How to read this console
- est. wait
- Waiting ÷ what the room can actually deliver per minute: the measured outflow when capacity binds, the configured rate otherwise. "≥" marks a lower bound: the room is at its concurrency ceiling and nothing has come out for a minute, so raise max-concurrent or shorten the session TTL; raising the rate will not help. "∞" means nothing is being let through: the room is paused or its rate is 0/min.
- worst est. wait
- The longest est. wait across the rooms with people in them. A room admitting nobody has no estimate, so it is named beside the number ("excludes N at ∞") rather than erasing it.
- Active · Paused · Bypass
- Active promotes visitors at the rate dial. Paused stops automatic outflow: the queue only grows until you resume, and "Let through" is the one way in. Bypass switches the queue off: everyone who arrives, and everyone already in line, goes straight to the target site at full speed.
- Autotune
- The server moves the rate dial from the target's measured response time: a slow or unreachable target lowers it, a healthy target with a queue raises it. The cog sets the latency threshold and the bounds the rate may move within.
- Export CSV
- The full retained history (1h) as one row per 5s bucket, with the measured actual-wait percentiles and the timeline annotations on the row they happened in.
- Alerts
- Desktop notifications in this tab only, on surge start, target-URL failure and stalled outflow. They stop when the tab closes, so they are a watch you are keeping rather than one the server keeps for you. Paging that outlives the tab is configured on the server with ALERT_WEBHOOK_URL and shown under Diagnostics → Alerting.
- Chart minimap
- Drag on the minimap to zoom, drag the window to pan, double-click to reset. From the keyboard, Tab to the window's start or end handle: the arrow keys move that edge one 5s bucket, Page Up and Page Down move it a minute, Shift with any of those slides the whole window, Home and End take the edge as far as it goes, Escape resets. The 5m · 15m · 1h buttons pick the range the window sits inside.
- Chart legend
- Two scales share the left axis: queue depth on the numbers outside it, joins/min and passes/min (each a trailing 60s window) on the amber /min numbers inside the left axis. When passes/min is too small to share, both buttons relabel to "own scale" and passes reads off the green numbers instead. Est. wait is computed, depth ÷ outflow at that moment, on the right axis in minutes. Actual wait p50 and p90 are measured: the join→pass time of visitors admitted in the trailing 60s, recorded by the server at promotion, so they redraw identically after a reload or a restart. Event marks are what changed the room: rate moves, state flips, threshold activation, autotune, flushes, purges and protection blocks. Each button shows or hides its series.
- All rooms
- One chart for the whole estate. Depth, joins/min and passes/min are summed across rooms; the wait lines show the worst room at each point in time, not an average and not any one room.
- Security filter
- Substring match against the client digest, the signal ids, the room and the User-Agent ("agent" in the placeholder).
- Download CSV (Reports)
- Exactly the rows the Reports table shows for the room, bucket and range picked above it: the BI export. Export CSV beside the chart is the other file, the full 1h live history.
- Visitor drill-down
- The ticket number from the queue, or the reference code the visitor reads off their waiting page, in the form 0060-6750.
No waiting rooms yet
Create your first room to start protecting a page with a queue.
This view is read-only. Rooms appear here as soon as an operator creates one.
Room
Every bucket in this window, as a table