Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.
Sign upNew imgadm has trouble doing initial upgrade if zones were created from a UUID that has no manifest #180
Comments
This comment has been minimized.
This comment has been minimized.
|
I'm not the only person hitting this: I'd also like to go back to using my machine, so a request for more information or a go-ahead to clean up my machine the inelegant way would be greatly appreciated. Thanks! |
This comment has been minimized.
This comment has been minimized.
|
@nshalman sorry for the delay. I'm looking at this now. |
This comment has been minimized.
This comment has been minimized.
|
@nshalman I think this commit should fix this. You should be able to try it out quickly per the "dev-install" notes here: https://github.com/joyent/smartos-live/blob/master/src/img/README.md#development I'd appreciate a note if that fixes it for you. Thanks, and sorry for the delay. |
This comment has been minimized.
This comment has been minimized.
|
I followed the instructions for doing a "dev-install" and things seem much better:
Other imgadm commands seem much happier now. It also looks like there is a persistent warning about a different image source that I created "properly" (via zfs send and receive of a zvol from another machine) but without metadata:
But that message doesn't break proper functionality and could be considered a feature (it's a "you created an image without metadata, fool" warning). Thanks for the fix!! |
How I encountered this:
My home system previously ran an older version of the platform.
I had created a blank KVM VM, and kickstarted fedora18 onto it.
I then took a snapshot of the disk zvol, cloned that snapshot to a new zones/$(uuid)
and used the resulting zvol UUID to create some additional VMs.
Upon rebooting the machine onto joyent_20130222T000747Z, imgadm is sad:
For reference, some zfs output showing relationships
Since these VMs aren't critical, I can probably do a workaround of destroying them so that imgadm will be happy and then recreate them from scratch, but I'll hold off for a while in case I can do some useful diagnostics or testing.