bug#30522: offloading requires guile-readline on the remote system

bug#30522: offloading requires guile-readline on the remote system

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 482 bytes --]

Hello!

After reconfiguring a remote machine on latest 'master' (post recent
core-updates merge), I found that offloading to it would just "hang".

After some debugging I found that `guile --listen` failed to import
(ice-9 readline).  Installing guile-readline for the offload user did
the trick.

It would be great to make Guile fail gracefully and print a message
about readline being unavailable instead of just looping endlessly.  Or
at the very least document the requirement.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

bug#30522: offloading requires guile-readline on the remote system

[Ludovic Courtès]

Hi Marius,

Marius Bakke <mbakke@fastmail.com> skribis:

> After reconfiguring a remote machine on latest 'master' (post recent
> core-updates merge), I found that offloading to it would just "hang".
>
> After some debugging I found that `guile --listen` failed to import
> (ice-9 readline).  Installing guile-readline for the offload user did
> the trick.

Could it be that ~/.guile on that machine imports (ice-9 readline)?
(On GuixSD the default ~/.guile used to import it unconditionally.)

Otherwise I see no reason why things would go wrong.  For instance, this
appears to work fine:

--8<---------------cut here---------------start------------->8---
$ ./pre-inst-env guix environment --ad-hoc guile -C -- guile --listen
substitute: updating list of substitutes from 'https://berlin.guixsd.org'... 100.0%
substitute: updating list of substitutes from 'https://mirror.hydra.gnu.org'... 100.0%
The following derivations will be built:
   /gnu/store/5y0fr7fbmpvbry09ks7i0fqaqvrl2yjn-profile.drv
   /gnu/store/ffp06ykw5qjlj3c2drys30nf73lgf6h4-info-dir.drv
   /gnu/store/209l6232jpic6cj90ijss2122xz743ia-fonts-dir.drv
   /gnu/store/1hj38jwa8bhkgwxxhrflbaslkql37db5-ca-certificate-bundle.drv
   /gnu/store/3hk4zjhl5g3i94yfdw5mzmbyhfrwqwxx-manual-database.drv
Creating manual page database...
1 entries processed in 0.1 s
GNU Guile 2.2.3
Copyright (C) 1995-2017 Free Software Foundation, Inc.

Guile comes with ABSOLUTELY NO WARRANTY; for details type `,show w'.
This program is free software, and you are welcome to redistribute it
under certain conditions; type `,show c' for details.

Enter `,help' for help.
scheme@(guile-user)> +
$1 = #<procedure + (#:optional _ _ . _)>
scheme@(guile-user)> -
$2 = #<procedure - (#:optional _ _ . _)>
--8<---------------cut here---------------end--------------->8---

Thanks,
Ludo’.

bug#30522: offloading requires guile-readline on the remote system

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 892 bytes --]

Ludovic Courtès <ludo@gnu.org> writes:

> Hi Marius,
>
> Marius Bakke <mbakke@fastmail.com> skribis:
>
>> After reconfiguring a remote machine on latest 'master' (post recent
>> core-updates merge), I found that offloading to it would just "hang".
>>
>> After some debugging I found that `guile --listen` failed to import
>> (ice-9 readline).  Installing guile-readline for the offload user did
>> the trick.
>
> Could it be that ~/.guile on that machine imports (ice-9 readline)?
> (On GuixSD the default ~/.guile used to import it unconditionally.)

Ah, good catch!  This was using the default .guile from before
2ba2c98d8aa695cce608313833a36c61b0fb3428.  I have verified that copying
the new .guile from /etc/skel works even without guile-readline
installed.

It would still be good with a better failure mode, but I'll file this
under "pebkac" for now.

Thanks!

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

bug#30522: offloading requires guile-readline on the remote system

[Ludovic Courtès]

Heya,

Marius Bakke <mbakke@fastmail.com> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> Hi Marius,
>>
>> Marius Bakke <mbakke@fastmail.com> skribis:
>>
>>> After reconfiguring a remote machine on latest 'master' (post recent
>>> core-updates merge), I found that offloading to it would just "hang".
>>>
>>> After some debugging I found that `guile --listen` failed to import
>>> (ice-9 readline).  Installing guile-readline for the offload user did
>>> the trick.
>>
>> Could it be that ~/.guile on that machine imports (ice-9 readline)?
>> (On GuixSD the default ~/.guile used to import it unconditionally.)
>
> Ah, good catch!  This was using the default .guile from before
> 2ba2c98d8aa695cce608313833a36c61b0fb3428.  I have verified that copying
> the new .guile from /etc/skel works even without guile-readline
> installed.

