Re: master df1e553688 4/5: Accommodate nonstandard turbo file senders in erc-dcc

Re: master df1e553688 4/5: Accommodate nonstandard turbo file senders in erc-dcc

[Lars Ingebrigtsen]

Something in recent erc changes is making the EMBA build test fail.  For
instance:

https://emba.gnu.org/emacs/emacs/-/jobs/46150

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master df1e553688 4/5: Accommodate nonstandard turbo file senders in erc-dcc

[F. Jason Park]

Hi Lars,

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Something in recent erc changes is making the EMBA build test fail.  For
> instance:
>
> https://emba.gnu.org/emacs/emacs/-/jobs/46150

This links to a job on my branch for fix/bug-48598, so I'm guessing you
meant to highlight something else. (More info on that specific failure
below, just in case anyone's curious.) The subject line of this email
references a commit on master. If it triggered a pipeline named after
it, I can't find it. Do you have a different link by any chance?

My initial push for that same bug set triggered this pipeline:

  https://emba.gnu.org/emacs/emacs/-/pipelines/18010

which didn't build anything. But since you say "build test," I'm
guessing you mean build-image-inotify (or similar), which it seems is
included in those "scheduled 3x daily" pipelines and some but not all
pipelines triggered by routine pushes.

There actually was a more recent failure on master (triggered by a push
from me) that ended in a weird timeout or hang, namely:

> Subject: Failed pipeline for master | emacs | 4d60dcd8
> Newsgroups: gmane.emacs.buildstatus
> Date: Tue, 24 May 2022 23:01:08 -0400 (1 day, 11 hours, 21 minutes ago)

> Pipeline #18061 has failed!
> Project: emacs ( https://emba.gnu.org/emacs/emacs )
> Branch: master ( https://emba.gnu.org/emacs/emacs/-/commits/master )
> Commit: 4d60dcd8 ( https://emba.gnu.org/emacs/emacs/-/commit/4d60dcd87c72f7b96cac04233ffb253bd79f5bb1 )
> Commit Message: ; Fix inevitable failure in erc-dcc test

> * tes...
> Commit Author: F. Jason Park

> Pipeline #18061 ( https://emba.gnu.org/emacs/emacs/-/pipelines/18061 )
> triggered by EMBA bot ( https://emba.gnu.org/bot )
> had 1 failed build.

> Job #46114 ( https://emba.gnu.org/emacs/emacs/-/jobs/46114/raw )

> Stage: build-images
> Name: build-image-inotify

But I don't think this is what you're referring to because there were
other, similar build failures preceding it, such as

https://emba.gnu.org/emacs/emacs/-/pipelines/18058
https://emba.gnu.org/emacs/emacs/-/pipelines/18060

and ones following it that succeeded, like

https://emba.gnu.org/emacs/emacs/-/pipelines/18063

So the failure(s) you're talking about must have occurred elsewhere
between 18010 and 18061. I clicked through that interval in the web UI
and couldn't find anything. Perhaps someone deleted them? (If there's an
API token I could borrow to better investigate, I have tooling made for
this sort of thing.)

BTW, the job you linked to above, 46150, failed because I neglected to
quote the `make_params' variable properly in the generated YAML document
test/infra/test-jobs.yml. And it failed in the test stage, not the build
stage. In case anyone cares, I modified that variable in that throwaway
branch to ensure my expensive tests always run, something I inferred was
acceptable from my conversations with Michael Albinus on the tracker for
bug#48598. If that's not the case, please say so. Regardless, I was
alerted to that job failure shortly after it occurred and tried to
address it promptly, as was hopefully shown by the subsequent pipeline:

  https://emba.gnu.org/emacs/emacs/-/jobs/46152

Anyway, I'd like people to know that ERC (and I) take these tests
seriously. So, whatever it is, we'll get to the bottom of it.

Thanks,
J.P.

Re: master df1e553688 4/5: Accommodate nonstandard turbo file senders in erc-dcc

[Lars Ingebrigtsen]

F. Jason Park <jp@neverwas.me> writes:

> This links to a job on my branch for fix/bug-48598, so I'm guessing you
> meant to highlight something else.

I was probably just confused by the EMBA web page, as usual.

> There actually was a more recent failure on master (triggered by a push
> from me) that ended in a weird timeout or hang, namely:

EMBA is difficult to deal with, as it has a tendency to fail for
mysterious reasons not seen in the real world anywhere, so for it to be
useful at all, tests that time out has to be disabled on EMBA.
Currently all the jobs are red, which doesn't help anybody.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: master df1e553688 4/5: Accommodate nonstandard turbo file senders in erc-dcc

[Michael Albinus]

Lars Ingebrigtsen <larsi@gnus.org> writes:

>> There actually was a more recent failure on master (triggered by a push
>> from me) that ended in a weird timeout or hang, namely:
>
> EMBA is difficult to deal with, as it has a tendency to fail for
> mysterious reasons not seen in the real world anywhere, so for it to be
> useful at all, tests that time out has to be disabled on EMBA.
> Currently all the jobs are red, which doesn't help anybody.

Emba has been rebooted by the FSF sysadmins, shall be back to normal.

Best regards, Michael.

Re: master df1e553688 4/5: Accommodate nonstandard turbo file senders in erc-dcc

[Lars Ingebrigtsen]

Michael Albinus <michael.albinus@gmx.de> writes:

> Emba has been rebooted by the FSF sysadmins, shall be back to normal.

Thanks.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no