unofficial mirror of bug-guix@gnu.org 
 help / color / mirror / code / Atom feed
From: "Ludovic Courtès" <ludo@gnu.org>
To: Danny Milosavljevic <dannym@scratchpost.org>
Cc: 38167@debbugs.gnu.org
Subject: bug#38167: guix pull takes over 8 GiB of memory to finish if there are no substitutes
Date: Sat, 28 Dec 2019 15:52:40 +0100	[thread overview]
Message-ID: <877e2gbiyv.fsf@gnu.org> (raw)
In-Reply-To: <20191228120128.010d72ee@scratchpost.org> (Danny Milosavljevic's message of "Sat, 28 Dec 2019 12:01:28 +0100")

Hi,

Danny Milosavljevic <dannym@scratchpost.org> skribis:

> On Fri, 27 Dec 2019 19:11:08 +0100
> Ludovic Courtès <ludo@gnu.org> wrote:
>
>> Danny Milosavljevic <dannym@scratchpost.org> skribis:
>> 
>> > USER       PID %CPU %MEM    VSZ   RSS TTY      STAT START   TIME COMMAND
>> > dannym   19221 20.8 87.4 9404812 6884184 pts/0 Tl   20:34   2:40 /gnu/store/sc7z07gim1iq5zvfz1amdwf2irxrzifg-guile-2.2.6/bin/guile --no-auto-compile /home/dannym/.config/guix/current/bin/guix pull  
>> 
>> Oh, that’s an RSS of 6 GiB for ‘guix pull’ itself?  Weird, I don’t see
>> how that can happen.
>> 
>> Could it be that ~/.cache/guix/checkouts/THE-THING is too big, which in
>> turn causes libgit2 to consume too much memory somehow?
>
> $ du -hs ~/.cache/guix/checkouts/
> 395M    /home/dannym/.cache/guix/checkouts/

OK, nothing bad.

>> What happens if you attach strace to this process at the moment where
>> it’s consuming a lot of memory?  Is it traversing Git pack files or
>> something like that?
>
> Right now the process is in a paused state and I think when I attach strace
> it will continue.  Should I still do it?  I don't want to destroy our debugging
> opportunity.

I think you should resume it (SIGCONT), attach strace, and grab a few
seconds of log so we have an idea of what it’s doing.  You can also
check /proc/PID/fd and keep a copy of /proc/PID/maps while we’re at it.

Thanks!

Ludo’.

  reply	other threads:[~2019-12-28 14:53 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-11-11  7:06 bug#38167: guix pull takes over 8 GiB of memory to finish if there are no substitutes Danny Milosavljevic
2019-11-11  7:37 ` Christopher Baines
2019-11-11 20:42   ` Ludovic Courtès
2019-12-26 19:45     ` Danny Milosavljevic
2019-12-26 19:48       ` Danny Milosavljevic
2019-12-27 18:11         ` Ludovic Courtès
2019-12-28 11:01           ` Danny Milosavljevic
2019-12-28 14:52             ` Ludovic Courtès [this message]
2019-12-28 17:55               ` Danny Milosavljevic
2019-12-29 22:41                 ` Ludovic Courtès
2019-12-31 15:44                   ` Danny Milosavljevic
2019-12-31 18:18                     ` Ludovic Courtès
2020-01-01 12:51                       ` Danny Milosavljevic
2020-10-09  2:11                         ` Maxim Cournoyer
2019-12-27 18:08       ` Ludovic Courtès
2019-12-28 10:56         ` Danny Milosavljevic

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://guix.gnu.org/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=877e2gbiyv.fsf@gnu.org \
    --to=ludo@gnu.org \
    --cc=38167@debbugs.gnu.org \
    --cc=dannym@scratchpost.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/guix.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).