Service levels

Admin, SLA (/admin/sla). Needs module.sla.

Set response and resolution targets per priority. Plugboard flags overdue work and raises an alert when a target is breached.

Setting targets

One row per priority. Each carries:

Response target	Minutes until somebody should have responded
Resolution target	Minutes until it should be done
Enabled	Whether this priority is tracked at all

A starting set that suits most school desks:

Priority	Respond	Resolve
Urgent	1 hour	4 hours
High	4 hours	1 day
Normal	1 day	3 days
Low	2 days	10 days

Set these to what you actually intend to achieve, not to what sounds impressive. A target nobody meets is a queue permanently coloured red, which is the same as no targets at all.

Define priorities first

Targets are per priority, so the priorities have to exist before you can set them. See workflow.

The clock

The clock runs from creation. It stops when the work is settled, meaning nobody is waiting on the desk.

Which statuses count as settled is derived from their meaning (resolved and closed) rather than from their names, so renaming a status does not change the clock’s behaviour.

Pausing. Work waiting on somebody outside the desk (a part, a vendor, a reply from the person who raised it) can pause, so time spent blocked does not count against you.

This matters for honest numbers. Three weeks waiting on a vendor is not your desk being slow.

What a breach does

Three things:

1. The submission or ticket is flagged, and shows as overdue in the queue.
2. An alert is raised, one shot per item rather than repeatedly.
3. The overdue count on the dashboard goes up.

The alert fires once. A breach that stays breached does not generate a new alert every minute, because people mute an alert that repeats.

Working the overdue list

The daily habit worth building: filter the queue to Overdue first thing, before Unassigned.

The most common cause of an overdue item at a school desk is not that it is hard, but that it is waiting on somebody and nobody set the status to say so. Setting the right status is both accurate and stops the clock.

Reporting

The repair turnaround report shows how long repairs actually took, by month. Compare against your targets rather than assuming.

If the average is climbing, either volume is up or something is stuck. The overdue filter tells you which.

Applies to both queues

Submissions and tickets are both covered, using the same per-priority targets.

Turning it off

If your desk does not work to targets, leave module.sla off. Everything else works without it, and the queue is simply not colour-coded by lateness.

That is a legitimate choice for a two-person desk where the queue is short enough to see all of.