bug#19248: installation of 0.8 stops at guile prompt.

bug#19248: installation of 0.8 stops at guile prompt.

[th3kent]

hello guix,

as requested by civodul on #guix, here is a bug report.

hardware: 8-year-old hp nx6125 with 512mb memory and 60gb h.d.d

first attempt at installation failed with the following:

--8<---------------cut here---------------start------------->8---
unionfs invoked oom-killer: gfp_mask=0x...
unionfs cpuset=/ mems_allowed=0
[...]
Out of memory: Kill process 2068 (guix) score 526 or sacrifice child
Killed process 2068 (guix) total-vm:829080kB, anon-rss:201792kB, file-rss:0kB
guix system: error: build failed: unexpected end-of-file
2470 operations
--8<---------------cut here---------------end--------------->8---

i went past this by enabling a swap device.

second attempt at installation failed with "no permissions" when copying
pata_* kernel modules to /gnu/store/*-linux-libre-3.17.2/.

after adding write permissions for user on directory
/gnu/store/*-linux-libre-3.17.2/, the installation completed
successfully (no errors).

on restarting the laptop and booting from the h.d.d, the reboot stops
at the guile-user prompt.  the backtrace shows the following:

--8<---------------cut here---------------start------------->8---
In /gnu/store/*-module-import/gnu/build/linux-boot.scm:
  385:7  2 (#<procedure 1e943c0 at /gnu/store/...
In srfi/srfi-1.scm:
  619:17  1 (for-each #<procedure load-linux-module* (file)> ...
--8<---------------cut here---------------end--------------->8---

regards,
... kennedy tembo.
-- 
former club president
midrand toastmasters
http://midrand.tmclub.co.za
 /"\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
 \ /       "[He has] campaigned steadfastly for freedom.
  X                I admire Nelson Mandela ..."
 / \  - Richard M. Stallman (rms), President, Free Software Foundation.
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

bug#19248: installation of 0.8 stops at guile prompt.

[Ludovic Courtès]

th3kent <tembokm@mweb.co.za> skribis:

> hardware: 8-year-old hp nx6125 with 512mb memory and 60gb h.d.d
>
> first attempt at installation failed with the following:
>
> unionfs invoked oom-killer: gfp_mask=0x...
> unionfs cpuset=/ mems_allowed=0
> [...]
> Out of memory: Kill process 2068 (guix) score 526 or sacrifice child
> Killed process 2068 (guix) total-vm:829080kB, anon-rss:201792kB, file-rss:0kB
> guix system: error: build failed: unexpected end-of-file
> 2470 operations

This is a sign that too much data was written to the unionfs that sits
on the root file system.

Could it be because you forgot to run ‘deco start cow-store /mnt’?  

> second attempt at installation failed with "no permissions" when copying
> pata_* kernel modules to /gnu/store/*-linux-libre-3.17.2/.

That is weird.  Do you remember at what point that happened?  Was it at
the end of the ‘guix system init config.scm /mnt’ run?

Files and directories under /gnu/store are meant to be immutable.

> on restarting the laptop and booting from the h.d.d, the reboot stops
> at the guile-user prompt.  the backtrace shows the following:
>
> In /gnu/store/*-module-import/gnu/build/linux-boot.scm:
>   385:7  2 (#<procedure 1e943c0 at /gnu/store/...
> In srfi/srfi-1.scm:
>   619:17  1 (for-each #<procedure load-linux-module* (file)> ...

Does it show an actual error below that?

Could you post your OS config?

Thanks for the report!

Ludo’.

bug#19248: installation of 0.8 stops at guile prompt.

[th3kent]

ludo@gnu.org (Ludovic Courtès) writes:
> th3kent <tembokm@mweb.co.za> skribis:
>> [...]
>> Out of memory: Kill process 2068 (guix) score 526 or sacrifice child
>> Killed process 2068 (guix) total-vm:829080kB, anon-rss:201792kB, file-rss:0kB
>> guix system: error: build failed: unexpected end-of-file
>> 2470 operations
>
> This is a sign that too much data was written to the unionfs that sits
> on the root file system.
>
> Could it be because you forgot to run ‘deco start cow-store /mnt’?  

i definitely ran ‘deco start cow-store /mnt’ after i created my o.s
config. 

>> second attempt at installation failed with "no permissions" when copying
>> pata_* kernel modules to /gnu/store/*-linux-libre-3.17.2/.
>
> That is weird.  Do you remember at what point that happened?  Was it at
> the end of the ‘guix system init config.scm /mnt’ run?
>
> Files and directories under /gnu/store are meant to be immutable.

yes, the "no permission" error was after running a second 'guix system
init ...' command.

