#1144803 lxc: autopkgtest fails under incus-lxc: several test failures

Package:
src:lxc
Source:
src:lxc
Submitter:
Antonio Terceiro
Date:
2026-10-01 19:53:05 UTC
Severity:
normal
Tags:
#1144803#5
Date:
2026-08-18 23:49:48 UTC
From:
To:
Hi,

Debian CI is switching away from lxc containers in favor of incus Containers.
This is motivated by security concerns from us; incus is based on lxc, but
orchestrates containers substantially different: containers are not privileged
(so root in the container is not uid 0 outside of it, and incus imposes a
stricter isolation from the host system.

lxc passes its tests under lxc, but fails under incus. This is a bit ironic. :-)

There are several test failures:
https://ci.debian.net/experiments/8/regressions/
(please beware of pagination)

Common types of failure and suggested fixes are being documented at:
https://wiki.debian.org/ContinuousIntegration/LxcToIncus

Note that for the time being, lxc is still being tested under lxc to
avoid disturbing its testing migration test results.

If you decide to add the `isolation-machine` restriction to get this package
tested under qemu, please mention that explicitly when closing this bug (it's
fine to do that only in the package changelog entry that closes the bug) so
that we can configure your package for qemu on ci.debian.net.

#1144803#10
Date:
2026-09-26 17:45:40 UTC
From:
To:
control: tags -1 + pending

  Yeah, that's funny, but caused by exactly what makes Incus a better
CI environment -- unprivileged containers by default. :) I've added a
commit[0] for lxc's packaging that's similar to what Leah (CC'ed) has
proposed coming from Ubuntu's packaging of lxc.

  I think there will be a new lxc release (7.0.1) in the near future,
so I'll plan to include this change in the next release of Debian's
packaging of lxc.

  Running the autopkgtests within a VM will give full test coverage, so
in an ideal world that would be preferable to running a subset of the
tests in containers. If I remember correctly, in the past QEMU runners
in the CI infrastructure were limited to just amd64 (and maybe arm64?),
is that still true? Or are QEMU runners available for all architectures
that run CI tests? If not, I'd choose to keep running lxc's subset of
tests within containers on all CI architectures versus running all of
lxc's tests on just amd64 and/or arm64 within VMs.

Mathias

[0] -- https://salsa.debian.org/lxc-team/lxc/-/commit/68f9f3d4a36afc954cb96002db372f8d637c1579

#1144803#17
Date:
2026-10-01 19:52:20 UTC
From:
To:
Thanks.

We have qemu only for amd64. It will fallback to whatever is
available on the other architectures, so any tests where you declare
`Restrictions: isolation-machine` will be skipped on non-qemu, while the
ones that don't will still run. You can also not declare anything, and
change the tests themselves to detect whether they are being run in a VM
or not and have them do the right thing.

Should I switch lxc to run under qemu?