GNU bug report logs - #26035
guix pull does not handle url lists and unreachable requests

Previous Next

To reply to this bug, email your comments to 26035 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: ng0 <contact.ng0 <at> cryptolab.net>
To: bug-guix <at> gnu.org
Subject: guix pull does not handle url lists and unreachable requests
Date: Thu, 9 Mar 2017 14:51:05 +0000

efficient enough and sets the times out much too high
Reply-To: 
In-Reply-To:
When you configure a system to use bayfront, for example with these
lines included in cons* of (services):
            (modify-services
              %base-services
              (guix-service-type
                config =>
                (guix-configuration
                  (inherit config)
                  (substitute-urls
                    (cons* "https://bayfront.guixsd.org"
                           %default-substitute-urls))
                  (authorized-keys
                    (cons*
                      (plain-file "bayfront.guixsd.org.pub"
                        (string-append "(public-key (ecc curve
Ed25519) "
                                       "(q
#8D156F295D24B0D9A86FA5741A840FF2"

"D24F60F7B6C4134814AD55625971B394#)))"))
                    %default-authorized-guix-keys))))))))

please consider that this does not reflect the actual indent and that I
can't paste all of the config here, but those are the relevant parts,
you will experience that: "guix package", and "guix system" (or guix
buld?) query all listed subsitute urls, while "guix pull" only queries
the first one. If the first one is experiencing issues (for example
server is offline or can't handle requests for whatever reasons), the
timeout is much too high. I had to cancel guix pull processes which
tried to request from bayfront.guixsd.org for 9, 12, and 3 hours.

As bayfront wasn't the only server, a switch should occur even with the
guix pull part where substitute servers are queried before the build of
the fetched master tarball starts.

Message #8 received at 26035 <at> debbugs.gnu.org (full text, mbox):

From: ng0 <contact.ng0 <at> cryptolab.net>
To: 26035 <at> debbugs.gnu.org
Subject: Re: bug#26035: guix pull does not handle url lists and unreachable
 requests
Date: Thu, 9 Mar 2017 15:02:23 +0000

ng0 transcribed 1.6K bytes:
> efficient enough and sets the times out much too high

Curses, my mailing program broke the subject. I thought I fixed it.

> Reply-To: 
> In-Reply-To:
> When you configure a system to use bayfront, for example with these
> lines included in cons* of (services):
>             (modify-services
>               %base-services
>               (guix-service-type
>                 config =>
>                 (guix-configuration
>                   (inherit config)
>                   (substitute-urls
>                     (cons* "https://bayfront.guixsd.org"
>                            %default-substitute-urls))
>                   (authorized-keys
>                     (cons*
>                       (plain-file "bayfront.guixsd.org.pub"
>                         (string-append "(public-key (ecc curve
> Ed25519) "
>                                        "(q
> #8D156F295D24B0D9A86FA5741A840FF2"
> 
> "D24F60F7B6C4134814AD55625971B394#)))"))
>                     %default-authorized-guix-keys))))))))
> 
> please consider that this does not reflect the actual indent and that I
> can't paste all of the config here, but those are the relevant parts,
> you will experience that: "guix package", and "guix system" (or guix
> buld?) query all listed subsitute urls, while "guix pull" only queries
> the first one. If the first one is experiencing issues (for example
> server is offline or can't handle requests for whatever reasons), the
> timeout is much too high. I had to cancel guix pull processes which
> tried to request from bayfront.guixsd.org for 9, 12, and 3 hours.
> 
> As bayfront wasn't the only server, a switch should occur even with the
> guix pull part where substitute servers are queried before the build of
> the fetched master tarball starts.
> 
> 
> 

Message #13 received at 26035 <at> debbugs.gnu.org (full text, mbox):

From: zimoun <zimon.toutoune <at> gmail.com>
To: Maxim Cournoyer <maxim.cournoyer <at> gmail.com>
Cc: 26035 <at> debbugs.gnu.org, 26833 <at> debbugs.gnu.org
Subject: Re: bug#26833: `guix substitute --substitute' dies on unreachable
 substitute server
Date: Fri, 20 Aug 2021 13:04:06 +0200

Hi Maxim,

Your recent investigation bug#30290 [0] rings this bell:

   <http://issues.guix.gnu.org/issue/26833>

If it appears the same issue, maybe we could merge them.  WDYT?

On Mon, 08 May 2017 at 08:40, Maxim Cournoyer <maxim.cournoyer <at> gmail.com> wrote:
> Problem: When using multiple substitute servers, if one of the server is
> temporarily unreachable, `guix substitute --substitute' will throw an
> exception. Example (while bayfront server was down, attempting to build
> a package only available locally (emacs-dvc)):
>
> guix package -i emacs-dvc
> ;;; note: source file /home/maxim/.config/guix/latest/gnu/packages/emacs.scm
> ;;;       newer than compiled /home/maxim/.config/guix/latest/gnu/packages/emacs.go
> ;;; note: source file /home/maxim/.config/guix/latest/gnu/packages/emacs.scm
> ;;;       newer than compiled /gnu/store/nqy9m6hhnkkfwr5wyq5bac96v9s9hc9i-guix-0.12.0-9.25a4/lib/guile/2.0/site-ccache/gnu/packages/emacs.go
> ;;; note: source file /home/maxim/.config/guix/latest/gnu/packages/emacs.scm
> ;;;       newer than compiled /run/current-system/profile/lib/guile/2.0/site-ccache/gnu/packages/emacs.go
> ;;; note: source file /home/maxim/.config/guix/latest/gnu/packages/emacs.scm
> ;;;       newer than compiled /home/maxim/.cache/guile/ccache/2.0-LE-8-2.0/home/maxim/src/guix/gnu/packages/emacs.scm.go
> The following package will be installed:
>    emacs-dvc    trunk-1.591     /gnu/store/sraxmg5qz9i4338s4ks7asgy4v68dgqs-emacs-dvc-trunk-1.591
>>
> substitute: updating list of substitutes from 'https://mirror.hydra.gnu.org'... 100.0%
> substitute: updating list of substitutes from 'https://mirror.hydra.gnu.org'... 100.0%
> Downloading https://bayfront.guixsd.org/nar/gzip/6h7ym07plwxfn4zq53ld8zfpbx3a09al-at-spi2-core-2.22.0 (1.1MiB installed)...
> guix substitute: error: connect: No route to host
> killing process 13896
> killing process 13896: No such process
>
> Expected: Since multiple substitute servers are being used, rather than
> bombing out on the first unavailable one, `guix substitute --substitute'
> (or any other command implicated with substitute servers) should simply
> warn about it before attempting the next one.
>
> Bonus point: `guix substitute --query' should also print a warning
> message that substitute server X couldn't be reached (it doesn't print
> anything about that right now, just skips to the next one).


Cheers,
simon

0: <http://issues.guix.gnu.org/issue/30290#10>

Previous Next

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