vcswatch reports that
this package seems to have a new changelog entry (version
6.1, distribution
UNRELEASED) and new commits
in its VCS. You should consider whether it's time to make
an upload.
Here are the relevant commit messages:
commit 5d182bdb8c55448de72d612fd114cc4549a5c6ab
Merge: f72de96 6e620a3
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 15:42:59 2026 +0200
Merge branch 'paride/doc-virt-server-quit' into 'master'
protocol spec: on 'quit', virt-servers do not reply 'ok'
See merge request ci-team/autopkgtest!693
commit f72de96fba535cc66fe0756589771c9a23f6710e
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 14:26:28 2026 +0200
a-v-ssh: drop maas ssh-setup script
The script is unmaintained, unused, and likely broken. I don't think it
got really used on Canonical infrastructure outside of experiments, and
there are no plans to do so.
commit 257a4891ad23e74f6b34f70e5d61009377b479ea
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 11:44:40 2026 +0200
Update d/changelog 6.1 release (UNRELEASED)
commit 6e620a371a6af29043a85f98038da043c3fa5033
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 11:16:56 2026 +0200
Handle virt-server exit using subprocess.communicate()
The new approach is slightly more robust (the previous one could cause
the virt-server to hit a SIGPIPE in case it tries to write to stdout,
given the self.sp.stdout.close() - not something we have seen in
practice).
In general, subprocess communicate() is more likely to do the right
thing in terms of file descriptor handling and wait for exit than us
calling the lower level close() and wait().
commit 3b7927fcd75b2bf025dcd5ec484dcaf0b3bb2882
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 11:13:30 2026 +0200
protocol spec: on 'quit', virt-servers do not reply 'ok'
This adapts the spec to how virt-server actually behave.
Closes: #1092808
Gbp-Dch: Full
commit e9c735cb523ae32e9ec86ae726100de2e660df53
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 11:28:08 2026 +0200
pre-commit: update pinned ruff hook version
Gbp-Dch: Ignore
commit 677abc47018fcec404e53cf101a318d01231585a
Merge: 374fc01 4fbafc5
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 21 10:37:24 2026 +0200
Merge branch 'paride/qemu-use-host-cpu' into 'master'
qemu: use host CPU passthrough for KVM
See merge request ci-team/autopkgtest!691
commit 374fc01dc095250dc6519204dc4c134ecc3d91d6
Merge: 28dca49 5bcb304
Author: Paride Legovini <paride@debian.org>
Date: Sun Sep 20 19:31:55 2026 +0200
Merge branch 'timeout-test-noprogress' into 'master'
Introduce timeout for lack of forward progress
See merge request ci-team/autopkgtest!456
commit 5bcb304aaab8510cc5a50d2df3c05ba89318b03e
Author: Christian Kastner <ckk@debian.org>
Date: Wed Dec 17 12:06:46 2025 +0100
Introduce timeout (25mins) for lack of forward progress
In cases where the global --timeout-test is necessarily large (hours),
it may be possible to detect a hung test and bail out earlier, rather
than wait for the global timeout to expire, thus freeing resources.
Some hung tests can be detected by a lack of forward progress, meaning
they stopped producing output.
This introduces a --timeout-test-noprogress for this case: if nothing
has been written to stdout or stderr for longer than this period,
VirtSubproc.Timeout is raised, which then triggers the already
existing test timeout functionality.
For this to work, we need to direct output to pipes, because (among
other things) Popen.communicate() waits for the process to exit, which
is precisely what we don't want.
The default of this timeout is set to 25mins.
commit 28dca49f209fe42103543da93e3a8444b13d0a9b
Merge: 5ef0cf0 7662e6d
Author: Paride Legovini <paride@debian.org>
Date: Sun Sep 20 16:33:27 2026 +0200
Merge branch 'paride/simplify-arch-matching' into 'master'
testdesc: improve architecture matching logic
See merge request ci-team/autopkgtest!690
commit 5ef0cf09d41d8d952f9c37ab399dec6590fcd7f5
Author: Paride Legovini <paride@debian.org>
Date: Sun Sep 20 14:50:40 2026 +0200
Salsa CI: suppress lintian debian-news-entry-has-unknown-version
Suppressing the tag allows having a d/NEWS entry for a version for
which the corresponding d/changelog stanza has not been created yet.
commit 7662e6d46e5debd30b8afd9a2af08c3b35b0dd58
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 18 18:23:44 2026 +0200
testdesc: handle unknown architectures with old python-debian
commit 5816bdf492421a18882c8ae5805c7e63d9e10d4f
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 18 14:51:30 2026 +0200
testdesc: fail on wrong restriction: Architecture: all
Technically `Architecture: all` is not wrong, but it will restrict the
test to run only on an architecture named `all` (literal), which in
practice means that the test would always be SKIPped.
