diff options
Diffstat (limited to 'deployment')
| -rw-r--r-- | deployment/doot-health.service | 18 | ||||
| -rw-r--r-- | deployment/doot-health.timer | 14 |
2 files changed, 32 insertions, 0 deletions
diff --git a/deployment/doot-health.service b/deployment/doot-health.service new file mode 100644 index 0000000..33e8e77 --- /dev/null +++ b/deployment/doot-health.service @@ -0,0 +1,18 @@ +[Unit] +Description=doot.terst.org health check (gateway routes + sub-services) +Documentation=file:///site/doot.terst.org/app-scripts/health-check +After=network-online.target + +[Service] +Type=oneshot +# Runs as root: recovery needs systemctl start and chmod on www-data files. +User=root +ExecStart=/site/doot.terst.org/app-scripts/health-check + +# A failed run should be visible in `systemctl --failed`, so do not swallow the +# exit code. The unit going failed is a large part of the alerting here, given +# there is no mail/ntfy on this host. +SuccessExitStatus=0 + +[Install] +WantedBy=multi-user.target diff --git a/deployment/doot-health.timer b/deployment/doot-health.timer new file mode 100644 index 0000000..22b687f --- /dev/null +++ b/deployment/doot-health.timer @@ -0,0 +1,14 @@ +[Unit] +Description=Run the doot.terst.org health check every 5 minutes + +[Timer] +# 2 minutes after boot: the failure this exists for first appeared at a reboot, +# when sub-services could not exec and systemd gave up on them. +OnBootSec=2min +OnUnitActiveSec=5min +# Spread the load off exact-minute boundaries. +RandomizedDelaySec=30s +Unit=doot-health.service + +[Install] +WantedBy=timers.target |