cronix history

history reads run records for one (app, job) pair from the backend’s native source — journalctl for systemd-timer, Pod logs for kubernetes — and prints one row per terminal run. The trigger shim emits one slog-JSON record per attempt; History folds those into one entry per terminal outcome (success, app-rejected, retries-exhausted, lock-contended, timeout).

The crontab backend currently returns nil pending a syslog reader; use journalctl _SYSTEMD_UNIT=cron.service or your distro’s equivalent until that lands. The kubernetes and systemd-timer backends return real data.

Synopsis

cronix history <app>.<job> [flags]

Flags

Flag	Default	Purpose
`--since`	(none)	Look back this duration. Accepts Go duration syntax: `1h`, `24h`, `168h`
`--until`	(now)	Stop at this duration ago — also a Go duration. Useful for windowed queries: `--since 48h --until 24h`
`--status`	(any)	Filter to one status: `ok`, `failed`, `lock-contended`, `timeout`, `unknown`
`--limit`	`50`	Max records to show. `0` means no limit
`-o, --output`	`table`	`table` or `json`

Plus all backend flags.

Examples

Last 24 hours of runs from a systemd-timer-backed job:

cronix history billing.reconcile-payments \
  --backend systemd-timer --since 24h
# WHEN                  RUN_ID        ATTEMPT  STATUS  SOURCE
# 2026-05-05T10:15:00Z  abc123def456  1        ok      journal
# 2026-05-05T10:00:00Z  abc123def012  1        ok      journal
# 2026-05-05T09:45:00Z  abc123def789  2        ok      journal

Failed runs only, in JSON, from a Kubernetes-backed job:

cronix history billing.reconcile-payments \
  --backend kubernetes --in-cluster --k8s-namespace billing \
  --since 7d --status failed -o json

A windowed query — yesterday’s runs only:

cronix history billing.reconcile-payments \
  --backend systemd-timer \
  --since 48h --until 24h --limit 0

Notes

since and until are durations, not timestamps. Both are interpreted as “this long ago”; --until 0s is identical to omitting it. Go duration syntax: 30m, 2h, 48h, 168h (no 7d shortcut — use 168h).
One row per terminal run, not per attempt. A job that succeeded after two retries appears as one ok row with attempt=3. The intermediate retry attempts are folded into that record by the History reader.
crontab returns no rows today. The crontab backend’s History is a nil stub awaiting a syslog reader. Use cronix list to confirm the schedule is installed and watch your distro’s syslog for cronix trigger invocations until then.
For “is it installed and what’s the schedule?” use cronix show. For “is it diverged from my manifest?” use cronix drift.