How do I hear when a job fails?
Each [[notify.targets]] entry has a name, a kind (ntfy, discord or webhook), a url, and optional headers and events:
[[notify.targets]]name = "phone"kind = "ntfy"url = "https://ntfy.sh/halyard-ci-6f3a9c1e" # anyone who knows the topic can read itReload or restart runnerd, then runnerctl notify test sends every target a test message, or only --target <name>. On a self-hosted ntfy server with access control, add headers = { Authorization = "Bearer tk_…" }. A Discord url is the channel’s webhook; each message is one embed that pings no one. A webhook target posts JSON (api::Notification in crates/core/src/api/insights.rs); Discord rejects it.
| Event | Sent when | Default |
|---|---|---|
failed |
a job failed or timed out | on |
lost |
a job’s runner vanished before the job completed | on |
runner |
a runner could not start | on |
insights |
a new warning insight, told again only after runnerd saw it gone for a day; host-held once in each 6-hour window, host-low once a day |
on |
digest |
daily at digest_at (default 09:00): the day’s jobs, new insights, longest waits, the host’s memory while CI ran |
off |
A burst is one message: runnerd waits 10 seconds after the first event and sends to a target at most once a minute, so twenty failures arrive as one. A failed delivery is retried with backoff, 5 seconds to 10 minutes.
A target’s URL and header values are secrets: logs, errors and runnerctl notify test never show them, and the config editor, runnerctl config show and GET /v1/config show "<secret>". Change them in runnerd.toml itself.