Resizing Instances on NFS Shared Storage

Question

3

Resizing Instances on NFS Shared Storage

asked 2013-12-20 10:16:34 -0600

abdi ibrahim
41 ●1 ●2 ●6

updated 2013-12-20 16:25:16 -0600

smaffulli
6058 ●28 ●50 ●87 http://maffulli.net/

I'm running Havana multi node deployment: 1 Controller and 3 Compute Nodes. NFS is the back-end shared storage where /var/lib/nova/instances are exported to the compute nodes. Spawning instances and live migrating (using CLI) works great. However; when I do a resize of an instance either from the Dashboard or from CLI it returns an error with the following exit code:

 stdr: error while executing command ssh <*ip address> to <remote server>.... Invalid Host Key

Not sure why the resize command is trying to create another folder to another compute node where the storage is already shared. The source and destination directory are hosted on the same storage. I've configured the following:

allow_resize_to_same_host=true
allow_migrate_to_same_host=true

restarted the compute services on the compute nodes and the same error occurs.

edit retag flag offensive close merge delete

Comments

You can solve the host key issue by creating a key for the nova user and propagating it to all your compute nodes. Basically passwordless ssh needs to work, however, that won't solve the issue of why it's trying to copy at all when it's shared storage. That's where I'm at, did you ever solve it?

mfischer ( 2014-05-07 17:18:01 -0600 )edit

You can solve the host key issue by creating a key for the nova user and propagating it to all your compute nodes. Basically passwordless ssh needs to work, however, that won't solve the issue of why it's trying to copy at all when it's shared storage. That's where I'm at, did you ever solve it?
mfischer ( 2014-05-07 17:18:01 -0600 )edit

Answer 1

1

answered 2013-12-31 14:14:39 -0600

abdi ibrahim
41 ●1 ●2 ●6

So here is a quick update. After making the changes in above I restarted all the nova compute services and I am now able to run the resize command either in the CLI and in the dashboard. As expected the resize does occur on the same host rather than the previously attempting to move to another host where it error out.

However the migrate option now does the same thing (migrating on the same host). I'm assuming that I need to set the following option in the nova.conf: allow_migrate_to_same_host=false

Does this means that migration would produce the similar error? meaning SSH key issues? the reality is that the instances disks and config files are on the shared storage and should not be a problem. I'll test it out and post my results.

Thanks, Abdi

edit flag offensive delete link

Comments

Here is my update after changing the config flag allow_migrate_to_same_host to false the same error does occur which makes me think that there is a bug in the code that does not allow the command to understand that the instances are hosted in a shared storage (NFS in my case). Here is the nova-compute.log entries at pastbin: "http://pastebin.com/FbLW5jxM"

abdi ibrahim ( 2014-01-02 11:26:55 -0600 )edit

Answer 2

0

answered 2013-12-21 23:33:29 -0600

DennyZhang
76 ●5 ●9 ●12

updated 2013-12-21 23:34:33 -0600

What's the console output, if

 - ssh from  "<*ip address>" to <remote server>
 - qemu-img resize /var/lib/nova/instances/XXX/disk

edit flag offensive delete link

Resizing Instances on NFS Shared Storage

Comments

2 answers

Comments

Your Answer

Get to know Ask OpenStack

Question Tools

Stats

Related questions

Feedback About This Page

OpenStack

Community

Documentation

Branding & Legal