Erroring out with InvalidControl() is a safeguard for users, so that
`all` is not specified by mistake instead of `any`.
commit 52df80a23e4be5b12a24b780e19086669028ad88
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 18 14:42:29 2026 +0200
testdesc: simplify architecture restriction matching
Delegate more of the architecture matching logic to the standard
architecture_is_concerned() from python-debian.
commit cfadb32061b12e6b547f38e436587d2edaed6a30
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 18 10:06:43 2026 +0200
testdesc: reject invalid architecture syntax
Closes: #1113955
commit fb590516bea0d44ff923e7c9a7e0d2676515637a
Merge: beccf34 91303f0
Author: Paride Legovini <paride@debian.org>
Date: Sun Sep 20 13:50:21 2026 +0200
Merge branch 'ruff' into 'master'
ruff: refactoring for increased readability
See merge request ci-team/autopkgtest!692
commit 91303f03b75d7d36146b3e0edcf7432fd29d0d45
Author: Benjamin Drung <benjamin.drung@canonical.com>
Date: Fri Sep 18 22:32:48 2026 +0200
ruff: refactoring for increased readability
Commit e3d427d0a7f469249d87077b79e19365c29e7909 ("ruff: fix remaining
violations") replaced most % formatting by f-strings.
Convert more code to use f-strings and refactor the code to make it more
readable.
commit 4fbafc58b8357a49be887d557a9332a158764506
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 18 17:03:31 2026 +0200
qemu: use host CPU passthrough for KVM
Since autopkgtest 3.20 (2016), x86_64 guests have used the baseline
kvm64 cpu model with the vmx and lahf_lm flags enabled. This kept the
virtual CPU reproducible across hosts while exposing the minimum
features then needed for nested KVM.
AMD was switched to host passthrough in 3.20.3 (2016), because kvm64
model did not work reliably there. QEMU now considers kvm64 a legacy
model [1], and current KVM documentation recommends host passthrough
[2]. Using kvm64 hides modern cpu capabilities and can make guest load
run very slowly.
Use host passthrough for native x86_64 KVM on both Intel and AMD.
The explicit user-provided -cpu option is still honored.
[1] https://www.qemu.org/docs/master/system/qemu-cpu-models.html#other-non-recommended-x86-cpus
"Legacy models just for historical compatibility with ancient QEMU versions."
[2] https://www.qemu.org/docs/master/system/i386/cpu.html#two-ways-to-configure-cpu-models-with-qemu-kvm
"This is the recommended CPU to use"
Closes: #1123760
commit beccf34b4ce24b15f11fc2b49d4db5779a90db27
Merge: e986b30 8135a54
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 18 14:42:55 2026 +0200
Merge branch 'paride/register-doc-base' into 'master'
doc: register HTML documentation with doc-base
See merge request ci-team/autopkgtest!689
commit 8135a544135aab049e1757edf685cc4232a9616b
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 23:52:15 2026 +0200
doc: register HTML documentation with doc-base
Closes: #1029414
commit 05399f49461ec80f412d575d58095316be9b3d8c
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 23:52:02 2026 +0200
doc: fail the build on rst warnings
commit ac4a39402f5c902209612e826606c940a6e3b9c7
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 23:51:56 2026 +0200
doc: fix sysvinit example formatting
commit e986b30839c75a3d08320c369baf65f319d29d00
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 22:21:56 2026 +0200
autopkgtest.1: clarify override control path
Closes: #1135606
commit 5aea6e56ccc4cd19e76304b75501c81828ce88b2
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 22:14:12 2026 +0200
Update name of schroot virt server across project
Closes: #1096110
commit 27b6afdd31398f46be7c53069da49e9c9a98b4ea
Merge: ca0d7c2 9c0edea
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 22:05:34 2026 +0200
Merge branch 'paride/virtsubproc-term-before-kill' into 'master'
VirtSubproc: terminate processes before killing them
See merge request ci-team/autopkgtest!688
commit 9c0edea17ecce8462be7d9222825586ed8eaf525
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 18:14:56 2026 +0200
VirtSubproc: terminate processes before killing them
In execute_timeout(), when the timeout is reached, send a SIGTERM
instead of a SIGKILL.
