Monitor a Cluster¶

On this page

Metrics Displayed in the Clusters View
View All Metrics

Atlas collects and displays metrics for your servers, databases, and MongoDB processes. Atlas displays three charts in the Clusters view and additional charts in the Metrics view.

Note

Free Tier clusters have limited available metrics. For more information, see Atlas M0 (Free Tier) Limitations.

Metrics Displayed in the Clusters View¶

The Clusters view displays charts that monitor your disk usage, IOPS, and database operations. If the collected metrics indicate you need to add capacity to your cluster, see Scale a Cluster.

From the Clusters view you can see the following:

Chart Data

Disk Usage

The total bytes of used disk space for the cluster. For a replica set, the chart displays disk use on the server that hosts the primary. For a sharded cluster, the chart displays the sum of disk usage on all servers that host a primary. The top line of the chart shows the maximum storage capacity on the server or servers.

To view disk use on a specific server in the cluster, click the server’s name, then click the Hardware tab, and then locate the Disk Space Used chart.

Disk IOPS

The input/output operations per second (IOPS) for the cluster. Each point on the chart is the sum of read IOPS and write IOPS.

For a replica set, the chart shows IOPS for the primary. For a sharded cluster, the chart shows the sum of the IOPS on all primaries in the sharded cluster. The top line of the chart shows the maximum IOPS configured for the server or servers.

To view IOPS for a specific server in the cluster, click the server’s name, then click the Hardware tab, and then locate the Disk IOPS chart.

Operations

The aggregated read and write operations for the cluster’s databases. The Read line aggregates all query and getMore operations. The Write line aggregates update, delete, and insert operations.

For a replica set, the chart shows operations for the primary. For a sharded cluster, the chart shows the sum of the operations on all primaries in the sharded cluster.

To view a specific operation by server, click the server’s name to open the server’s metrics. Click the Status tab. Select the Display Opcounters on Separate Charts checkbox to display operations on separate charts. Deselect the checkbox to display operations on a single Opcounters charts.

View All Metrics¶

To view metrics, locate the cluster on the Clusters tab and click Metrics. Atlas displays metrics at the granularity last selected. You can select different granularities.

To add or remove charts, click either the Display or Add Chart drop-down list, depending on which is available. To display chart controls, hover the mouse pointer over the chart. To view metrics at a different level of the cluster topology, use the breadcrumb.

For more information on chart metrics, see the following on this page:

Sharded Cluster Metrics
Replica Set Metrics
Real-Time Tab
Status Tab
Hardware Tab
DB Stats Tab
Chart Controls

Sharded Cluster Metrics¶

When you view metrics for a sharded cluster, you choose to display either shards, mongos instances, or config servers. When displaying shards, you can use the buttons below the chart to select primaries, secondaries, or both. Atlas displays different processes as different lines.

To sum or average the data, click the Display Data drop-down list and select the desired option. To change charts, click the Display drop-down list.

Replica Set Metrics¶

When you view metrics for a replica set, you can limit the display to particular member types using the P, S, and A options.

Real-Time Tab¶

Availability

Real-time metrics is available for the owners of the group.

Click the Real-Time tab to display various real-time metrics.

You can view the stats in either Table view or the Graph view.

Table View¶

Fields	Description
Resident Virtual Mapped	Displays the resident/virtual/mapped memory stats as reported by `mongostat`. For a description of the fields, see mongostat.
Bytes In Bytes Out Connections	Displays the connection stats (number of bytes in, number of bytes out, number of connections) as reported by `mongostat`. For a description of the fields, see mongostat.
Commands Queries Updates Deletes Inserts GetMores	Displays the number of the specified operations (commands, queries, etc.) as reported by `mongostat`. For a description of the fields, see `mongostat`.
Active Readers Active Writers Queued Readers Queued Writers	Displays the number of active reads, queued reads, active writes, queued writes as reported by `mongostat`. For a description of the fields, see mongostat.

Click on the elipsis button (...) to hide or display fields from the table.

