Running by hand

The Run now button is on the job list and on the job's own page, behind a confirmation, for the operator role and above. It runs the job off the schedule, at once, exactly as a tick would have: the same command, the same agents, the same secrets.

Reporting per agent

A job on several agents has no single verdict worth returning — one agent may be mid-run while the others stand idle. So the response says what happened on each, and the application sums it up in one sentence:

"Nightly backup" started on 2 of 3 agents — skipped: web-03 (a run of this job is already in flight on this agent)

The three reasons an agent was given no run are the same as on a tick and are listed under assigning agents. An agent that is offline is not skipped: a run is created for it and waits, for a bounded time.

Three cases where running by hand produces nothing:

• the job is disabled — the button says Enable the job to run it,
• the job has no agents,
• the workspace is suspended; reading works, running does not.

What a manual run is and is not

It carries the manual trigger with a pointer icon and is in the run history like any other, output and exit status included. A failed manual run is a failure like any other — it raises the same email a failed scheduled one does.

It does not count as the last automatic run for catching up: Run now at noon does not cancel a nightly occurrence missed at three in the morning. Somebody wanting to deal with a missed occurrence by hand does so in the sense of the work, not in the sense of the bookkeeping — Runjet will catch it up again at startup if the job has catch-up switched on.

The run is a row in the audit log with the name of whoever clicked, and the count of agents where a run was made and where it was not.