MTConnect: live-gate the refused-connection fast path; expose CommsLossConfirmMs in the AdminUI #852

New Issue

2026-11-11T05:24:28-05:00

dohertj2 commented

2026-11-11 05:24:28 -05:00

From #846. A refused Agent connection now marks subscribed values BadNoCommunication after CommsLossConfirmMs (default 2000 ms) instead of waiting for the ~90 s watchdog. Remaining:

only unit-tested; live-gate it by stopping an Agent (a dedicated one, not the shared fixture other work relies on) and watching an OPC UA client;
CommsLossConfirmMs is JSON-only — the AdminUI form preserves it but has no field;
native MTConnect alarm conditions still get Bad quality only when the watchdog fires; the fast path covers data values only.

From #846. A refused Agent connection now marks subscribed values BadNoCommunication after `CommsLossConfirmMs` (default 2000 ms) instead of waiting for the ~90 s watchdog. Remaining: - only unit-tested; live-gate it by stopping an Agent (a dedicated one, not the shared fixture other work relies on) and watching an OPC UA client; - `CommsLossConfirmMs` is JSON-only — the AdminUI form preserves it but has no field; - native MTConnect alarm conditions still get Bad quality only when the watchdog fires; the fast path covers data values only.

dohertj2 added the type:followup priority:p3 labels 2026-11-11 05:24:28 -05:00

dohertj2 referenced this issue

2026-11-11 05:24:40 -05:00

MTConnect: detect Agent loss sooner; clamp the raised sample count to the Agent's bufferSize #846

dohertj2 referenced this issue from a commit

2026-11-11 07:55:03 -05:00

feat(mtconnect): CommsLossConfirmMs AdminUI field, refused-connection fast path reaches native conditions, dedicated live-gate Agent (#852)

dohertj2 referenced this issue from a commit

2026-11-11 07:55:03 -05:00

Merge #852 MTConnect CommsLossConfirmMs field, condition fast path, gate overlay

dohertj2 commented

2026-11-11 07:55:34 -05:00

Progress (Pushed to master in 03e51457., 340d9d26): AdminUI MTConnect form has a "Comms-loss confirm (ms)" field (negatives preserved; JSON pinned through the driver parser); native MTConnect conditions now get quality-only BadNoCommunication on a confirmed refused connection and recover to Good without re-raise. A dedicated gate Agent overlay docker-dev/docker-compose.mtconnect-gate.yml keeps the hostname resolving while the Agent is stopped (a plain container stop gives a DNS failure = ambiguous = watchdog path); recipe in docs/drivers/MTConnect.md. Keeping open for the live gate on docker-dev.

Progress (Pushed to master in `03e51457`., `340d9d26`): AdminUI MTConnect form has a "Comms-loss confirm (ms)" field (negatives preserved; JSON pinned through the driver parser); native MTConnect conditions now get quality-only `BadNoCommunication` on a confirmed refused connection and recover to Good without re-raise. A dedicated gate Agent overlay `docker-dev/docker-compose.mtconnect-gate.yml` keeps the hostname resolving while the Agent is stopped (a plain container stop gives a DNS failure = ambiguous = watchdog path); recipe in docs/drivers/MTConnect.md. Keeping open for the live gate on docker-dev.

dohertj2 commented

2026-11-11 08:33:26 -05:00

Live gate PASS on docker-dev (03e51457, gate overlay docker-dev/docker-compose.mtconnect-gate.yml, driver seeded against http://mtconnect-gate:5000, CommsLossConfirmMs 2000):

Agent killed 13:27:15.19 (container FinishedAt — mtconnect/agent ignores SIGTERM, so compose stop kills at the 10 s grace) ⇒ log refused every connection for 2001 ms (definite loss); marking 2 subscribed value(s) BadNoCommunication + annotated 1 native condition(s) BadNoCommunication at 13:27:17; subscribed x_pos went BadNoCommunication at 13:27:17 (2.0 s, was ~94–118 s via the watchdog). The x_travel condition got one event, state kept Inactive/Acknowledged.
Agent restarted 13:27:42 ⇒ the condition got one more event (Inactive/Acknowledged — no re-raise), values Good again at 13:28:05; the ~23 s lag is the /sample reconnect backoff (attempt 9 ≈ 26 s), not the fast path. RE-BROWSE chip raised for the new instanceId as documented.
Recorded in docs/drivers/MTConnect.md (b4ab403c). The UI field was not round-tripped in the browser on this pass (unit-pinned).

Live gate PASS on docker-dev (`03e51457`, gate overlay `docker-dev/docker-compose.mtconnect-gate.yml`, driver seeded against `http://mtconnect-gate:5000`, `CommsLossConfirmMs` 2000): - Agent killed 13:27:15.19 (container FinishedAt — `mtconnect/agent` ignores SIGTERM, so `compose stop` kills at the 10 s grace) ⇒ log `refused every connection for 2001 ms (definite loss); marking 2 subscribed value(s) BadNoCommunication` + `annotated 1 native condition(s) BadNoCommunication` at 13:27:17; subscribed `x_pos` went `BadNoCommunication` at 13:27:17 (**2.0 s**, was ~94–118 s via the watchdog). The `x_travel` condition got one event, state kept Inactive/Acknowledged. - Agent restarted 13:27:42 ⇒ the condition got one more event (Inactive/Acknowledged — no re-raise), values Good again at 13:28:05; the ~23 s lag is the `/sample` reconnect backoff (attempt 9 ≈ 26 s), not the fast path. RE-BROWSE chip raised for the new instanceId as documented. Recorded in docs/drivers/MTConnect.md (`b4ab403c`). The UI field was not round-tripped in the browser on this pass (unit-pinned).

dohertj2 closed this issue

2026-11-11 08:33:26 -05:00

dohertj2 referenced this issue from a commit

2026-11-11 11:28:50 -05:00

docs(mtconnect): record the #852 refused-connection live gate (PASS)

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: dohertj2/lmxopcua#852