Graph View¶

Chart/Table	Description
Memory	Displays the memory stats as reported by `mongostat`. For a description of the fields, see mongostat.
Network	Displays the connection stats as reported by `mongostat`. For a description of the fields, see mongostat.
Operations	Displays the number of operations as reported by `mongostat`. For a description of the fields, see `mongostat`.
Reads & Writes	Displays the number of active reads, queued reads, active writes, queued writes as reported by `mongostat`. For a description of the fields, see mongostat.
Hottest Collections	Displays the collections with the most operations as reported by `mongotop`. For each hot collection, the table also displays the Scaled Load percentage for the collection. Scaled Load is calculated from the read and write times as reported by `mongotop` during a sample period. Specifically, the Scaled Load is the percentage (rounded to the nearest 0.1%) of the read and write times for a collection relative to the read and write times for all collections in the deployment during the sample period. If no read and write operation occur during this period, the Scaled Load may be 0%. The hottest collections correspond to the most current time displayed in the charts. That is, if the display is running (i.e. not paused), the collections correspond to the hottest collections at the current timestamp. If the display is paused, the collections correspond to the hottest collections at the paused time.
Slowest Operations	Displays the slowest operations as reported by db.currentOp(). The operations correspond to the most current time displayed in the charts. That is, if the display is running (i.e. not paused), the operations correspond to the slowest operations at the current timestamp. If the display is paused, the operations correspond to the slowest operations at the paused time. Click on the operation to view the detailed JSON document as reported by db.currentOp() for the operation. In the details panel for a slow operation, you can click on the Kill Op button to kill the slow operation.

To hide/show a metric in the graph, click on the color box for the metric.

Stop and Restart Display¶

To stop the update of the stats displayed, click the Pause button. The Pause button only stops the update of the displays and does not pause the collection of the underlying data.

To restart the display updates, click Play.

Status Tab¶

This tab displays the process metrics Atlas collects from the selected mongod or mongos process. Atlas collects a subset of the output of the MongoDB serverStatus command. See serverStatus in the MongoDB manual.

Hardware Tab¶

Atlas collects the following hardware metrics for your mongod and mongos processes:

Chart	Data
System CPU	The CPU usage of all processes (not just MongoDB processes) on the host partition. For partitions with more than 1 CPU core, this value can exceed 100%.
Normalized System CPU	The CPU usage of all processes (not just MongoDB processes) on the host partition, scaled to a range of 0-100% by dividing by the number of CPU cores.
Process CPU	The CPU usage of the MongoDB process. For partitions with more than 1 CPU core, this value can exceed 100%.
Normalized Process CPU	The CPU usage of the MongoDB process, scaled to a range of 0-100% by dividing by the number of CPU cores.
Disk Space Free	The total bytes of free disk space on the disk partition used by MongoDB.
Util %	The percentage of time during which requests are being issued to and serviced by the partition. This includes requests from any process, not just MongoDB processes.
Disk IOPS	The throughput of I/O operations per second for the disk partition used by MongoDB.
Disk Space Percent Free	The percent of free disk space on the partition used by MongoDB.
Disk Latency	The latency in milliseconds of the partition used by MongoDB.
Disk Space Used	The total bytes of used disk space on the partition that runs MongoDB.

DB Stats Tab¶

This tab displays database metrics Atlas collects through the MongoDB dbStats command. Atlas collects a subset of dbStats output. See dbStats in the MongoDB manual.

Atlas retrieves database metrics every 20 minutes by default but adjusts frequency when necessary to reduce the impact on database performance. If the collection of database statistics still affects performance, you can disable collection by clicking Settings, then Group Settings, and then setting Collect Database Specific Statistics to No.

To select which database to display, click the Select Database drop-down list.

Chart Controls¶

When viewing charts, you can do the following:

Task	Action
View a chart’s description.	Click the i icon next to the chart name. (Not available for cluster-level chart)
Expand a chart.	Click the two-way arrow at the top right of the chart.
Zoom in on a period of time.	Click and drag the mouse pointer over a portion of the chart. To reset, double-click the chart. Note When you zoom, the time selections above the charts automatically update to reflect the new period displayed.
View point-in-time statistics.	Hover the mouse pointer over a point on the chart.
Move a chart.	Click and hold the grabber in the upper left corner of the chart, and drag the chart to the new position.
Export the charts.	Click the Export drop-down list and select the format.
Link to a chart.	Click the curved arrow on the chart and select Chart Permalink.
Email the chart.	Click the curved arrow on the chart and select Email Chart.

← Monitoring and Alerts Respond to Alerts →