Difficulty after attempted template re-install
trueriver
https://www.qubes-os.org/doc/reinstall-template/
I tried first the automated method, and when that did not work then tried the manual method.
Question: Are these supposed to work with R 4.0 ? It says R 3.1+ and I am now wondering if this means ONLY R 3.n where n>1.
The result now is that I cannot even open the template domain to start adding new software. I am in dom0 and type
qvm-run -u root fedora-26-minimal xterm
(with or without a preceding 'sudo') and the result I get is a failure message saying
fedora-26-minimal: VM volume does not exist: /var/lib/qubes/VMtemplates/fedora-26-minimal
I know I am typing the command correctly because if I edit the command in Bash to take out the '-minimal' then xterm opens nicely in the fedora-26 domain
How can I sort this out?
Why has the action=reinstall not re-created everything the VM needs?
Any tips welcome...
awokd
> Hi, I have attempted to re-install the fedora-26-minimal template using
> the instructions here:
>
> https://www.qubes-os.org/doc/reinstall-template/
>
>
>
> I tried first the automated method, and when that did not work then tried
> the manual method.
>
> Question: Are these supposed to work with R 4.0 ? It says R 3.1+ and I am
> now wondering if this means ONLY R 3.n where n>1.
> How can I sort this out?
> Why has the action=reinstall not re-created everything the VM needs?
trueriver
>
> > How can I sort this out?
>
> Try the manual procedure again. Do you get any errors at any step?
No, none. Not till I try to run something.
It looks almost like it has re-installed it, it appears in the menu but with no apps, just the settings.
In settings the app pane is empty, and the refresh button greys out to ''Refresh in progress' and never returns after much much longer than it usually takes in a healthy domain.
>
> > Why has the action=reinstall not re-created everything the VM needs?
>
> Not sure, it should work!
OK, short of a complete re-install after backing up the domains that still work, what do I do now to repair this please?
Chris Laprise
functions was refactored late in the 4.0 pre-release cycle. Maybe this
should be opened as an issue.
Its conceivable that a bug left behind a similarly named meta-volume
that is now preventing a normal installation from completing. Comparing
the output of 'qvm-volume' with 'sudo lvs' may provide a clue if that's
the case.
--
Chris Laprise, tas...@posteo.net
https://github.com/tasket
https://twitter.com/ttaskett
PGP: BEE2 20C5 356E 764A 73EB 4AB3 1DC4 D106 F07F 1886
awokd
>
> No, none. Not till I try to run something.
>
>
> It looks almost like it has re-installed it, it appears in the menu but
> with no apps, just the settings.
>
> In settings the app pane is empty, and the refresh button greys out to
> ''Refresh in progress' and never returns after much much longer than it
> usually takes in a healthy domain.
confirm, you are:
sudo dnf remove qubes-template-fedora-26-minimal
sudo qubes-dom0-update qubes-template-fedora-26-minimal
Right? That's really all it should take. If it doesn't work, I can try it
too but it might be a day or two.
trueriver
> On Fri, April 20, 2018 8:42 pm, trueriver wrote:
>
> >
> > No, none. Not till I try to run something.
> >
> >
> > It looks almost like it has re-installed it, it appears in the menu but
> > with no apps, just the settings.
> >
> > In settings the app pane is empty, and the refresh button greys out to
> > ''Refresh in progress' and never returns after much much longer than it
> > usually takes in a healthy domain.
>
> Strange. Try to reboot maybe, sounds like something got confused.
Yes already tried that, several times over.
> Then, to
> confirm, you are:
>
> sudo dnf remove qubes-template-fedora-26-minimal
> sudo qubes-dom0-update qubes-template-fedora-26-minimal
I think so. Certainly what I intended to do.
trueriver
> Its conceivable that a bug left behind a similarly named meta-volume
> that is now preventing a normal installation from completing. Comparing
> the output of 'qvm-volume' with 'sudo lvs' may provide a clue if that's
> the case.
>
ran it again and it worked this time. Not sure why it took three re-installs but it is working now.
However, after the dnf remove, sudo lvs|grep minimal showed that vm-fedora-26-minimal-root-tmp was still there. Nothing shown at that point by qvm-volume|grep minimal.
After the qubes-dom0-update that volume is still there in sudo lvs, but not shown in qvm-volume.
Should I lvremove it, or do I need some Qubic magic command?
trueriver
> The code that supports template re-install (and other volume-related)
> functions was refactored late in the 4.0 pre-release cycle. Maybe this
> should be opened as an issue.
