Ok, now I'm afraid to turn off my computer or even stop any Debian template based VM's...
Here is what happened.
I was going to do a general update on Dom0 and my Debian-10 and Fedora-32 templates.
As is my habit, I deleted the older clones of those template VM's and was creating new clones with qvm-clone from a Dom0 command window.
While attempting to create a new clone of the Debian-10 template, it halted with an error:
file pool cannot export dirty volumes.
Searching for that issue suggested I start the template VM and exit it cleanly...although I don't have a recollection of a "dirty" exit (crash, kill etc...).
I went to start the template with qvm-start and it won't, giving the error:
file pool cannot start a VM with an exported volume.
How in the world do I recover from this?
Stuart
'Stuart Perkins' via qubes-users:
Ok, now I'm afraid to turn off my computer or even stop any Debian template based VM's...
Don't panic, it's just a bug* in qubes-core-dom0-4.0.56. Your VM data
is still okay.
Here is what happened.
I was going to do a general update on Dom0 and my Debian-10 and
Fedora-32 templates.
As is my habit, I deleted the older clones of those template VM's
and was creating new clones with qvm-clone from a Dom0 command
window.
While attempting to create a new clone of the Debian-10 template, it
halted with an error:
file pool cannot export dirty volumes.
Searching for that issue suggested I start the template VM and exit
it cleanly...although I don't have a recollection of a "dirty" exit
(crash, kill etc...).
I went to start the template with qvm-start and it won't, giving the
error:
file pool cannot start a VM with an exported volume.
How in the world do I recover from this?
If you restart your computer (or only qubesd), it will drop the
lingering export lock and you'll be able to start the original
template again, etc.
Rusty
* Refuse to export a file volume of a running VM · QubesOS/qubes-core-admin@0eb9504 · GitHub
I figured that out, eventually...after some sweat. I rely on this machine way too much. All critical data was backed up, but I was not looking forward to a reinstall and the time it would take out of my "way-too-busy" week.
Aside: I found out the command line "qvm-clone" will NOT work. That is what left it in this state. Oddly enough, using the Qubes Manager gui it succeeds in the clone...but leaves it where you can't start the template to do the update until you reboot again.
Just FYI. In my install I chose NOT to do the LVM as it adds an unnecessary layer of complexity to a laptop, which is already an encrypted installation. I want to add additional drive space ONCE, and do that with just formatting the second drive and symbolic links from the appvm directory to the "too large for my ssd" drives. SSD is 260G. Hard drive is 2TB. Like most Qubes users, I work the crap out of my machine...a coreboot Lenovo T420i.
Now that I'm up to 5.4.88, is this fixed?
Stuart