Good.

> It would still be good with a better failure mode, but I'll file this
> under "pebkac" for now.

Yes, I’m surprised it ended up looping.  Did you try to strace it or
something to see what was going on?  Perhaps guile failed to start
altogether and Guile-SSH kept retrying?

Ludo’.

bug#30522: offloading requires guile-readline on the remote system

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 410 bytes --]

Ludovic Courtès <ludo@gnu.org> writes:

>> It would still be good with a better failure mode, but I'll file this
>> under "pebkac" for now.
>
> Yes, I’m surprised it ended up looping.  Did you try to strace it or
> something to see what was going on?  Perhaps guile failed to start
> altogether and Guile-SSH kept retrying?

Indeed.  Guile-SSH kept trying to start `guile --listen` over and over.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

bug#30522: Guile-SSH retries upon “guile --listen” failure to start

[Ludovic Courtès]

Hello Artyom & Marius,

Marius Bakke <mbakke@fastmail.com> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>>> It would still be good with a better failure mode, but I'll file this
>>> under "pebkac" for now.
>>
>> Yes, I’m surprised it ended up looping.  Did you try to strace it or
>> something to see what was going on?  Perhaps guile failed to start
>> altogether and Guile-SSH kept retrying?
>
> Indeed.  Guile-SSH kept trying to start `guile --listen` over and over.

Artyom, do you think we could arrange for Guile-SSH to properly detect
failure to start ‘guile --listen’ and report it instead of retrying?

Perhaps it could check the exit code of ‘guile --listen’ and distinguish
127 (“command not found”) from other exit codes?

(For more context, see <https://bugs.gnu.org/30522>.)

Thanks,
Ludo’.

bug#30522: Guile-SSH retries upon “guile --listen” failure to start

[Artyom Poptsov]

[-- Attachment #1: Type: text/plain, Size: 587 bytes --]

Hello Ludovic and other Guilers,

> Artyom, do you think we could arrange for Guile-SSH to properly detect
> failure to start ‘guile --listen’ and report it instead of retrying?

sure, I'll investigate the problem in a week or so.  Currently busy with
my day job and personal hacktivities.

Added the task to my TODO list.

Thanks for reporting the bug!

- Artyom

-- 
Artyom V. Poptsov <poptsov.artyom@gmail.com>
Home page: http://poptsov-artyom.narod.ru/
CADR Hackerspace co-founder: https://cadrspace.ru/
GPG: D0C2 EAC1 3310 822D 98DE  B57C E9C5 A2D9 0898 A02F

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

bug#30522: Guile-SSH retries upon “guile --listen” failure to start

[Artyom Poptsov]

[-- Attachment #1: Type: text/plain, Size: 454 bytes --]

Hello Ludovic and Marius,

I implemented a fix for the looping problem; could you please check if
the patch works for you?

The changes are committed in e24401f1ef56b1f3d7ef847c54fb0c2fe9cf3cf8 on
the 'master' branch.

Thanks!

- Artyom

-- 
Artyom V. Poptsov <poptsov.artyom@gmail.com>
Home page: http://poptsov-artyom.narod.ru/
CADR Hackerspace co-founder: https://cadrspace.ru/
GPG: D0C2 EAC1 3310 822D 98DE  B57C E9C5 A2D9 0898 A02F

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 832 bytes --]

bug#30522: Guile-SSH retries upon “guile --listen” failure to start

[Ludovic Courtès]

Hi Artyom,

Artyom Poptsov <poptsov.artyom@gmail.com> skribis:

> I implemented a fix for the looping problem; could you please check if
> the patch works for you?
>
> The changes are committed in e24401f1ef56b1f3d7ef847c54fb0c2fe9cf3cf8 on
> the 'master' branch.

I’d rather let someone else test it :-), but the patch LGTM.

Thank you!

Ludo’.