[one-users] VMs IP Clash after failure

[Duverne, Cyrille]

Dear mailing list,

Following my last VM failure, I removed the VM directly from the
Sunstone DB, not a so good idea :)

To sum up the context :
- I created a batch of VMs
- Only 1 failed (FAILED state)
- I tried to redeploy it several times, and get an error message
saying that "deployment.1.transfer" was not accessible or something
- I then manually removed the working files of the $VMID directory on
my mutualized storage
- I removed the $VMID record from the DB (since the delete wasn't
working properly)
- And then the situation was back to normal, at least in the display.
- Afterward, I tried to create a new VM, which got the old IP of the
failed VM
- I faced the issue that the 1st VM has been correctly deployed but
appeared as failed in Sunstone, causing an IP clash with the new one.

So, why a VM could be deployed and appears as FAILED in OpenNebula ?
And how ONE could be sure that an IP adress isn't already used when
creating a new VM ?

Thanks in advance for your feedback.

Kind regards
Cyrille

"Imagination is more important than Knowledge"
Albert Einstein


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20121113/0d2186cd/attachment.htm>