Control view
Check center health in one place: external connections, agents, record counts, Edge logs, and handled errors, from top to bottom.
When to use it
Start here when you ask, 'Is the center working normally?' When an agent is unresponsive, external access fails, or errors repeat, this view helps distinguish connection problems from agent problems. Open Control view from the sidebar's system row.
It queries once on opening, then every five minutes. Refreshing stops when you leave the screen.
assets/turtlecenter/ui-control-room-1.png · Recommended size: 1600×900Refresh row, three status cards, Agent status/Work data/Edge logs, and Exception logs. Numbered annotations are planned.Layout
| Order | Area | What it shows |
|---|---|---|
| 1 | Top row | Last checked: time on the left; a Refresh button on the right |
| 2 | Three status cards | ngrok, Edge, and TurtleGate connection status |
| 3 | Agent status | Each center agent's state, queue, and current task |
| 4 | Work data | Record counts by record-room category |
| 5 | Edge logs | Recent records from the Edge connection service |
| 6 | Exception logs | How errors were handled, with details |
Top row: last check and refresh
Last checked is when the cards were last queried. Refresh checks again immediately. The button is disabled during a query. Failure changes the line to Last check failed · Try again later. A record-count-only failure adds a count-query failure time to the same line.
Three status cards
Cards check the three connections needed by the center and show results with color and text. Each has four lines: name, state/details, query time, and recommended action.
| Card | What is checked | How |
|---|---|---|
| ngrok | Is the inbound tunnel open? | Probes the public address for a response within five seconds. If none arrives, uses the ngrok connection service status instead. |
| Edge | Does the center's connection endpoint respond? | Queries the local Edge endpoint on port 5714. A Ready response within five seconds is considered normal. |
| TurtleGate | Is installation authorization and registration connected? | To be confirmed In the current source this card does not query status and shows Unknown · Query failed. Documentation will be updated when querying is implemented. |
Card states and colors
| State | Color | Meaning | Action line |
|---|---|---|---|
| Normal | Green | Working normally | Operating normally |
| Starting | Yellow | Starting or connecting | Waiting for startup |
| Attention | Yellow | Working, but needs checking | Check detailed logs |
| Stopped | Red | Not running | Check service status |
| Stale | Gray | Information is old | Refresh required |
| Unknown | Gray | Not yet verified | Checking status information |
Card detail labels
| Card | State · Details | Meaning |
|---|---|---|
| ngrok | Normal · Connected | Public address responds, or connection service reports connected |
| ngrok | Attention · Connected · Authentication required | Tunnel is open but the public address requires authentication |
| ngrok | Starting · Connecting | Opening the tunnel |
| ngrok | Stopped · Not connected | Tunnel is closed or not configured |
| ngrok | Attention · Authentication required | ngrok authentication is needed |
| ngrok | Attention · Abnormal status | Connection reports failure |
| Edge | Normal · Response normal | Edge responded Ready |
| All cards | Unknown · Query failed | No response or query error, including no answer within five seconds |
Agent status
Agent status gives each center agent a row showing its state and backlog. Sorted by name and limited to agents directly operated by the center.
| Column | Meaning |
|---|---|
| Agent | Agent name |
| State | Connected · Working · Received · Limit reached · Fault · Unknown (see below) |
| Pending queue | Unprocessed messages in the agent inbox, or — if unavailable |
| Current task | Working if processing a task; otherwise Waiting |
| State value | Meaning |
|---|---|
| Connected | Ready to accept work, corresponding to the blue sidebar indicator |
| Working | Processing assigned work |
| Received | Just received a message |
| Limit reached | The AI service usage allowance is exhausted |
| Fault | An agent or connection has a problem |
| Unknown | The state is not yet known |
If only one queue keeps growing, open that agent's room. For sidebar colors, see the Screen overview.
Work data
Work data is a two-column table of record-room category counts. Columns are Category and Record count. It initially shows Waiting for query and changes to Query failed on failure. Open the Record roomto inspect contents.
Edge logs
Edge logs list connection activity newest first. Reads the newest 200 events from the connection service log file and refreshes every five minutes.
| Column | Meaning |
|---|---|
| Event time | Recorded timestamp |
| Level | ERROR is red, WARN yellow, and others light blue |
| Activity | Type of action |
| Stage/result | Where and how the action finished |
This view shows the table, not agent filters or Refresh/Clear controls. Selecting a row puts sender, recipient, preview, and full details in the detail area of Exception logs below.
assets/turtlecenter/ui-control-room-2.png · Recommended size: 1600×900Close-up of the three middle areas, preferably showing both Working and Waiting agents.Exception logs
Exception logs show how the center handled errors. Only errors and warnings appear, not informational events. Shows the newest 200 and refreshes every five minutes.
| Column | Meaning |
|---|---|
| Event time | Time the error occurred |
| Level | ERROR is red; WARN is yellow |
| Agent | Related agent, or System for an unrelated error |
| Component | The center component where it occurred |
| Operation | The action underway |
| Exception type | Name of the error type |
| Catch message | Text left by the error |
| Handling result | Continued · Canceled · Closed · Failed and closed |
| Details | Additional explanation |
| Handling result | Meaning |
|---|---|
| Continued | Handled the error and continued the work. |
| Canceled | Canceled only that action. |
| Closed | Finished and closed the action. |
| Failed and closed | Closed the action as failed for safety. |
Above the table is a Refresh button with Latest event time · Refresh time · N events. Problems append DB query failed, N emergency log corruptions, or N missing events. Click headings to sort or drag to reorder. Selecting a row combines time, level, agent, component/operation, result, exception type, message, and details below.
Workflows
When an agent does not respond
- Open Control view and click Refreshto query the latest state.
- Check the agent state and queue. Limit reached or Fault indicates an agent-side cause.
- If Connected but the queue keeps growing, use the Communication monitorto check message arrival.
- Find matching agent entries in Exception logs and read the detail message.
When the center cannot be reached externally
- Check the ngrok and Edge cards.
- If Edge is normal and ngrok is Stopped, investigate the tunnel. Settingsunder External connections to check its state.
- If Edge is Unknown, check the center's connection endpoint first.
Things to know
- Control view is read-only. It has no agent on/off buttons.
- Automatic refresh is every five minutes. Use Refresh for a recent change.
- If an error repeats, save the first message and time, then follow Troubleshootingto request support.
Related screens
Communication monitor · Record room · Settings · Screen overview