Try Free
Documentation

System Monitor

System Monitor

The System Monitor is a live dashboard of your MongoDB server's health. Seven charts stream once per second over a server-sent events connection — connections, new connections, network throughput, request rate, operation counters, active clients, and queue length. Turn on recording and the same metrics are stored server-side, so you can scroll back to what the server was doing an hour ago instead of guessing.

Quick Start

  1. Connect to a MongoDB server in the sidebar.
  2. Open System Monitor from the connection's action bar or section menu.
  3. Charts start streaming immediately — the Live pill in the header turns green.
  4. Click Start Recording if you want history kept for this connection.
  5. Use the time-window selector to look back over recorded data, or Pause to freeze the view while you read a spike.

Opening the Monitor

The monitor is scoped to one connection. Open it from the connection's action bar in the sidebar, or from the connection's section menu (System Monitor). Opening it a second time for the same connection focuses the existing tab rather than creating a duplicate — each connection gets exactly one monitor, and the tab title carries the connection name.

The Charts

Every chart plots a rolling window of live values. Each series has a colored dot in the chart header — click a name to toggle that line on or off, which is the fastest way to read a chart where one series dwarfs the others. Hover anywhere on the plot for a tooltip with the timestamp and every visible series value.

ChartSeriesUnitWhat it tells you
Connections Current, Active, Available count Whether you're approaching the server's connection ceiling. Available trending toward zero means a pool leak or an undersized limit.
New Connections Connections conn/s Connection churn. A high, steady rate usually means a client that isn't pooling.
Network Throughput In, Out KB/s Bytes moving in and out of the server. A large Out with a small request rate points at unprojected queries returning whole documents.
Request Rate Requests req/s Overall traffic the server is fielding — the baseline you compare every other chart against.
Operations Insert, Query, Update, Delete, GetMore, Command ops/s The workload mix. GetMore climbing on its own means large cursors being paged through.
Active Clients Readers, Writers count How many operations are actually executing right now, split by read and write.
Queue Length Readers, Writers count Operations waiting on a lock. Anything sustained above zero is the clearest sign of contention.

Charts follow the app theme — switch between light and dark and the axes, grid, and tooltips repaint in place. They also resize with the panel, so the monitor works split-screen next to a query tab.

Header Controls

  • Connection badge — the connection the charts belong to.
  • Status pill — Live (green, pulsing) while metrics are arriving, Paused (yellow) after you pause, Disconnected (red) if the stream drops.
  • Recording badge — a red pulse shown whenever recording is active for this connection.
  • Last — the clock time of the most recent data point, so you can tell a flat line from a stalled stream.
  • Time window — the range the charts display; see below.
  • Pause / Resume — stops and restarts the metric stream. Pausing dims the charts and holds the current data in place.
  • Start / Stop Recording — controls server-side persistence of the metrics.
  • Delete History — permanently removes every stored metric for this connection.

Time Windows and History

The selector at the top right sets how far back the charts reach: 5 minutes, 10, 15, 30, 1 hour, 2, 6, 12, 24 hours, 7 days, or Custom Range. The default is 5 minutes.

Picking a window loads the recorded metrics for that period and replays them into the charts, then live points continue to arrive on top. Custom Range swaps the selector for two date-time fields — pick a start and an end and click Apply; the range defaults to the last 24 hours, and the start must come before the end.

History needs recording

Historical windows are served from recorded data. If recording has never run for a connection there is nothing stored, so a longer window simply leaves the live charts as they are instead of filling in the past. Start recording before the window you want to be able to inspect later.

Recording Metrics

Start Recording begins persisting this connection's metrics server-side; Stop Recording ends the session. While recording is active the stream keeps running even after you close the monitor tab — reopen it and the charts pick straight back up, with the gap you were away already filled in from history.

Delete History clears every stored metric for the connection. It cannot be undone, so use it to reclaim space or reset a noisy baseline, not as a way to pause collection — that's what Stop Recording is for.

Pro Tips

  1. Read Queue Length first. Latency complaints almost always show up there before they show up anywhere else. Active Clients tells you how much work is in flight; Queue Length tells you how much is stuck waiting for it.
  2. Turn recording on before a load test or a release. Live charts only answer "what is happening"; recorded history is what lets you answer "what happened at 14:05".
  3. Isolate a series instead of squinting. On the Operations chart, click every legend entry except the one you're chasing — the Y axis rescales to that series.
  4. Pause before you screenshot. Pausing freezes the view so a spike stays on screen while you copy the timestamp into a ticket.
  5. Pair it with the Query Monitor. The System Monitor shows the shape of the load; the Query Monitor names the individual operations behind it.
  6. Watch New Connections during deploys. A step change after a release usually means the new build lost its connection pooling.

Ready to try VisuaLeaf?

Download and start managing your MongoDB databases with ease.

Download Free Trial