Registry status
Is the registry up, and is the log moving. Uptime comes from an outside workflow that probes the registry every 5 minutes and appends one line per run to a public history; the job ages come from the registry's own job table. Every number here is computed from those two sources.
Uptime is computed from the public history file, one line per uptime workflow run (every 5 minutes). A run is up when no check failed; skipped checks count toward neither side.
Last daily root 14 h ago
The nightly pings the dead-man's switch only when every step succeeded, so a missed ping pages the operator even when the host is down. A failing step posts one alert per step per UTC day to the pager webhook.
| JOB | LAST SUCCESS | AGE | STATE |
|---|---|---|---|
| checkpoint | 2026-11-10T06:00:00.020Z | 14 h | OK |
| dump | 2026-11-10T06:00:00.020Z | 14 h | OK |
| git | 2026-11-10T06:00:00.020Z | 14 h | OK |
| nightly | 2026-11-10T06:00:00.020Z | 14 h | OK |
| ots | 2026-11-10T06:00:00.020Z | 14 h | OK |
| ots-upgrade | 2026-11-10T19:41:00.042Z | 57 min | OK |
| rekor-v2 | 2026-11-10T20:23:00.018Z | 15 min | OK |
| root | 2026-11-10T06:00:00.020Z | 14 h | OK |
| sequencer | 2026-11-10T16:37:52.719Z | 4 h | OK |
| timestamps | 2026-11-10T06:00:00.020Z | 14 h | OK |
| tsa-witness | 2026-11-10T20:23:00.025Z | 15 min | OK |
| webhooks | 2026-11-10T20:35:00.016Z | 3 min | OK |
| witness-lists | 2026-11-10T20:23:00.021Z | 15 min | OK |
| REGION | ROLE | STATE | LAST SEEN | LAST FAILOVER DRILL |
|---|---|---|---|---|
| us-east4-eqdc4a | leader: runs the sequencer and the jobs, serves reads | HEALTHY | 2026-11-10T20:38:10.116Z | none |
Only the replica holding the leader lease runs the sequencer and the scheduled jobs, so the tree never forks. Every replica reads the same Postgres index and bucket.
No failover drill recorded yet.
A drill stops the leader's region on purpose. "No leader for" is the time from the old leader's last lease renewal to the standby's takeover, as the takeover recorded it. Each drill is also on the admin ledger.
No uptime history yet.
None in the history read.
An incident is a run of consecutive uptime runs with a failing check. It is open while the newest run still fails.