The queued-job test passed in a disposable, network-isolated amd64 VM running Debian systemd 262.2-1 as PID 1. I verified that the matching Debian source returns the setter result for JobTimeoutUSec and that the running manager executable and mapped systemd core library matched the authenticated package binaries.
I submitted a transient target service with JobTimeoutUSec set to uint64 20000000. The target had Requires= and After= on a Type=oneshot service running /usr/bin/sleep 35. While the dependency was activating, the manager reported target job 53 as waiting with the full 20,000,000 µs timeout configured. JobRemoved for that exact target job reported “timeout”. Using the guest monotonic clock, the signal arrived 20.135 seconds after sending StartTransientUnit and 20.028 seconds after its acknowledgement. Those readings bracket enqueue-to-observation time and include signal delivery. At that point the dependency was still activating, with the same sleep process (PID 140 and matching process start time). The target’s ExecStart marker was absent. The dependency’s own job, 56, subsequently reported “done” 36.471 seconds after submission. The target marker was still absent. After cleanup there were no fixture jobs, cgroups, or surviving sleeper. A preliminary run ended before expiry due to a cgroup observation error. The timings above are from the corrected run in a fresh guest. This demonstrates that the whole-job timeout includes time waiting for an ordered dependency. Verification of an Ubuntu Resolute backport remains pending. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2174933 Title: [Resolute] StartTransientUnit rejects JobTimeoutUSec on amd64 (v259.10 backport requested) To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/2174933/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
