bug#74967: About /etc/subuid and /etc/subgid commits

bug#74967: About /etc/subuid and /etc/subgid commits

[Nicolas Graves]

Hi Giacomo,

I noticed the last 11 commits are breaking my peculiar tmpfs root
configuration.  More precisely, I have more or less this layout :
tmpfs: /, $HOME
vfat32: /boot/efi
btrfs: /var/guix, /gnu/store, /var/lib, /boot ... + several directories
under $HOME

That means that at startup, I won't have anything in /etc nor $HOME
basically. To deploy guix home properly, I run a pam_exec on login which
symlinks /var/guix/profiles/per-user/${USER}/guix-home && current-guix
to their expected places, then runs the home activation script, and then
logins.

It is a bit circumvoluted, but it worked properly on my side on
guix@478b9ccea8. Do you see a reason why your patches could break my
user login?

I can at least confirm that the $PATH for my $USER is not set properly
after login after these patches, before it was.

-- 
Best regards,
Nicolas Graves

bug#74967: About /etc/subuid and /etc/subgid commits

[paul via Bug reports for GNU Guix]

[-- Attachment #1: Type: text/plain, Size: 924 bytes --]

Hi Nicolas, apologies for breaking your system :( Is is possible for you 
to try and bisect the history to find the culprit commit?

On 12/19/24 10:36, Nicolas Graves wrote:
> It is a bit circumvoluted, but it worked properly on my side on
> guix@478b9ccea8. Do you see a reason why your patches could break my
> user login?

I'm not sure, I was wondering: do you use/extend the subids-service-type 
in your configuration or did you already have manually setup /etc/subuid 
or /etc/subgid? Otherwise I'm not sure how it can impact your system, if 
you could share even some parts of your operating-system configuration 
it would make finding the problem a little easier.

Since you mentioned PAM, these two changes that I authored were recently 
merged. Could you try and see whether your system work at commit 
a1ecd7f56c4ffadc49d5501a0df7f4c4556120c2 which is the parent of the 
first pam change?

Thank you very much :)


[-- Attachment #2: Type: text/html, Size: 1391 bytes --]

bug#74967: About /etc/subuid and /etc/subgid commits

[paul via Bug reports for GNU Guix]

I forgot to attach the two PAM related commits, apologies for the noise.

[0]: 
https://git.savannah.gnu.org/cgit/guix.git/commit/?id=3eb720bd0b4bab233918f979bb51b58437db06ad
[1]: 
https://git.savannah.gnu.org/cgit/guix.git/commit/?id=9411a1495f7cb2b116129a71a1fc1817e962fa5d

bug#74967: About /etc/subuid and /etc/subgid commits

[Nicolas Graves]

On 2024-12-19 12:32, paul wrote:

> Hi Nicolas, apologies for breaking your system :( Is is possible for you 
> to try and bisect the history to find the culprit commit?
>
> On 12/19/24 10:36, Nicolas Graves wrote:
>> It is a bit circumvoluted, but it worked properly on my side on
>> guix@478b9ccea8. Do you see a reason why your patches could break my
>> user login?
>
> I'm not sure, I was wondering: do you use/extend the subids-service-type 
> in your configuration or did you already have manually setup /etc/subuid 
> or /etc/subgid? Otherwise I'm not sure how it can impact your system, if 
> you could share even some parts of your operating-system configuration 
> it would make finding the problem a little easier.
>
> Since you mentioned PAM, these two changes that I authored were recently 
> merged. Could you try and see whether your system work at commit 
> a1ecd7f56c4ffadc49d5501a0df7f4c4556120c2 which is the parent of the 
> first pam change?
>
> Thank you very much :)

I've tried bisecting and this time it seems to work properly.  I'm not
sure what it was then, but it works fine now!

-- 
Best regards,
Nicolas Graves

bug#74967: About /etc/subuid and /etc/subgid commits

[paul via Bug reports for GNU Guix]

[-- Attachment #1: Type: text/plain, Size: 1349 bytes --]

Hi Nicolas, awesome! I'm closing this.


Cheers,

giacomo

Il 21 dicembre 2024 17:49:08 CET, Nicolas Graves <ngraves@ngraves.fr> ha scritto:
>On 2024-12-19 12:32, paul wrote:
>
>> Hi Nicolas, apologies for breaking your system :( Is is possible for you 
>> to try and bisect the history to find the culprit commit?
>>
>> On 12/19/24 10:36, Nicolas Graves wrote:
>>> It is a bit circumvoluted, but it worked properly on my side on
>>> guix@478b9ccea8. Do you see a reason why your patches could break my
>>> user login?
>>
>> I'm not sure, I was wondering: do you use/extend the subids-service-type 
>> in your configuration or did you already have manually setup /etc/subuid 
>> or /etc/subgid? Otherwise I'm not sure how it can impact your system, if 
>> you could share even some parts of your operating-system configuration 
>> it would make finding the problem a little easier.
>>
>> Since you mentioned PAM, these two changes that I authored were recently 
>> merged. Could you try and see whether your system work at commit 
>> a1ecd7f56c4ffadc49d5501a0df7f4c4556120c2 which is the parent of the 
>> first pam change?
>>
>> Thank you very much :)
>
>I've tried bisecting and this time it seems to work properly.  I'm not
>sure what it was then, but it works fine now!
>
>-- 
>Best regards,
>Nicolas Graves

[-- Attachment #2: Type: text/html, Size: 1984 bytes --]

bug#74967: About /etc/subuid and /etc/subgid commits

[paul via Bug reports for GNU Guix]

[-- Attachment #1: Type: text/plain, Size: 1347 bytes --]

Hi Nicolas, awesome! I'm closing this.

Cheers,

giacomo

Il 21 dicembre 2024 17:49:08 CET, Nicolas Graves <ngraves@ngraves.fr> ha scritto:
>On 2024-12-19 12:32, paul wrote:
>
>> Hi Nicolas, apologies for breaking your system :( Is is possible for you 
>> to try and bisect the history to find the culprit commit?
>>
>> On 12/19/24 10:36, Nicolas Graves wrote:
>>> It is a bit circumvoluted, but it worked properly on my side on
>>> guix@478b9ccea8. Do you see a reason why your patches could break my
>>> user login?
>>
>> I'm not sure, I was wondering: do you use/extend the subids-service-type 
>> in your configuration or did you already have manually setup /etc/subuid 
>> or /etc/subgid? Otherwise I'm not sure how it can impact your system, if 
>> you could share even some parts of your operating-system configuration 
>> it would make finding the problem a little easier.
>>
>> Since you mentioned PAM, these two changes that I authored were recently 
>> merged. Could you try and see whether your system work at commit 
>> a1ecd7f56c4ffadc49d5501a0df7f4c4556120c2 which is the parent of the 
>> first pam change?
>>
>> Thank you very much :)
>
>I've tried bisecting and this time it seems to work properly.  I'm not
>sure what it was then, but it works fine now!
>
>-- 
>Best regards,
>Nicolas Graves

[-- Attachment #2: Type: text/html, Size: 1980 bytes --]