Monitor and tune Azure Cosmos DB for PostgreSQL

  • Article

APPLIES TO: Azure Cosmos DB for PostgreSQL (powered by the Citus database extension to PostgreSQL)

Monitoring data about your servers helps you troubleshoot and optimize for your workload. Azure Cosmos DB for PostgreSQL provides various monitoring options to provide insight into the behavior of nodes in a cluster.

Metrics

Azure Cosmos DB for PostgreSQL provides metrics for nodes in a cluster, and aggregate metrics for the group as a whole. The metrics give insight into the behavior of supporting resources. Each metric is emitted at a one-minute frequency, and has up to 30 days of history.

In addition to viewing graphs of the metrics, you can configure alerts. For step-by-step guidance, see How to set up alerts. Other tasks include setting up automated actions, running advanced analytics, and archiving history. For more information, see the Azure Metrics Overview.

Per node vs aggregate

By default, the Azure portal aggregates metrics across nodes in a cluster. However, some metrics, such as disk usage percentage, are more informative on a per-node basis. To see metrics for nodes displayed individually, use Azure Monitor metric splitting by server name.

Note

Some clusters do not support metric splitting. On these clusters, you can view metrics for individual nodes by clicking the node name in the cluster Overview page. Then open the Metrics page for the node.

List of metrics

These metrics are available for nodes:

Metric Metric Display Name Unit Description
active_connections Active Connections Count The number of active connections to the server.
apps_reserved_memory_percent Reserved Memory Percent Percent Calculated from the ratio of Committed_AS/CommitLimit as shown in /proc/meminfo.
cpu_credits_consumed CPU credits consumed Credits Total number of credits consumed by the node. (Only available when burstable compute is provisioned on the node.)
cpu_credits_remaining CPU credits remaining Credits Total number of credits available to burst. (Only available when burstable compute is provisioned on the node.)
cpu_percent CPU percent Percent The percentage of CPU in use.
iops IOPS Count See the IOPS definition and Azure Cosmos DB for PostgreSQL throughput
memory_percent Memory percent Percent The percentage of memory in use.
network_bytes_ingress Network In Bytes Network In across active connections.
network_bytes_egress Network Out Bytes Network Out across active connections.
replication_lag Replication Lag Seconds How far read replica nodes are behind their counterparts in the primary cluster.
storage_percent Storage percentage Percent The percentage of storage used out of the server's maximum.
storage_used Storage used Bytes The amount of storage in use. The storage used by the service may include the database files, transaction logs, and the server logs.

Azure supplies no aggregate metrics for the cluster as a whole, but metrics for multiple nodes can be placed on the same graph.

Next steps