Problems and known errors
/tickets?filter=problems, and /tickets/<id> for one. Needs
module.tickets and ticket.view.
A problem is the desk’s record of the cause behind repeated incidents. An incident is what somebody reports; a problem is what the desk finds out.
A problem is a ticket of type Problem. It has a number, a queue, an assignee, a conversation and service level targets like any other ticket, and three fields of its own:
| Root cause | What is actually wrong |
| Workaround | What people can do until it is fixed |
| Known error | A mark, with the date it was set, saying the cause and workaround are understood |
The Problems view
Section titled “The Problems view”Problems is a view on Tickets, the last tab after Mine, Unassigned, My queue, Overdue and All open. It lists the open problems, with their count on the tab; with none open, the tab shows no count. Each row says how many incidents the problem caused, and carries a Known error badge where it is one.
Like the other tabs, the queue keeps it while you work: open a problem, come back, and Tickets is still on Problems. Status, Priority and Filter work on it as on any tab, and Save view keeps the result as a view of your own.
Problems is not a page in the menu. To get there from anywhere, type problems, known error, root cause or workaround into Search or ask.
Raising a problem
Section titled “Raising a problem”- On Tickets, select the incidents that share a cause.
- Choose Raise a problem in the bar that appears.
- Say what the problem is in Subject, then choose Raise problem.
The problem opens, with each selected ticket linked to it by Was caused by.
- The selected tickets have to sit on one desk, and the problem joins that desk. When they all share a campus, the problem takes that campus too.
- Select up to 200 tickets at a time. A problem cannot be one of them.
- Raising a problem needs
ticket.createandticket.update, and it is recorded in the audit log.
To link a later incident to a problem that already exists, open the incident and, under Parent, child and related tickets, choose Relate a ticket. Pick Was caused by as the relationship, then find the problem by its number or subject.
A major incident offers Raise a problem too, once it has ended and its ticket is resolved.
On the problem
Section titled “On the problem”The problem’s Work tab holds Cause and workaround. On a wide screen, where Work is a column of its own, it is there.
- Root cause and Workaround are saved with Save.
- Known error is a tick box. Ticking it records the date; clearing it removes it.
- Incidents it caused lists each incident with its status. Only the incidents you can open are listed. The rest are counted, as “2 more incidents on desks or campuses you cannot see”, and nothing else is said about them.
Changing the three fields needs ticket.update, and the audit trail names the
fields that changed. Somebody without ticket.update reads the fields and sees
a Known error badge where there is one.
Where the workaround shows
Section titled “Where the workaround shows”On every incident caused by a problem with a workaround, the Workaround shows in the ticket’s Related section, under the linked records. It names the problem it came from, and says “known error” when the problem is one. Only problems the reader can open are shown.
Where the school’s AI writes the ticket’s Summary, the model is given the workaround to read as well, but only from a problem on the same desk and campus as the incident.
A write-up in the desk’s documentation reaches the incidents as well: a document bound to the problem is listed first on each incident it caused. See known errors.
Resolving a problem
Section titled “Resolving a problem”Resolve a problem the usual way, from its status. When incidents it caused are still open, Plugboard asks whether to resolve them too.
Resolve and reply resolves each open incident you can see and change, and sends its requester one public reply: that the underlying cause is fixed, and that if it happens again they can reply and the desk will reopen the ticket. The reply is the same fixed wording every time, never text from a model. An incident that could not be resolved, because it has moved on, is named and left as it is.
If you say no, Resolve them too… stays on the problem for later. Resolving
the incidents needs ticket.update and is recorded in the audit log.
A recurring fault stays an incident
Section titled “A recurring fault stays an incident”A person reporting the same fault again is still reporting an incident. Plugboard’s built-in rules never make a requester’s ticket a problem:
- A ticket that says the fault is happening again (“again”, “keeps failing”, “third time”) stays an incident. Its triage note adds “it may be recurring”, so the desk can link it to a problem.
- The built-in rules set the type Change only on a ticket the desk raises: by phone, at the desk, through the API or on a schedule. A ticket from the portal, a kiosk, email, Teams or another help desk Plugboard mirrors stays an incident.
- AI triage never proposes Problem for a requester’s ticket.
A school’s own classifier rules can still set any type.
In reports
Section titled “In reports”The Open problems report lists the open problems with their known errors and the incidents each one caused. The tickets dataset in the report builder has Caused by problem and Known error. See reports.
A problem belongs to the service it is filed under and to
the services of the incidents it caused. For people with service.view, Open
problems shows each problem’s services and can group by service, Issue
insights shows each service’s open problems and known errors beside its
issues, and the Service review counts them for
each service.
Flows do not raise problems or resolve their incidents. Deciding that tickets share a cause, and closing the incidents it caused, are a person’s decisions.
Permissions
Section titled “Permissions”| Permission | Allows |
|---|---|
ticket.view | Open the Problems view and a problem |
ticket.create and ticket.update | Raise a problem from selected tickets |
ticket.update | Edit the root cause and workaround, mark a known error, resolve the incidents |