Maintenance & Conformance
Two kinds of thing run in the background on a schedule. Sweeps keep the system tidy — clearing out what has expired, finishing what was left half-done. Conformance runners prove a subsystem still works and raise the alarm when it does not. Each has its own page, and each gives you the same two facts: when it last ran, and what happened.
Where to find it
| Screen | Where it lives | What it is for |
|---|---|---|
Maintenance — /admin/maintenance | Settings → System → Maintenance | the clean-up sweeps |
Conformance — /admin/maintenance/conformance | Settings → System → Conformance | the checks that prove things work |
Each page has its own view permission, and Run now on each needs its own run permission.
Maintenance sweeps
The table lists each Sweep — its name and a one-line description — with its Interval, whether it is Enabled (On or Off), its Last run and its Next due. A sweep that is due says due now under its next-due time; one retrying after a failure says retrying after the failure.
The Last run cell gives the result and the time, and underneath, what the run did:
| Result | What it means |
|---|---|
| OK | it ran and finished; underneath, what it changed, or no changes when there was nothing to do — not a failure |
| Failed | it ran and reported a failure; underneath, the failure in its own words |
After a failure, the same cell also says when the sweep last worked — Last succeeded: and the time — or Has never succeeded yet. rather than showing a blank. A sweep that has not run at all says Never. Every run is recorded in the system event log.
Run now in the Actions column runs one immediately, whatever its schedule. If another worker is already running that sweep, it says Already running elsewhere — skipped. rather than running it twice.
Conformance runners
A conformance runner is a background check that proves a subsystem still works: the phone stack, mail, the gateway, sign-in, integrations. It runs on a schedule and notifies operators when something breaks. The point is that a failure announces itself, instead of waiting to be discovered by a person trying to use the thing.
Each card shows the runner's name and description, its state, its Last run (with manual or scheduled), how often it runs (Every), when it is Next due, and the checks of its last run, each marked ✓ or ✗. When something failed, the card gives the failure in words; while a run is under way it says Running since and the time.
| State | What it means |
|---|---|
| Passing | every check passed |
| Failing | at least one check failed; the card names the first |
| Skipped | it did not apply, and the card says why after Skipped: |
| Never run | no run recorded yet |
Run history lists past runs — all of them, or one runner's, picked in Runner — with their Runner, Started time, Trigger (scheduled or manual), Result, how many Checks ran (and how many failed), and the First failure, so a check that passes most of the time and fails on Tuesdays is visible as a pattern rather than as one bad card.
Run now starts a runner by hand. It warns you — Started — this can take a while; the card refreshes when it finishes. If the runner is already running, it says so rather than starting a second copy.
All runners passing or skipped. is the healthy summary. When some are failing, the header says how many — for example 2 failing.