Package: release.debian.org
Severity: normal
Tags: bookworm
User: release.debian....@packages.debian.org
Usertags: pu
X-Debbugs-Cc: netplan...@packages.debian.org, sl...@ubuntu.com
Control: affects -1 + src:netplan.io
[ Reason ]
Starting with systemd v252.15 in proposed-updates for bookworm
Netplan's autopkgtests started to tmpfail and kept DebCI looping.
elbrus created an RC bug about it:
- https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1053496
This is related to a change in systemd-stable's "wait-online" behavior:
-
https://github.com/systemd/systemd-stable/commit/f4831559171033ab044758e7fd68627e3bfe84a5
We've seen the same issue with Netplan's autopkgtests in unstable
when systemd v253 hit experimental. Here's some more context of the
change and the old bugreport:
-
https://github.com/systemd/systemd-stable/commit/0531c47e96d13ea62e4b54bc16c96a573513caf9
- https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1033265
The same fix as in Debian#1033265 can be applied to fix the timeout issues.
[ Impact ]
Users should not be directly affected by this Netplan patch. They
might already be affected by systemd's change in behavior, though,
which is independent of Netplan.
This patch impacts Netplan's autopkgtests only. So might be of
interest to the cloud team, who is utilizing Netplan in their
bookworm images. Having working Netplan autopkgtests would be
good to check reverse-depends for cloud-images and others.
[ Tests ]
Running the 'autostart' and 'cloud-init' tests locally, I was able
to reproduce the testbed timeout. After applying the debdiff, all
autopkgtests can be run without hitting any timeouts.
Failure case:
$ sudo autopkgtest netplan.io --test-name=autostart -- lxc
autopkgtest-bookworm-amd64
[...]
autopkgtest [11:54:51]: test process requested reboot with marker config
Unexpected error:
Traceback (most recent call last):
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 829, in mainloop
command()
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 758, in command
r = f(c, ce)
^^^^^^^^
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 453, in cmd_reboot
caller.hook_wait_reboot(**wait_reboot_args)
File "/usr/bin/autopkgtest-virt-lxc", line 311, in hook_wait_reboot
wait_booted(lxc_container_name)
File "/usr/bin/autopkgtest-virt-lxc", line 117, in wait_booted
VirtSubproc.wait_booted(sudoify(['lxc-attach', '--name', lxc_name, '--']),
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 247, in wait_booted
check_exec(
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 192, in check_exec
(status, out, err) = execute_timeout(None, timeout, real_argv,
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 154, in execute_timeout
(out, err) = sp.communicate(instr)
^^^^^^^^^^^^^^^^^^^^^
File "/usr/lib/python3.11/subprocess.py", line 1197, in communicate
stderr = self.stderr.read()
^^^^^^^^^^^^^^^^^^
File "/usr/share/autopkgtest/lib/VirtSubproc.py", line 74, in alarm_handler
raise Timeout(to)
VirtSubproc.Timeout: 60
autopkgtest [11:56:04]: ERROR: testbed failure: unexpected eof from the testbed
[ Risks ]
The changes are trivial and only affect Netplan's autopkgtests.
Specifically debian/tests/autostart and debian/tests/cloud-init
Furthermore, we're adding the "KeepConfiguration=yes" setting on
the management interface's systemd-networkd configuration, to make
sure the container stays reachable over the network.
[ Checklist ]
[x] *all* changes are documented in the d/changelog
[x] I reviewed all changes and I approve them
[x] attach debdiff against the package in (old)stable
[x] the issue is verified as fixed in unstable
[ Changes ]
We're changing the test-interface priority to "optional", to avoid
systemd's "wait-online" blocking on it during reboot.
[ Other info ]
The same fix has been applied in unstable/testing (Bug #1033265).
diff -Nru netplan.io-0.106/debian/changelog netplan.io-0.106/debian/changelog
--- netplan.io-0.106/debian/changelog 2023-03-02 17:40:56.000000000 +0100
+++ netplan.io-0.106/debian/changelog 2023-10-24 12:48:28.000000000 +0200
@@ -1,3 +1,9 @@
+netplan.io (0.106-2+deb12u1) bookworm; urgency=medium
+
+ * debian/tests: Fix testbed timeout with systemd v252.15 (Closes: #1053496)
+
+ -- Lukas Märdian <sl...@debian.org> Tue, 24 Oct 2023 12:48:28 +0200
+
netplan.io (0.106-2) unstable; urgency=medium
* Fix DBus .Config/.Get APIs using upstream commits (PR#331) (LP: #1997467)
diff -Nru netplan.io-0.106/debian/tests/autostart.sh
netplan.io-0.106/debian/tests/autostart.sh
--- netplan.io-0.106/debian/tests/autostart.sh 2023-02-09 09:49:42.000000000
+0100
+++ netplan.io-0.106/debian/tests/autostart.sh 2023-10-24 11:48:04.000000000
+0200
@@ -35,6 +35,7 @@
[Network]
DHCP=yes
+KeepConfiguration=yes
EOF
case "${AUTOPKGTEST_REBOOT_MARK:-}" in
@@ -58,6 +59,7 @@
version: 2
bridges:
brtest00:
+ optional: true # ignore in systemd-networkd-wait-online.service to avoid
testbed timeouts
addresses: [10.42.1.1/24]
EOF
diff -Nru netplan.io-0.106/debian/tests/cloud-init.sh
netplan.io-0.106/debian/tests/cloud-init.sh
--- netplan.io-0.106/debian/tests/cloud-init.sh 2023-02-09 09:49:42.000000000
+0100
+++ netplan.io-0.106/debian/tests/cloud-init.sh 2023-10-24 11:48:04.000000000
+0200
@@ -33,6 +33,7 @@
[Network]
DHCP=yes
+KeepConfiguration=yes
EOF
case "${AUTOPKGTEST_REBOOT_MARK:-}" in
@@ -45,6 +46,7 @@
version: 2
bridges:
brtest00:
+ optional: true # ignore in systemd-networkd-wait-online.service to avoid
testbed timeouts
addresses: [10.42.1.1/24]
EOF
# Prepare a dummy netplan service unit, which will be moved to
/run/systemd/system/
