Control: tags -1 wontfix Control: close -1 On Wed, 07 Aug 2024 13:11:42 +0000 Bastien =?ISO-8859-1?Q?Roucari=E8s?= <[email protected]> wrote: > Package: systemd > Version: 247.3-7+deb11u5 > Severity: important > Tags: patch upstream jessie stretch buster bullseye > Forwarded: https://github.com/systemd/systemd/commit/b2c7d1bbc2 > > Dear Maintainer, > > Without this commit autopkgtest on salsa are broken. > > See for instance > https://salsa.debian.org/apache-team/apache2/-/jobs/5960590 > > Can you consider to release a PU release this patch ? > > I can do the work. > > It breaks your testing infrastructure, particularly for testing daemon, particularly security update testing.
As mentioned on IRC, I am not very comfortable with shipping code changes in oldstable at this point in time for a specific corner case that only happens due to a particular setup in a particular container environment. Also as mentioned, this is only a problem for units that set PrivateTmp=yes, so what the autopkgtest branch for that affected package can do is add a drop-in in /run/systemd/system/foo.service.d/disabletmp.conf with: [Service] PrivateTmp=yes and do a daemon-reload at the beginning of the test case, and the problem should be bypassed. If the unit really needs an individual tmpdir, you can also add: TemporaryFileSystem=/tmp Which is very similar but doesn't add a dependency on the host's tmp.mount Hence I am closing for now. If it turns out that there are hundreds of packages affected that need to run tests on bullseye in salsa-ci and are also using PrivateTmp and it's impractical to add workarounds everywhere we can reconsider, but if it's just a handful and the workaround works, please use that instead, so that we can minimize risk. -- Kind regards, Luca Boccassi
signature.asc
Description: This is a digitally signed message part