Sending a SIGKILL is problematic when the command uses sudo, as sudo
cannot catch the signal and forward it to its privileged child. This was
worked around in 63eb7e60c by adding an extra timeout layer, making sudo
invocations like:
check_exec(sudoify(argv, 300), outp=True, timeout=310)
Note the two timeouts: one for check_exec(), one for the inner
timeout(1). Other than being ugly, this layering caused #1011509, which
778f0aac6df fixed by running "timeout sudo" instead of "sudo timeout".
It also left a gap: 63eb7e60c only wrapped lxc-start and friends, so
sudoify() calls without a timeout still rely on the SIGKILL that sudo
cannot forward. That is #987046, hit via wait_booted().
A better solution is: on timeout, send a SIGTERM first so sudo can
forward it, wait up to 10 seconds, and only then fall back to SIGKILL.
This also makes the extra timeout(1) layer redundant, so drop it, and
also drop the sudoify() timeout parameter.
Closes: #987046
commit ca0d7c23f9df5159d76e7f10b1288a1e9adec3ea
Merge: 1808545 0ab80b7
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 17 16:20:10 2026 +0200
Merge branch 'paride/no-built-tree-special-treatment' into 'master'
Drop built-tree special-casing; rename tree kind to source-tree
See merge request ci-team/autopkgtest!687
commit 1808545eeb26529eed1997a4960480b7e81d1068
Merge: 1b1edf6 4beb1a4
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 15:42:14 2026 +0200
Merge branch 'paride/reset-is-revert' into 'master'
Consolidate testbed "reset" naming onto "revert"
See merge request ci-team/autopkgtest!686
commit 0ab80b7369601dd35918c44b0345763f30a1652a
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 14:15:39 2026 +0200
Drop built-tree special-casing; rename tree kind to source-tree
Previously a source tree directory was classified as either an
"unbuilt-tree" or a "built-tree" (the latter when debian/files was
present), with built trees being treated slightly differently, with one
user-visible difference: built trees got --no-built-binaries by default.
Most of the time this "feature" is a source of confusion.
Drop that distinction: every source tree passed to autopkgtest is now
treated uniformly. Since there is no longer a built-vs-unbuilt
distinction, rename the remaining "unbuilt-tree" kind to "source-tree"
for clarity.
commit 4beb1a4c4b41a2aa9bd0d520c6fb880be056e76c
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 13:50:41 2026 +0200
Consolidate testbed "reset" naming onto "revert"
Function names, docstrings and log messages used both "reset" and
"revert" for the operation of bringing the testbed back to a clean
state, which is confusing. Consolidate on "revert".
Historically the two terms encoded a distinction:
- "revert" was the low-level mechanism: the virt-server capability and
protocol command that rolls the testbed back to a pristine state.
- "reset" was a higher-level, "smart" autopkgtest concept. The old
Testbed.reset() tracked testbed state (self.modified, deps_installed,
needs_reset()) and only issued a revert when necessary. Moreover, an
apt-only path provided a manual apt-config cleanup as a fallback for
testbeds that lack the revert capability.
That distinction was essentially removed when the apt-only path got
moved to DebBinaries, and even more when when Testbed.reset() was
simplified to always revert if the capability is present (commit "Always
revert testbed before running individual tests"). At that point reset()
became a thin wrapper over revert, so keeping a separate word is no
longer useful and creates confusion.
commit 1b1edf69b4ccc0114f234ed5c9da2aeb24763681
Merge: da6833e ff214e8
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 13:52:50 2026 +0200
Merge branch 'paride/setup-testbed-always-deb822' into 'master'
setup-testbed: always write deb822 apt sources
See merge request ci-team/autopkgtest!685
commit da6833ea430fa3ef2606a5fd583c6eb1bd177b65
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 13:21:49 2026 +0200
pre-commit: update hooks
Gbp-Dch: Ignore
commit 4e71fbf060f496d8e3f64a7be59647d5295d7c90
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 13:16:59 2026 +0200
a-virt-lxd: bring back --vm to force a VM over a container
Image type auto-detection does not work on simplestreams remotes, where
a VM image can share its alias with a container image and thus cannot be
requested specifically. Restore --vm as an explicit override: when given
it skips type detection, adds the isolation-machine capability, and is
passed through to launch so the VM variant is selected.