>> on restarting the laptop and booting from the h.d.d, the reboot stops
>> at the guile-user prompt.  the backtrace shows the following:
>>
>> In /gnu/store/*-module-import/gnu/build/linux-boot.scm:
>>   385:7  2 (#<procedure 1e943c0 at /gnu/store/...
>> In srfi/srfi-1.scm:
>>   619:17  1 (for-each #<procedure load-linux-module* (file)> ...
>
> Does it show an actual error below that?
>
> Could you post your OS config?

the last line of the backtrace starts with "In Unkown ...".

here is my o.s config:

--8<---------------cut here---------------start------------->8---
(use-modules (gnu)
             (srfi srfi-26))

(operating-system
  (host-name "gnusys01")
  (timezone "Africa/Johannesburg")
  (locale "en_US.UTF-8")

  (initrd (lambda (file-systems . rest)
            (apply base-initrd file-systems
                   #:extra-modules '("pata_acpi.ko" "pata_atiixp.ko")
                   rest)))

  (bootloader (grub-configuration (device "/dev/sda")))
  (file-systems (cons (file-system
                        (device "root")
                        (title 'label)
                        (mount-point "/")
                        (type "ext4"))
                      %base-file-systems))

  (users (list (user-account
                (name "kent")
                (comment "the wizard of gnu")
                (group "users")

                (supplementary-groups '("wheel"
                                        "audio" "video"))
                (home-directory "/home/kent")))))
--8<---------------cut here---------------end--------------->8---

regards,
... kennedy tembo.
-- 
former club president
midrand toastmasters
http://midrand.tmclub.co.za
 /"\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
 \ /       "[He has] campaigned steadfastly for freedom.
  X                I admire Nelson Mandela ..."
 / \  - Richard M. Stallman (rms), President, Free Software Foundation.
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

bug#19248: installation of 0.8 stops at guile prompt.

[Ludovic Courtès]

th3kent <tembokm@mweb.co.za> skribis:

> ludo@gnu.org (Ludovic Courtès) writes:
>> th3kent <tembokm@mweb.co.za> skribis:

[...]

>>> on restarting the laptop and booting from the h.d.d, the reboot stops
>>> at the guile-user prompt.  the backtrace shows the following:
>>>
>>> In /gnu/store/*-module-import/gnu/build/linux-boot.scm:
>>>   385:7  2 (#<procedure 1e943c0 at /gnu/store/...
>>> In srfi/srfi-1.scm:
>>>   619:17  1 (for-each #<procedure load-linux-module* (file)> ...
>>
>> Does it show an actual error below that?
>>
>> Could you post your OS config?

[...]

>   (initrd (lambda (file-systems . rest)
>             (apply base-initrd file-systems
>                    #:extra-modules '("pata_acpi.ko" "pata_atiixp.ko")
>                    rest)))

This may be the culprit.  First, it’s not needed, because 0.8 already
included these two modules by default.  Second, 0.8 would fail when
trying to load the same module twice, which is exactly what would happen
here (this is fixed in ‘master’, around commit 0e704a2.)

Could you try to remove these ‘initrd’ lines and try again?

Thanks in advance,
Ludo’.

bug#19248: installation of 0.8 stops at guile prompt.

[kennedy]

ludo@gnu.org (Ludovic Courtès) writes:
> th3kent <tembokm@mweb.co.za> skribis:
>> ludo@gnu.org (Ludovic Courtès) writes:
>>> [...]
>>> Could you post your OS config?
>>   (initrd (lambda (file-systems . rest)
>>             (apply base-initrd file-systems
>>                    #:extra-modules '("pata_acpi.ko" "pata_atiixp.ko")
>>                    rest)))
>
> This may be the culprit.  First, it’s not needed, because 0.8 already
> included these two modules by default.  Second, 0.8 would fail when
> trying to load the same module twice, which is exactly what would happen
> here (this is fixed in ‘master’, around commit 0e704a2.)
>
> Could you try to remove these ‘initrd’ lines and try again?

i deleted the lines for "initrd" and the installation completed
successfully on first attempt.

however, the reboot after installation results in a kernel panic.  the
messages just before the panic are the following,

--8<---------------cut here---------------start------------->8---
groupadd: cannot lock /etc/group; try again later
usermod: no changes
adding user 'kent'
ERROR: In procedure getpw: entry not found
--8<---------------cut here---------------end--------------->8---

regards,
... kennedy tembo.
-- 
former club president
midrand toastmasters
http://midrand.tmclub.co.za
 /"\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
 \ /       "[He has] campaigned steadfastly for freedom.
  X                I admire Nelson Mandela ..."
 / \  - Richard M. Stallman (rms), President, Free Software Foundation.
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

bug#19248: installation of 0.8 stops at guile prompt.

