Bug#1035983: libsoup3 (and libsoup2) autopkgtests are flaky: Address already in use: AH00072: make_sock: could not bind to address 127.0.0.1:47524

[Santiago Vila]

El 24/1/25 a las 12:40, Luca Boccassi escribió:
> Same experience with src:systemd uploads.
> 
> RT says that flaky autopkgtest are RC, so raising severity accordingly.
> Restrictions: flaky could also be added if the tests can't be fixed, as
> a quick bandaid.

Hi. Thanks a lot for raising the severity of this bug.
  
This bug was hidden for a long time because the package
had also a missing BD on ca-certificates (#1054962) which
took one year to be fixed.

After the BD was fixed, this package FTBFS 20% of the time
in my environment (in the way described by this bug), which imo is
way worse than our users deserve.

So, I wonder if we can apply the same principle that
flaky tests are RC to gcr and gcr4:

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1069402
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1057562

In the case of gcr4, there is an explicit request from
one of the release managers (Sebastian Ramacher) to
"either disable the flaky tests or make them non-flaky".

In fact, gcr4 fails to build in Salsa CI:

https://salsa.debian.org/gnome-team/gcr4/-/pipelines

and it also fails to build for me 99% of the time, so
imo it's even worse than libsoup3. I would say that if that's
not RC, it should be.

Not raising the severity myself, but I still expect
somebody else (maybe a RM) to do so.

Thanks.