Thanks: Alessandro Astone
commit ff214e885c76b9d3674fa40d85b7c979ab19e083
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 11:51:13 2026 +0200
setup-testbed: always write deb822 apt sources
The auto-detection branch used to preserve the style of the base
image's apt sources, writing either deb822 .sources or one-line-style
deb entries in sources.list. All supported testbeds now understand
deb822 sources, so drop the one-line-style output branches and always
write deb822.
Reading the base image's existing sources to auto-detect $RELEASE and
$MIRROR still handles both formats, as some base images continue to
ship a one-line sources.list.
commit 197df27c677a894f4c77e95a823dcd3a5393b168
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 11:30:51 2026 +0200
setup-testbed: drop code specific to upstart
Upstart is not used by any distribution supported by autopkgtest.
commit 721f993112d2f8627bd1a79748371185c7528062
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 10:26:33 2026 +0200
virt-unshare: drop no-op downtmp-host filter in hook_cleanup
A downtmp-host= entry is never added to capabilities for this backend.
commit 2dc9dc17e5db6980e9f9333a6dae1dee0ec5aff7
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 10:25:57 2026 +0200
virt-docker: drop dead hook_forked_inchild() no-op
Support for this hook was dropped project-wide long ago (see
debian/changelog: "Drop support for hook_forked_inchild() in
virt-runners. This has never been used.").
commit 83ca242242b50b3ea4ae902c0518b9415d19ac7a
Merge: 6a070d1 3336b2f
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 11:03:22 2026 +0200
Merge branch 'paride/no-more-trusty' into 'master'
Update baseline testbed requirements; drop legacy code
See merge request ci-team/autopkgtest!684
commit 3336b2f326bab1aff3f2d1478aeaf71ea49f8061
Author: Paride Legovini <paride@debian.org>
Date: Mon Sep 14 09:05:33 2026 +0200
tests: avoid apt segfault in test_persistent_apt_failure on old apt
Corrupt an apt.conf.d fragment rather than sources.list: a malformed
sources.list makes "apt-get build-dep" segfault on old apt (e.g.
bionic's 1.6), which is the code path used when apt lacks "satisfy".
A broken apt.conf produces the same permanent error across all apt
versions.
commit 5f73e5adf23c1b6a5ebef51e0ffff3d85dea90c9
Author: Paride Legovini <paride@debian.org>
Date: Sun Sep 13 19:34:02 2026 +0200
ci: use unconfined containers in the lxc jobs
Gbp-Dch: Ignore
commit 59ffacafc54fe836b5da1e7ad1ed52bbb252029c
Author: Paride Legovini <paride@debian.org>
Date: Sun Sep 13 18:48:06 2026 +0200
await-boot: also wait for network to be ready before returning
This tries be be defensive, returning early if configuration is broken,
or if there is no configuration to wait for.
commit 6a070d1e1ecce57dd2d0510532243cfbe6dd0a17
Author: Paride Legovini <paride@debian.org>
Date: Fri Sep 11 11:37:12 2026 +0200
a-b-qemu: fix image build as non-root with proxy
commit 569be2f7da5bfb0e31209bff920795570e1c3d80
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 23:49:30 2026 +0200
setup-testbed: drop the Ubuntu 12.04 grub fallback
Xenial and Stretch already support /etc/default/grub.d/*.cfg.
Create the directory if necessary and always use the drop-in file
instead of editing /etc/default/grub for older GRUB versions.
commit e5eced829136d9cc84276a2d22060e8368f207da
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 23:42:47 2026 +0200
build-lxc: drop the legacy lxcpath lookup
The supported LXC baseline already provides lxc-config lxc.lxcpath.
