From debbugs-submit-bounces@debbugs.gnu.org Thu Dec 27 23:22:35 2018 Received: (at 33515) by debbugs.gnu.org; 28 Dec 2018 04:22:35 +0000 Received: from localhost ([127.0.0.1]:39653 helo=debbugs.gnu.org) by debbugs.gnu.org with esmtp (Exim 4.84_2) (envelope-from ) id 1gcjfO-0000sG-LN for submit@debbugs.gnu.org; Thu, 27 Dec 2018 23:22:34 -0500 Received: from world.peace.net ([64.112.178.59]:53592) by debbugs.gnu.org with esmtp (Exim 4.84_2) (envelope-from ) id 1gcjfM-0000s2-GU for 33515@debbugs.gnu.org; Thu, 27 Dec 2018 23:22:32 -0500 Received: from mhw by world.peace.net with esmtpsa (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.89) (envelope-from ) id 1gcjfG-0007WG-CE; Thu, 27 Dec 2018 23:22:26 -0500 From: Mark H Weaver To: Ludovic =?utf-8?Q?Court=C3=A8s?= Subject: Re: [bug#33515] [PATCH 0/5] Cuirass/Hydra: evaluate jobs in an inferior References: <20181126163757.17399-1-ludo@gnu.org> <87bm56aniw.fsf@gnu.org> Date: Thu, 27 Dec 2018 23:21:18 -0500 In-Reply-To: <87bm56aniw.fsf@gnu.org> ("Ludovic \=\?utf-8\?Q\?Court\=C3\=A8s\=22'\?\= \=\?utf-8\?Q\?s\?\= message of "Thu, 27 Dec 2018 18:27:03 +0100") Message-ID: <87bm56waau.fsf@netris.org> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.1 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-Spam-Score: 0.0 (/) X-Debbugs-Envelope-To: 33515 Cc: 33515@debbugs.gnu.org X-BeenThere: debbugs-submit@debbugs.gnu.org X-Mailman-Version: 2.1.18 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: debbugs-submit-bounces@debbugs.gnu.org Sender: "Debbugs-submit" X-Spam-Score: -1.0 (-) Hi Ludovic, Ludovic Court=C3=A8s writes: > Ludovic Court=C3=A8s skribis: > >> This patch set changes the way we compute continuous integration jobs: >> instead of letting Guile auto-compile all of Guix from its checkout, >> we first build Guix in the same way as =E2=80=98guix pull=E2=80=99, open= an inferior >> to that Guix, and run the job evaluation code in that inferior. >> >> I think it=E2=80=99s cleaner and it should be faster and less resource-h= ungry >> than the current approach. >> >> The build-aux/hydra/gnu-system.scm file will now rely on the >> (guix channels) and (guix inferior) with the new =E2=80=98checkout->chan= nel-instance=E2=80=99 >> and =E2=80=98inferior-eval-with-store=E2=80=99 procedures, which means t= hat Cuirass >> (and Hydra) will need to be using a recent Guix to be able to perform >> the evaluation. Apart from that =E2=80=98gnu-system.scm=E2=80=99 is rat= her decoupled >> from the Guix APIs. > > I=E2=80=99ve pushed the patches to the =E2=80=98wip-ci-inferior=E2=80=99 = branch and created > these two jobsets: > > https://berlin.guixsd.org/jobset/wip-ci-inferior > https://hydra.gnu.org/jobset/gnu/wip-ci-inferior > > Evaluation with Cuirass on berlin went fine. > > I suspect evaluation on hydra will fail though, because it=E2=80=99s prob= ably > running an older Guix version; we=E2=80=99ll have to upgrade there. Mark= , would > you like to take a look? Indeed, the evaluation eventually failed on hydra.gnu.org. I then upgraded 'guix' in both root's and hydra's profiles to the latest version and tried again. This time it failed almost immediately, with the following error: --8<---------------cut here---------------start------------->8--- evaluator hydra-eval-guile-jobs hydra-eval-guile-jobs returned exit code 1: adding `/gnu/store/y4m61z51s28kmiff2hzbr7xm6f4lsk80-git-export' to the load= path Backtrace: In ice-9/eval.scm: 293:34 19 (_ #) In ice-9/boot-9.scm: 2862:4 18 (define-module* _ #:filename _ #:pure _ #:version _ # _ ?) 2875:24 17 (_) 222:17 16 (map1 (((guix git)) ((guix records)) ((guix gexp)) (#) ?)) 2788:17 15 (resolve-interface (guix git) #:select _ #:hide _ # _ # ?) 2714:10 14 (_ (guix git) _ _ #:ensure _) 2982:16 13 (try-module-autoload _ _) 2312:4 12 (save-module-excursion #) 3002:22 11 (_) In unknown file: 10 (primitive-load-path "guix/git" #) In ice-9/eval.scm: 721:20 9 (primitive-eval (define-module (guix git) #:use-module ?)) In ice-9/psyntax.scm: 1235:36 8 (expand-top-sequence ((define-module (guix git) # # ?)) ?) 1182:24 7 (parse _ (("placeholder" placeholder)) ((top) #(# # ?)) ?) 285:10 6 (parse _ (("placeholder" placeholder)) (()) _ c&e (eval) ?) In ice-9/eval.scm: 293:34 5 (_ #) In ice-9/boot-9.scm: 2862:4 4 (define-module* _ #:filename _ #:pure _ #:version _ # _ ?) 2875:24 3 (_) 222:17 2 (map1 (((git)) ((git object)) ((guix i18n)) ((guix ?)) ?)) 2791:6 1 (resolve-interface _ #:select _ #:hide _ #:prefix _ # _ ?) In unknown file: 0 (scm-error misc-error #f "~A ~S" ("no code for modu?" ?) ?) ERROR: In procedure scm-error: no code for module (git) Some deprecated features have been used. Set the environment variable GUILE_WARN_DEPRECATED to "detailed" and rerun the program to get more information. Set it to "no" to suppress this message. --8<---------------cut here---------------end--------------->8--- For now, I rolled back both of the profiles that I updated. This looks similar to the errors I encountered when trying to update 'guix' on the armhf build slaves, although the module it failed to load in that case was different. In both cases, the wrapper for 'guix' is failing to populate the GUILE_LOAD_PATH with all of the needed modules. Mark