Deploy OpenStack via Juju - Fail on ceph-osd deploy

Hello. I have received a failure when going through the OpenStack deployment via Juju/Charms and am having a failure on 2 of my 4 nodes

(one sticks at needing to reboot, one gives this error:

Node installation failure - 'curtin' curtin command install

Is there a way to deploy onto those 2 failed nodes, or do I just need to burn the cluster and start over? I’m not sure about the node that doesn’t want to reboot yet, but the curtin command seems to be tied to a disk issue so I’ve deployed Ubuntu to that node and wiped that disk.

This is the original command:

juju deploy -n 4 --config ceph-osd.yaml --constraints tags=compute ceph-osd

I have tried doing this command but it fails because ceph-osd is already deployed:

juju deploy -n 4 --config ceph-osd.yaml --constraints tags=compute ceph-osd

ERROR cannot add application "ceph-osd": application already exists:
deploy application using an alias name, or use remove-application to remove the existing one and try again

It would seem that removing the application leaves the deployed nodes in place, perhaps I can do that and re-associate the other nodes (assuming they deploy correctly)? What I did previously was just delete all 4 nodes, destroy the controller, and start over from scratch.

Maybe you deploy openstack fellow OpenStack Charms Deployment Guide — charm-deployment-guide 0.0.1.dev416 documentation .

you can rebulit model with :

 juju destroy-model model-you-named --force --no-wait 
 juju add-model model-you-named

Then redeploy application.

After you deploy openstack manually, you can deploy openstack again with bundle Openstack Base #78.But may you need edit the bundle.yaml ,because actually used ethernet interface of metal machines may differ with config in


$ juju list-models
Controller: maas-controller

Model       Cloud/Region    Type  Status     Machines  Cores  Units  Access  Last connection
controller  mymaas/default  maas  available         1      3  -      admin   just now
default*    mymaas/default  maas  available         4     48  4      admin   just now

So I should be able to just destroy that default model and recreate it, rather than destroying the whole system. Thanks!

May you creat a new model instead of default model.

Hi Sean. How is your Charmed OpenStack installation coming along? The curtin problem you mentioned in your original post indicates an issue at the hardware and MAAS level. If that is still happening you should investigate without bringing Juju into the mix by manually deploying nodes from the MAAS web UI and checking its logs.

I was able to get past that, I ended up replacing one of the hardware nodes.

In general I have not had any luck getting a full install with Juju and Charms working. I can often get to the end, but end up experience other issues like non-functional networking or the console not working and such.

I think it’s just one of those things you have to keep at to get the hang of it. Even MicroStack doesn’t work out of the box, but if you stick with it you can find someone talking about resolving the networking issue and you do end up with a functional install.

1 Like