Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > linux.debian.bugs.rc > #390011 > unrolled thread

Bug#1103406: ogdi-dfsg: flaky autopkgtest on arm64

Started byPaul Gevers <elbrus@debian.org>
First post2025-04-17 12:20 +0200
Last post2025-04-17 12:20 +0200
Articles 1 — 1 participant

Back to article view | Back to linux.debian.bugs.rc

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.


Contents

  Bug#1103406: ogdi-dfsg: flaky autopkgtest on arm64 Paul Gevers <elbrus@debian.org> - 2025-04-17 12:20 +0200

#390011 — Bug#1103406: ogdi-dfsg: flaky autopkgtest on arm64

FromPaul Gevers <elbrus@debian.org>
Date2025-04-17 12:20 +0200
SubjectBug#1103406: ogdi-dfsg: flaky autopkgtest on arm64
Message-ID<KCuvD-e9DC-5@gated-at.bofh.it>

[Multipart message — attachments visible in raw view] — view raw

Hi Bas,

On 17-04-2025 10:53, Sebastiaan Couwenberg wrote:
> Runner with unreliable network should not satisfy the needs-internet 
> requirement.


I agree with you on this, the problem is of course how to define 
unreliable network and where the border should be.

> I've made the wget and tar commands verbose, but that doesn't solve the 
> problem.


Thanks. It will help to *determine* if workers are really unreliable in 
networking. For me this and this package, it's currently just a 
hypothesis. Also, I'm not seeing this a lot on arm64, last time I ran 
into unreliable network issues in tests was at least several months ago. 
An additional option is to mark a test as skippable and exit 77 if the 
downloading (after retries) fails.

> Removing the autopkgtest or excluding unreliable architectures will be 
> needed instead. I don't really want to do either, but nuking the 
> autopkgtest from the package resolves this class of issues forever.


Let's work together to improve the situation. I understand this is 
frustrating from your side, but autopkgtesting has been a great 
improvement for Debian release management so it would be sad if for this 
reason you would remove them.

To avoid having to do this on the package side, I (ci.d.n maintainer hat 
on) could add your package to the ci.d.n reject_list until we have a way 
to disable needs-internet testing per host and can direct tests that 
need it to hosts which support needs-internet [1]. We currently can only 
switch needs-internet per architecture and I consider that a too big 
hammer for arm64 (release team member hat on).

Paul

[1] https://salsa.debian.org/ci-team/debci/-/issues/166

[toc] | [standalone]


Back to top | Article view | linux.debian.bugs.rc


csiph-web