Remove the fallback to the old lxcpath spelling and the hardcoded
/var/lib/lxc default, letting a failed lookup report an error.
commit 82cce948a3d927e0155d351ba5deeb0a5e3621f4
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 23:42:46 2026 +0200
testdesc: use python-debian for architecture wildcard matching
Use DpkgArchTable.matches_architecture() instead of spawning Perl to
call Dpkg::Arch::debarch_is(). The required python-debian version
already provides this API.
This removes the last runtime use of libdpkg-perl. Drop the package
dependency and its explicit installation in the quicktests CI job.
commit 37141c65032554d4d2e3c5b9c17c0138ee177c7a
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 22:32:02 2026 +0200
testdesc: drop the fallback for missing NativeVersion
The versioned dependency on python3-debian ensures that NativeVersion
is available.
commit af3bc0bbdca1b37f7dc8e90265dc5e5459ddf4ef
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 22:31:49 2026 +0200
testdesc: always use python-debian architecture matching
With the versioned dependency on python3-debian >= 0.1.48 we can
drop the perl fallback.
commit 6c55d28bd1084a2c770aa96dd0ff30bbbdb04ffc
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 22:31:34 2026 +0200
virt-qemu: drop the pipes.quote fallback
The generated auxiliary command runs with the host's sys.executable.
All supported host Python versions provide shlex.quote, so the
ImportError fallback to pipes.quote is unreachable.
commit ffee96f49d830acb0ecae188d3f55a6360fa7e78
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 22:31:32 2026 +0200
tests: drop the unused ChrootRunner apt version query
commit 9e619ebc24f5e131bf9fe61e1e203cfaece766a2
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 22:31:32 2026 +0200
runner: stop generating unused fake source metadata
The fake .dsc for unbuilt source trees was only read by the legacy
build-dependency installation path. With that path removed, nothing
consumes the file.
commit e1c180eb4fc17c66d0825f7acf5bc56771e24904
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 22:08:48 2026 +0200
ci: use Bionic for the lxc-old-testbed job
Xenial's systemd 229 cannot boot on cgroup-v2-only hosts: it attempts to
mount the legacy systemd cgroup hierarchy and freezes.
Use Bionic instead. Its systemd supports cgroup v2, while apt 1.6 is
still older than 1.9.0 and exercises the .dsc fallback for missing
apt-get satisfy.
commit 9838bbf4418d36003ceb8014ee7235ee4685596c
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 19:08:56 2026 +0200
Update baseline testbed requirements (apt 1.2.35, python3.5.1)
Debian 8 (jessie) is out of ELTS support [0]. Ubuntu 14.04 LTS (trusty)
is still in legacy ESM support [1], but the small number of Trusty
packages with autopkgtests does not justify keeping Trusty support until
(at least) May 2029. Running tests on Trusty will still be possible
using an older version of autopkgtest.
[0] https://wiki.debian.org/LTS/Extended
[1] https://ubuntu.com/security/esm
commit 051b3f2d3d5c513b4e7aec964603cc50bada2433
Merge: 517b513 5adb0a9
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 20:33:38 2026 +0200
Merge branch 'paride/fix-mypy' into 'master'
Fix mypy and make its failures fatal
See merge request ci-team/autopkgtest!683
commit 5adb0a967039a7cc1954a55dc4a06121fc40fce4
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 18:56:59 2026 +0200
tests: make mypy failures fatal
commit 4a633d455d20e86376c4a9371d25e00c54838fd0
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 18:48:46 2026 +0200
mypy: fix all the issues currently found by tests/mypy
commit 517b5138f0844b5c3f0da19d1855e4b5402c22db
Merge: 3776362 e3d427d
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 18:43:48 2026 +0200
Merge branch 'paride/new-ruff' into 'master'
Comply to newer ruff
See merge request ci-team/autopkgtest!682
commit e3d427d0a7f469249d87077b79e19365c29e7909
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 15:26:06 2026 +0200
ruff: fix remaining violations
Gbp-Dch: Ignore
commit f5346a7bf817b8051abb6d9f94b5ed4f106ac553
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 14:57:37 2026 +0200
ruff: apply automatically generated fixes
Gbp-Dch: Ignore
commit c9b58908959529abe3b0cc212e9c3a20a415744f
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 14:40:27 2026 +0200
ruff: rely on the default rule set
Ruff 0.16 expanded its default rule set. Let's stick with the defaults
instead of manually enabling all the pycodestyle W rules.
