GNU bug report logs - #42002
'guix pull' build processes get SIGKILL on ARMv7

Previous Next

Full log

Message #28 received at 42002 <at> debbugs.gnu.org (full text, mbox):

From: Ludovic Courtès <ludo <at> gnu.org>
To: Leo Famulari <leo <at> famulari.name>
Cc: Léon Lain Delysid <leon.lain.delysid <at> gmail.com>,
 Efraim Flashner <efraim <at> flashner.co.il>, 42002 <at> debbugs.gnu.org
Subject: Re: bug#42002: Bug: Signal kill during build
Date: Tue, 23 Jun 2020 23:26:11 +0200

Leo Famulari <leo <at> famulari.name> skribis:

> On Mon, Jun 22, 2020 at 03:47:08PM +0300, Efraim Flashner wrote:
>> The signal 9 (killed) makes me think a C++ program killed. I checked
>> wikipedia and it says the Banana Pi M3 has 2GB of RAM. Was there
>> anything else running at the time? 'guix pull' can be resource
>> intensive, especially on lower powered machines.
>
> I think it should be possible with 2GB of RAM, if you limit the number
> of concurrent builds.
>
> For example, something like `guix pull --max-jobs=1`.
>
> And I think the memory requirements should decrease with the recently
> released Guile 3.0.3:
>
> https://lists.gnu.org/archive/html/guile-devel/2020-06/msg00008.html

For the record, I’ve successfully run the following command on my A20
OLinuXino with a bit less than 1 GiB of RAM:

  guix pull --commit=380c3e459b6ebf0077008886954ff881a0cc7924

It takes ages, but fortunately substitutes were available for this
revision.

Before substitutes were available, it went OOM (out of memory) as in
your case while building guix-system.drv.

‘guix pull’ is still barely usable on these low-end devices, though with
Guile 3.0.3 it’s better.

Ludo’.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.