# Hourly reclaim of Docker resources act_runner LEAKS (per-job volumes, stale networks, old build # cache). Sibling of docker-prune.service, which handles what CI legitimately produces and then # abandons; the split matters because this one must stay conservative enough to run while jobs are # live (dangling-only volumes, age-gated networks) — see docker-reclaim.sh for the full why. # # Install: see the header of docker-reclaim.sh (note the installed unit name is # ci-docker-reclaim.service — existing fleet hosts already run it under that name). [Unit] Description=Reclaim disk leaked by Gitea act_runner (per-job volumes, networks, stale build cache) Documentation=https://git.unom.io/unom/punktfunk After=docker.service Requires=docker.service [Service] Type=oneshot ExecStart=/usr/local/sbin/ci-docker-reclaim.sh # Never let maintenance starve a running build. Nice=10 IOSchedulingClass=idle