[Ludovic Courtès]

kennedy <tembokm@mweb.co.za> skribis:

> ludo@gnu.org (Ludovic Courtès) writes:
>> th3kent <tembokm@mweb.co.za> skribis:
>>> ludo@gnu.org (Ludovic Courtès) writes:
>>>> [...]
>>>> Could you post your OS config?
>>>   (initrd (lambda (file-systems . rest)
>>>             (apply base-initrd file-systems
>>>                    #:extra-modules '("pata_acpi.ko" "pata_atiixp.ko")
>>>                    rest)))
>>
>> This may be the culprit.  First, it’s not needed, because 0.8 already
>> included these two modules by default.  Second, 0.8 would fail when
>> trying to load the same module twice, which is exactly what would happen
>> here (this is fixed in ‘master’, around commit 0e704a2.)
>>
>> Could you try to remove these ‘initrd’ lines and try again?
>
> i deleted the lines for "initrd" and the installation completed
> successfully on first attempt.

So you booted on the USB disk and re-run ‘guix system init’, is that
correct?

Did you reformat the root partition before doing that?

> however, the reboot after installation results in a kernel panic.  the
> messages just before the panic are the following,
>
> groupadd: cannot lock /etc/group; try again later

Could you try removing the /etc/.pwd.lock file on that partition?  (You
can do that on a system booted from the USB stick, and then reboot in
the installed system.)

Thanks,
Ludo’.

bug#19248: installation of 0.8 stops at guile prompt.

[th3kent]

ludo@gnu.org (Ludovic Courtès) writes:
> kennedy <tembokm@mweb.co.za> skribis:
>> i deleted the lines for "initrd" and the installation completed
>> successfully on first attempt.
>
> So you booted on the USB disk and re-run ‘guix system init’, is that
> correct?
>
> Did you reformat the root partition before doing that?

yes, i booted off my u.s.b disk, reformatted my root partition, and
re-installed.

>> however, the reboot after installation results in a kernel panic.  the
>> messages just before the panic are the following,
>>
>> groupadd: cannot lock /etc/group; try again later
>
> Could you try removing the /etc/.pwd.lock file on that partition?  (You
> can do that on a system booted from the USB stick, and then reboot in
> the installed system.)

i removed /etc/.pwd.lock, as well as /etc/group.lock, and the reboot
tries to enter the guile prompt ("Entering new prompt ...") and i get a
kernel panic.

just f.y.i,  i can't use shift-page.up to scrollback through the boot
messages.

regards,
... kennedy tembo.
-- 
former club president
midrand toastmasters
http://midrand.tmclub.co.za
 /"\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
 \ /       "[He has] campaigned steadfastly for freedom.
  X                I admire Nelson Mandela ..."
 / \  - Richard M. Stallman (rms), President, Free Software Foundation.
     ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

bug#19248: installation of 0.8 stops at guile prompt.

[Ludovic Courtès]

th3kent <tembokm@mweb.co.za> skribis:

> ludo@gnu.org (Ludovic Courtès) writes:
>> kennedy <tembokm@mweb.co.za> skribis:
>>> i deleted the lines for "initrd" and the installation completed
>>> successfully on first attempt.
>>
>> So you booted on the USB disk and re-run ‘guix system init’, is that
>> correct?
>>
>> Did you reformat the root partition before doing that?
>
> yes, i booted off my u.s.b disk, reformatted my root partition, and
> re-installed.
>
>>> however, the reboot after installation results in a kernel panic.  the
>>> messages just before the panic are the following,
>>>
>>> groupadd: cannot lock /etc/group; try again later
>>
>> Could you try removing the /etc/.pwd.lock file on that partition?  (You
>> can do that on a system booted from the USB stick, and then reboot in
>> the installed system.)
>
> i removed /etc/.pwd.lock, as well as /etc/group.lock, and the reboot
> tries to enter the guile prompt ("Entering new prompt ...") and i get a
> kernel panic.

But do you actually get the Guile prompt?

If you did and were brave enough, we could coordinate on IRC to try and
investigate, for instance to check whether / is writable, things like
that.

> just f.y.i,  i can't use shift-page.up to scrollback through the boot
> messages.

I’m running out of ideas, sorry.

If that is an option for you, could you try starting anew?  That is,
boot on the USB image, erase the target partition with “mkfs.ext4
/dev/xyzN”, and proceed with installation.  We must have missed
something, but I don’t see what.

Thanks,
Ludo’.

bug#19248: installation of 0.8 stops at guile prompt.

[Ludovic Courtès]

tags 19248 moreinfo unreproducible
close 19248
thanks

Hello,

I’m closing this bug for now.  I would be grateful if you could try
installing 0.8.1, released two weeks ago, and let us know if you can
reproduce this problem.

Thanks in advance,
Ludo’.