Problem
The current scripts/ci_host_maintenance.sh installs a daily systemd timer:
Description=Daily Verity CI host maintenance
OnCalendar=*-*-* 04:30:00
However, the operator-facing text still describes the timer as weekly:
scripts/ci_host_maintenance.sh help: Install and enable a weekly systemd timer
scripts/README.md: Install weekly cache, journald, and Docker image cleanup.
docs/CI_RUNNERS.md: Each CI host should have the weekly Verity maintenance timer installed.
Reproduction
- Run
scripts/ci_host_maintenance.sh install-systemd.
- Inspect the generated unit with
systemctl cat verity-ci-host-maintenance.timer.
- It schedules at
*-*-* 04:30:00 (daily), while the help and operator docs say weekly.
Why this matters
This is an operationally relevant documentation regression introduced when the maintenance cadence changed after the CI disk-exhaustion incident. An operator following the docs can form the wrong expectation about cache cleanup cadence and disk growth.
Proposed narrow fix
- Change the three stale descriptions to
daily.
- Extend
scripts/test_ci_infra_maintenance.py to assert the help/docs wording remains aligned with the generated timer.
- Do not change the timer or any cleanup behavior.
I'd be happy to prepare this as a small documentation/test PR if this scope matches the project's preference.
Problem
The current
scripts/ci_host_maintenance.shinstalls a daily systemd timer:However, the operator-facing text still describes the timer as weekly:
scripts/ci_host_maintenance.shhelp:Install and enable a weekly systemd timerscripts/README.md:Install weekly cache, journald, and Docker image cleanup.docs/CI_RUNNERS.md:Each CI host should have the weekly Verity maintenance timer installed.Reproduction
scripts/ci_host_maintenance.sh install-systemd.systemctl cat verity-ci-host-maintenance.timer.*-*-* 04:30:00(daily), while the help and operator docs say weekly.Why this matters
This is an operationally relevant documentation regression introduced when the maintenance cadence changed after the CI disk-exhaustion incident. An operator following the docs can form the wrong expectation about cache cleanup cadence and disk growth.
Proposed narrow fix
daily.scripts/test_ci_infra_maintenance.pyto assert the help/docs wording remains aligned with the generated timer.I'd be happy to prepare this as a small documentation/test PR if this scope matches the project's preference.