Upload Cirros "TestVM" image failed in multi-node with HA

[Charles Hsu]

Fuel version: 6.0 (MOS 6.0)
Deployment mode: multi-node with HA

I might ran into a bug when I deployment a OpenStack environment on VirtualBox. I used the Mirantis quickstart scripts (https://software.mirantis.com/quick-start/) to create the environment.

Upload TestVM failure
http://i.imgur.com/MvnTEg5.png

Astute_log
http://i.imgur.com/XYsBEL8.png

I tried to upload a image from primary controller with glance command and it's work.
http://i.imgur.com/Y3xurfb.png

I can list the image.
http://i.imgur.com/VlKC3t7.png

The attachment is my deployment info.
I can deploy a OpenStack environment with multi-node without any problem.

[Miroslav Anashkin (manashkin)]

Charles,

What was the backend for Cinder?

If I understand correct - there is no compute nodes in your environment and only single Cinder role, combined with one of controllers.
And this controller was not the first one deployed.

Could you please attach (better id compressed) the
/var/log/docker-logs/remote/nailgun/app.log from master node to this bug?

Or generate a diagnostic snapshot and attach it instead of the single log.

And last - 1.5 GB of RAM per controller in HA mode is not enough.
It may not work, but with small probability.
Please check the minimum requirements in the docs here:
http://docs.mirantis.com/openstack/fuel/fuel-6.0/planning-guide.html#controller-nodes-3

[Charles Hsu (charles0126)]

Miroslav,

Yes, you're correct. That's no compute node, only three controllers with single Cinder role.
I have limited resource on my machine, so I didn't keep the logs on the VMs, but I tried that many times and always fails.
It's easy to reproduce in my machine.

So I'm going to increase amount of memory to 2G for each fuel slaves. And try to deploy a new multi-node with HA environment.

I'll update more detail here.

[Charles Hsu (charles0126)]

Miroslav,

The backend of Cinder is LVM, I used default setting.

[Charles Hsu (charles0126)]

Miroslav,

I recreated a new environment and deployed it. I got same error.

Error
http://i.imgur.com/xpH7O08.png

Astute Log
http://i.imgur.com/T5TIbT1.png

Network Config
http://i.imgur.com/qOr2miO.png

?field.comment=Miroslav,

I recreated a new environment and deployed it. I got same error.

Error
http://i.imgur.com/xpH7O08.png

Astute Log
http://i.imgur.com/T5TIbT1.png

Network Config
http://i.imgur.com/qOr2miO.png

[Charles Hsu (charles0126)]

node-6: primary controller + cinder
node-7: controller + cinder
node-8: controller + cinder

[Vitaly Sedelnik (vsedelnik)]

Won't Fix for 6.0-updates as there is no real customer impact