Procedure
Monitoring a connection
Walks through watching a live connection, reading its recent messages and spotting any that need attention from an operator.
This page watches a live connection. You read its health, follow its recent exchanges, and spot the messages that need an operator's attention.
The health board
The console shows a health board, one card per connection, each reading healthy, degraded or down. A card carries the time of the last successful call and the run of consecutive failures, and a per-category roll-up sits above the cards.
Health is passive. It is read from the recorded outcome of real traffic rather than a synthetic ping, and the board refreshes live off one listener rather than a poll.
The bridge's own liveness
The orophin bridge reports its own liveness through a regular heartbeat, and the board reflects it, so a quiet connection and a dropped bridge read differently.
Drilling into a connection
- 1
Open a card
Click a degraded or down card to open the connection.
- 2
Read the recent history
See the last error class and the run of failures, together with the recent forensic rows.
- 3
Follow one exchange
The rows correlate on one trace, so a single exchange reads end to end.
The forensic trail
Behind the board sits a forensic trail, one row per exchange with an outside system, held free of patient data. A separate stream records security signals, a rate-limit breach or a flood, for a reviewer to read. Both are read only, and both correlate on the one trace, so an operator follows an exchange while the record stays private.
Messages that need attention
Most traffic settles itself. A few signals ask for an operator, and the board points to where each is handled.
| Signal | What it means | Where to look |
|---|---|---|
| Set-aside message | A message retried past its limit and held for review. | The outbox review, which raises an alert as the count rises |
| Unmapped code | A code with no map, queued for a person. | The terminology queue |
| Degraded connection | A run of failures against a system. | The connection card |
A switched-off connection
A connection switched off surfaces on the wire as a plain error, and the console shows a plain-language notice. A clinician's screen degrades softly and keeps working while the connection stands down. The switch itself is described in connections and credentials.
Files people uploaded
The console also lists the files staff uploaded, each shown by a hashed name in place of the raw filename, so an operator traces an upload while the record stays private. Read where those files land in uploads and documents.
Common questions
Why is a connection degraded rather than down?
Degraded reads a run of failures against a system that still answers sometimes. Down reads a system that stopped answering. Both verdicts come from real traffic.
Does the forensic trail hold patient data?
It holds one row per exchange, free of patient data, correlated on a trace, so an operator follows an exchange while the record stays private.
Review the bridge that carries every connection in the bridge.