Fix Lost Primary

Alert Conditions

You can configure alert conditions in the project-level alert settings page to trigger alerts.

For more information on the alert condition, see Replica set has no primary.

Common Triggers

• The workload exceeds the cluster's throughput limits and compute resources.

• A network issue with the cloud provider is preventing voting members of the replica set from communicating with each other, such that a primary cannot be elected.

Fix the Immediate Problem

• Check the cluster's metrics to determine if there are sufficient compute resources for your workload.

  • If the cluster is exhausting CPU, disk IOPS, connections, or other resources, upgrade to a cluster that supports your workload.

  • If the cluster's metrics are normal, there could be a network issue with the cloud provider. Atlas automatically attempts to repair these issues. If the problem persists, contact MongoDB Support.

Implement a Long-Term Solution

If Atlas collects data during an election, this alert might send a false positive. To prevent such false positives, set the alert configuration's after waiting interval (in the configuration's Send to section).

Monitor Your Progress

View the following charts to monitor whether the cluster is exhausting resources:

• Normalized System CPU

  Monitor CPU usage to determine whether data is retrieved from disk instead of memory.

  If you are unable to see the usage that triggered the alert, zoom in on the Normalized System CPU chart by clicking and dragging your mouse over the period of interest. With a higher-resolution view you may be able to identify acute spikes in CPU usage that weren't visible in the overview.

• Disk IOPs

  Monitor whether disk IOPS approaches the maximum provisioned IOPS. Determine whether the cluster can handle future workloads.

• Connections

  Monitor connections to determine whether the current connection limits are sufficient. If necessary, upgrade the cluster tier.

To learn more, see View Cluster Metrics.