Ask Your Question
0

neutron agent-list shows ovs is not up on compute nodes

asked 2016-01-13 12:51:55 -0500

aksan gravatar image

image description

When i do a neutron agent-list on controller node, it shows ovs agent is dead on both compute nodes. But, when i create a vm ,it is getting a ip and i can access the console of that vm. When i do a nova list, i can see the vm on compute node 1. When i do a ovs-vsctl show on compute node 1, i can see the bridge configurations. I understand there is some reachability issues between controller and compute node. How do i solve this? I understand it is not creating any problem for me, but i want to solve this issue.

Thanks

edit retag flag offensive close merge delete

Comments

What says service neutron-openvswtch-agent status on compute nodes ?

dbaxps gravatar imagedbaxps ( 2016-01-13 13:53:43 -0500 )edit

service neutron-openvswitch-agent status: unrecongnised service

service neutron-plugin-openvswitch-agent status: start/running

service openvswitch-switch status : start/running

aksan gravatar imageaksan ( 2016-01-13 14:59:49 -0500 )edit

service neutron-plugin-openvswitch-agent restart and after that try neutron agent-list

Bipin gravatar imageBipin ( 2016-01-13 21:48:22 -0500 )edit

1 answer

Sort by ยป oldest newest most voted
0

answered 2016-01-14 03:25:27 -0500

dbaxps gravatar image

Run several times openstack-service restart on Compute Nodes to update database.

edit flag offensive delete link more

Comments

can you please elaborate on this term "update database"

Bipin gravatar imageBipin ( 2016-01-14 03:58:56 -0500 )edit

I mean mariadb database containing meta-data for Openstack deployment.

dbaxps gravatar imagedbaxps ( 2016-01-14 05:48:50 -0500 )edit

Hi,Thanks dbaxps and Bipin.

I did nothing except running "service neutron-plugin-openvswitch-agent restart" and "service openvswitch-switch restart" and now neutron agent-list shows that ovs is up for both compute nodes. Why did this happened? Some kind of database inconsistency? Please explain

aksan gravatar imageaksan ( 2016-01-14 07:08:22 -0500 )edit

Would you have some experience with 3 Node Controllers using MultiMaster MariaGalera db replication, you wouldn't be surprised at all. Pretty often nova service-list reports service down say on second controller, however systemctl | grep nova reports it up and running on same box.

dbaxps gravatar imagedbaxps ( 2016-01-14 08:41:46 -0500 )edit

Hi dbaxps, I am surprised in my situation, but i believe you have encountered these a lot. Thanks for the help. I will definitely try the 3 node controller with db replication to get familiar with these things.

aksan gravatar imageaksan ( 2016-01-14 09:14:23 -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: 2016-01-13 12:51:55 -0500

Seen: 2,687 times

Last updated: Jan 14 '16