Skip to main content

Architecture

IPsaDataProvider.GetSnapshotAsync()
  -> PsaSnapshot
  -> PriorityEngine.Evaluate()
  -> DashboardState
  -> BusyBarRenderer.RenderAsync()

Components​

IPsaDataProvider​

The pluggable provider seam. One method, GetSnapshotAsync, returning a PsaSnapshot. Implementations are registered in dependency injection, and the Psa:Provider configuration setting selects which one runs at startup. See Adding a PSA provider for how to implement one.

HaloPsaDataProvider​

The built in provider. Authenticates to Halo's REST API through OAuth2 client credentials, exchanging a client ID and secret for a bearer token that it refreshes as needed, then queries Halo for the ticket data a PsaSnapshot needs.

PsaSnapshot​

The provider agnostic data contract: open ticket count, priority counts, SLA risk tickets, unassigned ticket count, and VIP customer tickets. See Adding a PSA provider for what each field means to a new implementation.

PriorityEngine​

A pure function, PsaSnapshot in, DashboardState out. No I/O, no PSA or BUSY Bar dependency. It evaluates a five level priority order, top to bottom, first match wins:

  1. Rank 1 tickets present: critical mode.
  2. SLA risk tickets present (breaching within the configured threshold): SLA warning mode.
  3. VIP customer tickets present: critical mode, VIP variant.
  4. Unassigned tickets present: normal mode with an unassigned count callout.
  5. Otherwise: normal mode, standard dashboard.

DashboardState​

Which display mode is active, plus the specific values to render: counts, priority names when the provider supplies them, and SLA timing.

BusyBarRenderer​

Turns a DashboardState into DisplayDrawAsync calls against the BUSY Bar, one method per display mode. For critical mode it can render several pages (whichever condition triggered it, then a rank 2 tier, then a rank 3 tier, then SLA risk tickets if any), picking the requested page through its cyclePage parameter. See PollingBackgroundService below for how those pages get shown in turn.

PollingBackgroundService​

A .NET BackgroundService driving the poll loop on Psa:PollIntervalSeconds. A failed poll (Halo unreachable, authentication failure, BUSY Bar unreachable) is logged, and the service tries again on the next interval rather than crashing the worker. When a poll produces critical mode with more than one page to show, the service cycles through those pages every Dashboard:DisplayCycleSeconds seconds for the rest of that poll interval, so every tier gets screen time rather than only whichever one triggered the alert. Normal and SLA warning mode have only one page each, so they are drawn once and left until the next poll. On shutdown, the service clears the display so a container restart or deploy does not leave stale dashboard state on screen.