>From autopkgtest (as in "the thing that prepares a testbbed and runs the
test in it") this is correct -- the test fails, and there is no robust
way to automatically tell whether this is to be blamed on the package
itself or a forward or reverse dependency. Also, autopkgtest is just a
command, it cannot "retry" itself. So moving to the infrastructure
project.
Just blindly retrying every 24 hours is arbitrary and pointless. E. g.
in this case repowerd is just broken and needs to be fixed, there is no
use in retrying it on a time basis. It *would* make sense to retry it
once repowerd gets fixed, but since this is neither the trigger nor the
test here, this would require complex state keeping of transitive test
dependencies, heuristics on parsing build logs (which are hard to
define, easy to get wrong, never complete, and just raise expectancies
that we can ever fulfill). I wontfix this, as creating and maintaining
this is magnitudes more work than just retrying tests after a package
got fixed. Note that bug 1491145 is about an important subset of this
problem which is reasonably easy to fix, so I will do that.
So, the real problem is that repowerd fails -- this needs to be fixed in
either case, will affect your project (thumbnailer) on actual
installations as well, and isn't an actual problem with the test or the
test infra.
** Package changed: autopkgtest (Ubuntu) => auto-package-testing
** Changed in: auto-package-testing
Status: New => Won't Fix
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1603373
Title:
Retry failures due to broken dependencies automatically
To manage notifications about this bug go to:
https://bugs.launchpad.net/auto-package-testing/+bug/1603373/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
