Ask Your Question
0

HAProxy, Keepalived and neutron.

asked 2016-03-18 03:43:56 -0500

Lucas LAMBOLEY gravatar image

updated 2016-03-18 03:51:14 -0500

Hi,

I'm in trouble. I want to do HA on my Openstack service, so i decide to do HAproxy and Keepalived. Keepalived provide HA for VIP using by HAProxy, And HAproxy do load-balancing using the VIP and the port of the service. But ... If by example, i have a network node. So i double that. If network node one go down, so HAproxy redirect to other API network node, but the network created in the network node one, i can't got him on the seconde network node, because there are one the one ... It's the same for glance image, cinder volume etc. What i don't understand ? Help me please, i can't work when i don't understand, and i don't see any response at my question ... Sorry for my english. Best Regards, L.L

EDIT1: Need i configure the same database for all node ?

edit retag flag offensive close merge delete

Comments

"EDIT1: Need i configure the same database for all node ?"
Which nodes you mean Controllers or Network ?
Might be better to stop developing bike and create 3 Controller/Network Nodes as was advised in Howto suggested to you last time.

dbaxps gravatar imagedbaxps ( 2016-03-18 04:08:50 -0500 )edit

By the way if you take a look on how Triple0 works . It is pushing to overcloud 3 HA Controller/Network Cluster using pacemaker. Same approach see how yaml templates look been used for overcloud deployment.

dbaxps gravatar imagedbaxps ( 2016-03-18 04:14:48 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
1

answered 2016-03-18 04:05:07 -0500

dbaxps gravatar image

If you will remind suggested to you :-
https://github.com/beekhof/osp-ha-dep...
External network provider is invoked only on Node1. However fail-over plays nicely with Neutron. Why that ?

1. MariadbGalera MultiMaster Synchronous Replica is running on all 3 Controller/Network Nodes
2.`neutron-db-manage --config-file /etc/neutron/neutron.conf --config-file /etc/neutron/plugin.ini upgrade head`
has been already run, before external network provider was invoked.
Hence, changes been done on Node 1 will be populated across Controller's cluster.
edit flag offensive delete link more

Comments

Sorry, It's works now. I was concentred on openstack service and i have completely forgotten the non-api service, like Galera cluster for maria, etc. I'm going to use link to the core non-openstack service Thx

Lucas LAMBOLEY gravatar imageLucas LAMBOLEY ( 2016-03-18 08:04:21 -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: 2016-03-18 03:43:56 -0500

Seen: 483 times

Last updated: Mar 18 '16