Claude
0cd7a970f2
The runner abandoned tasks Gitea had already assigned when FetchTask took longer than the default 5s fetch_timeout, leaving runs in_progress with no logs until Gitea's zombie sweep failed them (~15 min). Raise fetch_timeout to 30s, cap jobs at 1h, and keep the Fly config + runner config here so the runner can be rebuilt from the repo. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VVFi3wvLWKaEkpDw6eLnQC
21 lines
818 B
YAML
21 lines
818 B
YAML
# gitea-runner (act_runner 5.0.0) config for Fly app green-tea-runner.
|
|
# Mounted at /etc/gitea-runner/config.yaml via fly.toml [[files]]; CONFIG_FILE points at it.
|
|
log:
|
|
level: info
|
|
runner:
|
|
file: /data/.runner # registration state lives on the volume
|
|
capacity: 1
|
|
# Root cause of "job taken, container never starts" (Oct 25 2026): FetchTask to teahub.xyz
|
|
# sometimes takes >5s; the default 5s fetch_timeout made the runner abandon a task the
|
|
# server had already assigned, leaving it in_progress with no logs until Gitea's zombie
|
|
# sweep failed it (~15 min). Give the call room.
|
|
fetch_timeout: 30s
|
|
fetch_interval: 2s
|
|
fetch_interval_max: 10s
|
|
# Fail a hung job in 1h instead of the 3h default.
|
|
timeout: 1h
|
|
shutdown_timeout: 60s
|
|
report_close_timeout: 10s
|
|
container:
|
|
force_pull: false
|