Guix system becomes unresponsive after backtrace

  • Done
  • quality assurance status badge
Details
4 participants
  • Gábor Boskovits
  • Danny Milosavljevic
  • Giovanni Biscuolo
  • Ludovic Courtès
Owner
unassigned
Submitted by
Gábor Boskovits
Severity
normal
G
G
Gábor Boskovits wrote on 6 Nov 2019 15:15
(address . bug-guix@gnu.org)
CAE4v=pib7J-rrcR9JDXN8iLdy2bMtUK+b7kPVMH4No35doRcGw@mail.gmail.com
Hello,

I did not know how to get the info better, so here is a screenshot
about the situation.

This happened when I was trying a pre-inst-env guix build from a
core-updates checkout.
Previously python3 failed to build, and I was trying to build it again.

Best regards,
g_bor
--
OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21
Attachment: error_screenshot.png
D
D
Danny Milosavljevic wrote on 6 Nov 2019 16:52
(name . Gábor Boskovits)(address . boskovits@gmail.com)(address . 38088@debbugs.gnu.org)
20191106165235.614a4d8a@scratchpost.org
Hi Gábor,

On Wed, 6 Nov 2019 15:15:41 +0100
Gábor Boskovits <boskovits@gmail.com> wrote:

Toggle quote (4 lines)
> This happened when I was trying a pre-inst-env guix build from a
> core-updates checkout.
> Previously python3 failed to build, and I was trying to build it again.

Hmm, sounds like disk corruption. If there's a sudden read-only appearing then
it's often because the kernel found a file system error and doesn't want to make
the situation worse. It then remounts the affected file-system read-only.
According to the top of your screenshot, even /tmp is read-only. I think we
don't use a tmpfs, so that's the root file system.

Could you check dmesg for signs?
-----BEGIN PGP SIGNATURE-----

iQEzBAEBCAAdFiEEds7GsXJ0tGXALbPZ5xo1VCwwuqUFAl3C7EMACgkQ5xo1VCww
uqXHfwf/VmXKpfn4vPU6+d/KftehDAfZrjLCp9h2zo8hZYe67D/UAYM71jhKRI0t
SliQLYR5kIF93DnNjVrRcVDG8FTMVQ9QZIjwxhiUKxqBlK7bKFo2eV7pqG76lRQl
ZWP+VdD2gYAxSB0uT4DyiwV+G3w8Hfv8SnEd+LQNj4HLyrMCVb1mkzPt9DfcmbtH
iU6esIX32vWGfyDYmAZdHKKZ2Lr1T4eXU24Ty3RA/TR5QCdRmykQEHizFuq/eITp
l4VJxcb4CUs0ijtsWacY6IrfWz2WUeoGrriMin3HEMRgDzddH56Vd1+cY3rwlKCe
ZoUzh5t5lgr/wvv16pRQJM40csm+qA==
=yAZl
-----END PGP SIGNATURE-----


L
L
Ludovic Courtès wrote on 6 Nov 2019 18:42
(name . Gábor Boskovits)(address . boskovits@gmail.com)(address . 38088@debbugs.gnu.org)
87y2wsvr3m.fsf@gnu.org
Hi Gábor,

Gábor Boskovits <boskovits@gmail.com> skribis:

Toggle quote (7 lines)
> I did not know how to get the info better, so here is a screenshot
> about the situation.
>
> This happened when I was trying a pre-inst-env guix build from a
> core-updates checkout.
> Previously python3 failed to build, and I was trying to build it again.

From the screenshot, it seems that your root file system (or at least
/tmp and /gnu/store) became read-only, which in turn caused various
things to fail, including guix-daemon (hence the “broken pipe” when
‘guix build’ was talking to it, I suppose.)

Could you check what happened on your machine? Do /var/log/messages and
/var/log/guix-daemon.log contain any hints?

Thanks,
Ludo’.
G
G
Gábor Boskovits wrote on 6 Nov 2019 23:40
(name . Ludovic Courtès)(address . ludo@gnu.org)(address . 38088@debbugs.gnu.org)
CAE4v=pi6MGLf4GK3vgprt=10xhdRSVUBY_kgojgQNy+kiPkxgg@mail.gmail.com
Hello,

Ludovic Courtès <ludo@gnu.org> ezt írta (id?pont: 2019. nov. 6., Sze, 18:42):
Toggle quote (17 lines)
>
> Hi Gábor,
>
> Gábor Boskovits <boskovits@gmail.com> skribis:
>
> > I did not know how to get the info better, so here is a screenshot
> > about the situation.
> >
> > This happened when I was trying a pre-inst-env guix build from a
> > core-updates checkout.
> > Previously python3 failed to build, and I was trying to build it again.
>
> From the screenshot, it seems that your root file system (or at least
> /tmp and /gnu/store) became read-only, which in turn caused various
> things to fail, including guix-daemon (hence the “broken pipe” when
> ‘guix build’ was talking to it, I suppose.)

Yes, it also became corrupted. fsck on boot fixed it.
Since then it works happily again...

Toggle quote (4 lines)
>
> Could you check what happened on your machine? Do /var/log/messages and
> /var/log/guix-daemon.log contain any hints?

No idea actually, I don't see anything really suspicious, but it might
well be that it just refused
to write to the logs also...

Toggle quote (5 lines)
>
> Thanks,
> Ludo’.


