Bug in guix installer (seems to be unable to create swap) - reproducable

  • Open
  • quality assurance status badge
Details
4 participants
  • Andrew "Truck" Holland
  • Ludovic Courtès
  • Mathieu Othacehe
  • Andrew "Truck" Holland
Owner
unassigned
Submitted by
Andrew "Truck" Holland
Severity
normal
A
A
Andrew "Truck" Holland wrote on 31 May 2021 15:44
(address . bug-guix@gnu.org)
20210531134423.GS16306@docka.amigadns.com
I have attempted to install 1.3.0 from the iso, but the "guided"
install encounters an error after partitioning; which if I read
correctly, indicates swap cannot be created.

This happens when I use _any_ partitioning scheme, manual or
automatic.

The command 'mkswap -f /dev/sda3' executes correctly from the shell
session on tty3 from the installer.

Sadly, this is also fatal and the installation cannot continue, so
I'm a bit surprized no one has reported this - as I can recreate
this situation reliably every time I attempt it. (Maybe I just am
blind and cannot find the bug report at the guix bug tracker, if
so, my apologies for the duplicate.)

Error file attached.
---
In ice-9/ports.scm:
478:4 19 (with-error-to-port _ _)
In guix/build/utils.scm:
654:6 18 (invoke _ . _)
In ice-9/boot-9.scm:
1669:16 17 (raise-exception _ #:continuable? _)
1667:16 16 (raise-exception _ #:continuable? _)
1667:16 15 (raise-exception _ #:continuable? _)
1667:16 14 (raise-exception _ #:continuable? _)
1667:16 13 (raise-exception _ #:continuable? _)
1667:16 12 (raise-exception _ #:continuable? _)
1667:16 11 (raise-exception _ #:continuable? _)
1667:16 10 (raise-exception _ #:continuable? _)
1667:16 9 (raise-exception _ #:continuable? _)
1667:16 8 (raise-exception _ #:continuable? _)
1667:16 7 (raise-exception _ #:continuable? _)
1764:13 6 (_ #<&invoke-error program: "mkswap" arguments: ("-f" "/dev/sda3") exit-status: 1 term-signal: #f stop-signal: #f>)
In ice-9/eval.scm:
619:8 5 (_ #(#(#<directory (guile-user) 7f2d93296c80> #<<installer> name: newt init: #<procedure init ()> exit: #<procedure exit ()> exit-error: #<procedure exit-error (file key args)> f…>) …))
619:8 4 (_ #(#(#(#<directory (guile-user) 7f2d93296c80> #<<installer> name: newt init: #<procedure init ()> exit: #<procedure exit ()> exit-error: #<procedure exit-error (file key arg…>) …) #))
In ice-9/ports.scm:
463:17 3 (call-with-output-file _ _ #:binary _ #:encoding _)
In ice-9/eval.scm:
619:8 2 (_ #(#(#<directory (guile-user) 7f2d93296c80> %exception (#<&invoke-error program: "mkswap" arguments: ("-f" "/dev/sda3") exit-status: 1 term-signal: #f stop-signal: #f>)) #<output:…>))
159:9 1 (_ #(#(#<directory (guile-user) 7f2d93296c80> %exception (#<&invoke-error program: "mkswap" arguments: ("-f" "/dev/sda3") exit-status: 1 term-signal: #f stop-signal: #f>)) #<output:…>))
In unknown file:
0 (make-stack #t)
ice-9/eval.scm:159:9: ERROR:
1. &invoke-error:
program: "mkswap"
arguments: ("-f" "/dev/sda3")
exit-status: 1
term-signal: #f
stop-signal: #f
M
M
Mathieu Othacehe wrote on 1 Jun 2021 14:46
(name . Andrew "Truck" Holland)(address . gutruck9@trebel.org)(address . 48762@debbugs.gnu.org)
87mts969d6.fsf@gnu.org
Hello Andrew,

Toggle quote (3 lines)
> This happens when I use _any_ partitioning scheme, manual or
> automatic.

Thanks for the complete bug report. I think that this bug falls into the
https://issues.guix.gnu.org/47567,https://issues.guix.gnu.org/44872 and
probably some others.

The "free-parted" procedure that makes sure that all the partitions are
created and not in use probably fails in some configurations. Otherwise,
I cannot understand why the "mkswap" binary would fail.

We may need to come up with a new procedure that provides stronger
guarantees.

Thanks,

Mathieu
L
L
Ludovic Courtès wrote on 13 Jun 2021 15:12
(name . Mathieu Othacehe)(address . othacehe@gnu.org)
875yyh29j3.fsf@gnu.org
Hi,

Mathieu Othacehe <othacehe@gnu.org> skribis:

Toggle quote (4 lines)
> The "free-parted" procedure that makes sure that all the partitions are
> created and not in use probably fails in some configurations. Otherwise,
> I cannot understand why the "mkswap" binary would fail.

Do we know why ‘mkswap’ exited with code 1, though? EBUSY or could it
be something else?

Thanks,
Ludo’.
A
A
Andrew "Truck" Holland wrote on 13 Jun 2021 15:01
(name . Ludovic Courtès)(address . ludo@gnu.org)
20210613130152.GO16306@docka.amigadns.com
I would _asume_ it could not be "EBUSY" as I see nothing in the script that
would mount the drive... Again: this is the script, not custom. This is
100% reproducable in all my setups: just try using the defaults from the
installer, and you should hit this.

This occurs immediately after partitioning, which _does_ succeed.

- you say "some configurations" - try the DEFAULT. Let it wipe the HD,
partition, and: fail.

A guess, but ... one that makes no sense:

Surely it isn't one of those "old kernel" type problems where the new
partition scheme isn't known, and therefore, kernel can't find the partition
that the script expects? I haven't seen one of those issues for... over
15 years (:

On Sun, Jun 13, 2021 at 03:12:32PM +0200, Ludovic Courtès wrote:
Toggle quote (14 lines)
> Hi,
>
> Mathieu Othacehe <othacehe@gnu.org> skribis:
>
> > The "free-parted" procedure that makes sure that all the partitions are
> > created and not in use probably fails in some configurations. Otherwise,
> > I cannot understand why the "mkswap" binary would fail.
>
> Do we know why ‘mkswap’ exited with code 1, though? EBUSY or could it
> be something else?
>
> Thanks,
> Ludo’.
>
?