Gbp-Dch: Ignore
commit 262050232d9324f6f64e861511ff1e3ea28f0eef
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 12:23:00 2026 +0200
pyproject: document requires-python consumers
Gbp-Dch: Ignore
commit 5591490da71495bf41e1b9224221abbb3de51850
Author: Paride Legovini <paride@debian.org>
Date: Wed Sep 9 23:18:10 2026 +0200
pre-commit: upgrade to ruff v0.16.6
commit 37763620716c7525d8c888211fb4f705fd4710a2
Merge: 28eadf6 0822d9e
Author: Paride Legovini <paride@debian.org>
Date: Thu Sep 10 10:49:21 2026 +0200
Merge branch 'paride/normalize-unbuilt-tree-path' into 'master'
runner: normalize unbuilt-tree path
See merge request ci-team/autopkgtest!681
commit 0822d9e311e9d90ba60451a0ffc8233ff3d2b72c
Author: Paride Legovini <paride@debian.org>
Date: Wed Sep 9 18:53:26 2026 +0200
runner: normalize unbuilt-tree path
Without this, passing a path to a directory with a trailing slash
(/foo/bar/) would result in empty basename, and thus in the `ubtree-`
directory. Normalizing it fixes this.
commit 28eadf6a95e8f929b78551c6118867bad4fd8b6d
Author: Paride Legovini <paride@debian.org>
Date: Wed Sep 9 23:15:21 2026 +0200
pre-commit: update codespell to v2.4.3
Gbp-Dch: Ignore
commit b2055fb9a6992d4a10bb9cf116856f300d4d2b74
Merge: 9b2dc26 cca15ea
Author: Paride Legovini <paride@debian.org>
Date: Wed Jul 29 20:13:40 2026 +0200
Merge branch 'paride/qemu-timeouts' into 'master'
a-v-qemu: bump reboot timeout from 60s to 900s
See merge request ci-team/autopkgtest!677
commit 9b2dc26dceb67327d6005b3e305f4a1b7b945f48
Merge: 0841e27 8358da5
Author: Paride Legovini <paride@debian.org>
Date: Wed Jul 29 20:11:31 2026 +0200
Merge branch 'paride/lxd-revert-force-noninteractive' into 'master'
lxd auxverb: do not execute commands with --force-noninteractive
See merge request ci-team/autopkgtest!678
commit cca15eafb55c1642cc238ec496f75ff1021cb808
Author: Paride Legovini <paride@debian.org>
Date: Mon Jul 27 15:36:01 2026 +0200
a-v-qemu: warn about slowness when /dev/kvm is not available
Helps: #1077620
commit 3d7d60e342b0e09ba94d1f25ccecae02349a0940
Author: Paride Legovini <paride@debian.org>
Date: Mon Jul 27 15:34:22 2026 +0200
a-v-qemu: bump reboot timeout from 60s to 900s
60s is not enough time for emulated testbeds to complete a reboot.
Closes: #1077620
commit 8358da5b272288bc605658d8c8a66e7b3373d127
Author: Paride Legovini <paride@debian.org>
Date: Wed Jul 29 17:02:27 2026 +0200
lxd auxverb: do not execute commands with --force-noninteractive
Do not execute auxverb commands using --force-noninteractive as that
breaks --shell/--shell-fail.
The idea of using --force-noninteractive came from [1], with the problem
described there also being reproducible using Incus. I do not believe
--force-noninteractive made a big difference, while --disable-stdin did.
[1] https://github.com/canonical/lxd/issues/17797#issuecomment-4459954015
This reverts commit 509db28362e7d4e8dd64e4ff1842ad82bd04ebea.
Closes: #1142949
Thanks: Colin Watson
https://salsa.debian.org/api/v4/projects/ci-team%2Fautopkgtest API request failed: 401 Unauthorized at /srv/qa.debian.org/data/vcswatch/vcswatch line 410.
![[bug history graph]](https://qa.debian.org/data/bts/graphs/a/autopkgtest.png){kind=link}