Regards,
g_bor
--
OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21
G
G
Gábor Boskovits wrote on 6 Nov 2019 23:42
(name . Danny Milosavljevic)(address . dannym@scratchpost.org)(address . 38088@debbugs.gnu.org)
CAE4v=pggGehgnh4o701DVJRic3hqqn7Oza7cKW7zdujWE+aWHw@mail.gmail.com
Danny Milosavljevic <dannym@scratchpost.org> ezt írta (id?pont: 2019.
nov. 6., Sze, 16:52):
Toggle quote (16 lines)
>
> Hi Gábor,
>
> On Wed, 6 Nov 2019 15:15:41 +0100
> Gábor Boskovits <boskovits@gmail.com> wrote:
>
> > This happened when I was trying a pre-inst-env guix build from a
> > core-updates checkout.
> > Previously python3 failed to build, and I was trying to build it again.
>
> Hmm, sounds like disk corruption. If there's a sudden read-only appearing then
> it's often because the kernel found a file system error and doesn't want to make
> the situation worse. It then remounts the affected file-system read-only.
> According to the top of your screenshot, even /tmp is read-only. I think we
> don't use a tmpfs, so that's the root file system.

I also believe it was the root filesystem.

Toggle quote (3 lines)
>
> Could you check dmesg for signs?

I believe the dmesg info was lost on force-restart. Logs contain nothing....

Regards,
g_bor
--
OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21
G
G
Giovanni Biscuolo wrote on 7 Nov 2019 08:35
(name . Gábor Boskovits)(address . boskovits@gmail.com)(address . 38088@debbugs.gnu.org)
87eeykqgti.fsf@roquette.mug.biscuolo.net
Hello Gabor,

Gábor Boskovits <boskovits@gmail.com> writes:

[...]

Toggle quote (8 lines)
>> From the screenshot, it seems that your root file system (or at least
>> /tmp and /gnu/store) became read-only, which in turn caused various
>> things to fail, including guix-daemon (hence the “broken pipe” when
>> ‘guix build’ was talking to it, I suppose.)
>
> Yes, it also became corrupted. fsck on boot fixed it.
> Since then it works happily again...

if the problem depended on filesystem corruption and all is fine now for
you, could you also close this bug please?

[...]

Thanks! Gio'

--
Giovanni Biscuolo

Xelera IT Infrastructures
G
G
Gábor Boskovits wrote on 7 Nov 2019 08:40
(name . Giovanni Biscuolo)(address . g@xelera.eu)(address . 38088-done@debbugs.gnu.org)
CAE4v=ph00fDkb8hauKNx5=ORdWva42Sb8=UgRSopqdWO9=zwLw@mail.gmail.com
It seems this was a filesystem corruption. Everything seems fine after reboot,
so closing.

Giovanni Biscuolo <g@xelera.eu> ezt írta (id?pont: 2019. nov. 7., Cs, 8:36):
Toggle quote (29 lines)
>
> Hello Gabor,
>
> Gábor Boskovits <boskovits@gmail.com> writes:
>
> [...]
>
> >> From the screenshot, it seems that your root file system (or at least
> >> /tmp and /gnu/store) became read-only, which in turn caused various
> >> things to fail, including guix-daemon (hence the “broken pipe” when
> >> ‘guix build’ was talking to it, I suppose.)
> >
> > Yes, it also became corrupted. fsck on boot fixed it.
> > Since then it works happily again...
>
> if the problem depended on filesystem corruption and all is fine now for
> you, could you also close this bug please?
>
> [...]
>
> Thanks! Gio'
>
> --
> Giovanni Biscuolo
>
> Xelera IT Infrastructures



--
OpenPGP Key Fingerprint: 7988:3B9F:7D6A:4DBF:3719:0367:2506:A96C:CF63:0B21
Closed
L
L
Ludovic Courtès wrote on 7 Nov 2019 21:58
(name . Gábor Boskovits)(address . boskovits@gmail.com)(address . 38088@debbugs.gnu.org)
87tv7fmmjl.fsf@gnu.org
Hi Gábor,

Gábor Boskovits <boskovits@gmail.com> skribis:

Toggle quote (21 lines)
> Ludovic Courtès <ludo@gnu.org> ezt írta (id?pont: 2019. nov. 6., Sze, 18:42):
>>
>> Hi Gábor,
>>
>> Gábor Boskovits <boskovits@gmail.com> skribis:
>>
>> > I did not know how to get the info better, so here is a screenshot
>> > about the situation.
>> >
>> > This happened when I was trying a pre-inst-env guix build from a
>> > core-updates checkout.
>> > Previously python3 failed to build, and I was trying to build it again.
>>
>> From the screenshot, it seems that your root file system (or at least
>> /tmp and /gnu/store) became read-only, which in turn caused various
>> things to fail, including guix-daemon (hence the “broken pipe” when
>> ‘guix build’ was talking to it, I suppose.)
>
> Yes, it also became corrupted. fsck on boot fixed it.
> Since then it works happily again...

OK. My guess is that there was maybe a transient error with your hard
disk, which led the kernel to remount the root file system read-only,
which in turn led to the failure of various programs, including
‘guix-daemon’.

I’m closing the bug, but do reopen it if you have reasons to suspect a bug!

Thanks,
Ludo’.
L
L
Ludovic Courtès wrote on 7 Nov 2019 21:59
control message for bug #38088
(address . control@debbugs.gnu.org)
87sgmzmmhw.fsf@gnu.org
tags 38088 + notabug
quit
?
Your comment

This issue is archived.

To comment on this conversation send an email to 38088@debbugs.gnu.org

To respond to this issue using the mumi CLI, first switch to it
mumi current 38088
Then, you may apply the latest patchset in this issue (with sign off)
mumi am -- -s
Or, compose a reply to this issue
mumi compose
Or, send patches to this issue
mumi send-email *.patch