What do you think now?
Is that -root-tmp volume a sign of a bug, if so where?
I am not confident of reproducing the bug, if indeed it is one.
My gut feeling is that it may not enough to make a useful bugrep, but will do so if you or awokd think I should.
One thought I had is how do I know f I run out of pool space - might that have triggered something like this or should I get an elegant warning? Certainly my disk space is overcommitted, with the magic of sparse files.
awokd
> Is that -root-tmp volume a sign of a bug, if so where?
>
>
> I am not confident of reproducing the bug, if indeed it is one.
>
>
> My gut feeling is that it may not enough to make a useful bugrep, but
> will do so if you or awokd think I should.
>
> One thought I had is how do I know f I run out of pool space - might that
> have triggered something like this or should I get an elegant warning?
> Certainly my disk space is overcommitted, with the magic of sparse files.
Tried to reproduce this. First installed the minimal template, tested,
then did an --action=reinstall. The menu shortcut for xterm stopped
working. Did a refresh applications in Qube Settings but that also failed
to start the template qube. qvm-run gave me "VM directory does not exist:
/var/lib/qubes/vm-templates/fedora-26-minimal" and ls confirms it does
not. This is probably a bug, possibly
https://github.com/QubesOS/qubes-issues/issues/3169. I'll make a note in
there.
At that point, "sudo dnf remove qubes-template-fedora-26-minimal" followed
by "sudo qubes-dom0-update qubes-template-fedora-26-minimal" worked with
no errors and restored proper function.
Next, I tested with the template running. --action=reinstall shutdown the
template before doing its work, but resulted in the same bug as before.
Doing a dnf remove with the template running failed with an error message
about same. I didn't see -root-tmp at any time; not sure what might have
created it.
Chris Laprise
> On Fri, April 20, 2018 11:38 pm, trueriver wrote:
>
>> Is that -root-tmp volume a sign of a bug, if so where?
>>
>>
>> I am not confident of reproducing the bug, if indeed it is one.
>>
>>
>> My gut feeling is that it may not enough to make a useful bugrep, but
>> will do so if you or awokd think I should.
>>
>> One thought I had is how do I know f I run out of pool space - might that
>> have triggered something like this or should I get an elegant warning?
>> Certainly my disk space is overcommitted, with the magic of sparse files.
>
> Don't think it's pool space related.
organization issue which causes reinstall to abort part way through.
>
> Tried to reproduce this. First installed the minimal template, tested,
> then did an --action=reinstall. The menu shortcut for xterm stopped
> working. Did a refresh applications in Qube Settings but that also failed
> to start the template qube. qvm-run gave me "VM directory does not exist:
> /var/lib/qubes/vm-templates/fedora-26-minimal" and ls confirms it does
> not. This is probably a bug, possibly
> https://github.com/QubesOS/qubes-issues/issues/3169. I'll make a note in
> there.
there. As I mentioned, the storage layer was updated and its partly to
implement 3169.
>
> At that point, "sudo dnf remove qubes-template-fedora-26-minimal" followed
> by "sudo qubes-dom0-update qubes-template-fedora-26-minimal" worked with
> no errors and restored proper function.
>
> Next, I tested with the template running. --action=reinstall shutdown the
> template before doing its work, but resulted in the same bug as before.
>
> Doing a dnf remove with the template running failed with an error message
> about same. I didn't see -root-tmp at any time; not sure what might have
> created it.
Chris Laprise
stable yet. I think this should be tried after updating from qubes*testing.
awokd
>
> OK, problem may be that 'qubes-core-admin/pull/203' hasn't reached
> stable yet. I think this should be tried after updating from
> qubes*testing.
testing
(https://github.com/marmarek/qubes-core-admin-client/commit/c75c0176dc23dbb50dc1420c2bfe181844e4ae47).
I updated to testing and the bug is gone so --action=reinstall is working
correctly now. Thanks!
And a big thanks to whomever coded that new disk usage widget that came
with the testing updates, I think that's going to save a lot of people a
lot of grief.
trueriver
> On Sat, April 21, 2018 1:02 pm, Chris Laprise wrote:
> >
> > OK, problem may be that 'qubes-core-admin/pull/203' hasn't reached
> > stable yet. I think this should be tried after updating from
> > qubes*testing.
>
> Looks like Marek just pushed a fix for core-admin-client v4.0.17 to
> testing
MY thanks to Marek for that, and to everyone who responded in this thread.