ones provided on the openstack website) - and the only difference is that the additional node has a different ip address?

happy75 gravatar imagehappy75 ( 2020-01-22 21:35:02 -0500 )edit

and thanks for the links to those documents. I will take a look at them to get some ideas.

happy75 gravatar imagehappy75 ( 2020-01-22 21:35:33 -0500 )edit

You can't just create a second controller without adding a front-end. You will access your cloud using a single IP address - how is that possible with two or more controllers having several IP addresses? The typical solution for that is a virtual IP address and HA-Proxy.

Bernd Bausch gravatar imageBernd Bausch ( 2020-01-22 22:45:21 -0500 )edit

Thanks. i found something on the internet. it is like this? https://wiki.infn.it/progetti/cloud-a...

happy75 gravatar imagehappy75 ( 2020-01-22 23:49:29 -0500 )edit

will i be able to set up additional controller nodes WITHOUT using HA-proxy?

happy75 gravatar imagehappy75 ( 2020-01-23 00:09:21 -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: 2020-01-22 18:22:40 -0500

Seen: 57 times

Last updated: Jan 22