A bot’s words travel over a lane: a connected model subscription it answers with or a channel it posts to, such as Slack or SMS. When a lane breaks, the bot may be unable to reply. Settings → Status shows each tracked lane and its current health.

​
The Status panel

Open Settings → Status on Desktop. On the phone layout and the installed phone app, Settings → Status opens the same list full screen. Each row has:
  • A colour and a word: Working (green), Slow (amber) or Down (red). The word is always shown, never colour alone.
  • Since when, for anything that isn’t working.
  • The affected subscription and the reason in plain words, such as needing to sign in again.
  • The bots it affects.
  • Fix, which opens Settings → Subscriptions (vendor lanes) or Settings → Connected apps (channels).
Lanes that need a look are listed first. Check again asks for a fresh reading; otherwise the panel refreshes every minute while it is open. If your runtime can’t be reached, the panel says so and keeps showing the last result. A bot whose subscription lane is down also shows Brain offline, followed by the affected subscription and reason, on its line in the Bots list in place of its last message.

​
What is checked

Not on the board yet: model providers and the runtime itself.

​
Get told when a lane goes down

When a vendor lane turns red, Allternit opens a Lane down item in Needs you naming the bots that can’t reply, and sends a push. If any of those bots has a phone number or takes texts, the item is critical, because callers are waiting on it. When the lane works again the item resolves by itself. A lane left down is reminded about again until you fix it.

​
Channel pause

If a channel’s last five messages all fail within ten minutes, Allternit pauses sending on that channel for two minutes, then tries one message. If that works, sending resumes; if it fails, the pause starts again. Messages are not discarded while paused; the Needs you inbox lists the ones that did not go through.

​
For developers

  • Gateway. Migration 0010_lane_canary.sql adds canary_runs. worker/canary.ts ticks every SUBS_GATEWAY_CANARY_SEC seconds (default 180, 0 turns it off), calls WorkerPool.probeResident for each enabled account, and records the run. A pass calls recordAdapterProbePassed, so an open drift breaker closes. GET /v1/lanes (scope tasks:read) returns { lanes: [{ account_id, provider, label, health, state, breaker, lastProbeAt, lastOkAt, lastProbeOutcome, latency_ms }] }.
  • Runtime. subscription_sync pulls v1/lanes with each minute’s sync, writes state changes to lane_samples (V259, kept 14 days), opens lane_down through netops::alerts::open on a move into red and resolves it on recovery. GET /api/v1/lanes/status (netops/lanes.rs) returns [{ laneId, kind: vendor_account|channel|model_provider|runtime, label, state: green|amber|red, since, reason, affectedBots: [{id,name}], action? }]. channel_lane_open(conn, binding_id) is the channel breaker; it is computed from channel_message_log, so it needs no state of its own.
  • UI. src/lib/lanes/lane-status.ts (one shared poller), src/components/lanes/LaneStatusBoard.tsx, and bot-status-line.ts for the Bots list line.