Generator Service Scheduling

Every generator, grouped by pod — click a gen for full detail. Pods are grouped into sections by how many gens need attention, worst first.
Show gens matching
Source: live snapshot from status_api (/api/status via the 5022 read-only turnstile) merged with generator maintenance tags / PM bookings in master_config.json. This is a static pull, not a live-refreshing page — regenerate it to pick up changes since the timestamp above.
Service-interval method: per-gen PM countdown from apps/generators/pm_service.py — overdue = past due date; due = ≤7 days or ≤150 engine-hours out; upcoming ("due soon") = ≤21 days out; no data = anchored with no engine-hour reading from either feed, or the gen has never been seeded from a Mesa schedule (currently true for most/all Texas pods — ND-only seed source per CLAUDE.md). A gen box with no chip and a green edge is running with nothing flagged.
Inspect / Swap are the manual maintenance tags from Site Manager/gen tracker, independent of the PM countdown. Swap covers both active-swap (still running, swap scheduled) and swap (pulled, being swapped) — the same grouping the status-page map uses for "planned out" gens.
Pod grouping / red count: the red number after each pod name is how many of its gens need attention — with no filter checked that's Inspect + Swap + Overdue + Due + Upcoming combined; with filters checked it's just those. Pods are bucketed into sections by that count (3+, 2, 1, 0) so the worst pods sort to the top of the page.
Who to dispatch: today's mapping is by task type only — Swap needs an NGON employee, a PM (overdue/due/upcoming) needs a MESA PM tech, an Inspect needs a MESA SC. This is a placeholder: different Mesa fault/alarm codes likely need different tech levels, and that per-fault-code table doesn't exist yet — the live alarm text is captured per gen (visible in each gen's popup) so that mapping can be layered in later without re-pulling data.
Swap availability: pulled from master_config.json → Spares → "Available Spares" — currently 0 units shown there. That bucket is hand-maintained, not a live feed, so a "not avail" reading may just mean nobody moved a spare into it recently rather than there being no physical spare — verify with dispatch before treating it as ground truth.
Export: both buttons above act on whatever the checkboxes currently select (or every gen needing attention, if none are checked) — CSV for a spreadsheet, or a plain-text list sized for pasting straight into an email body. Neither button sends anything; this page has no ability to send email on its own.