Monitor cluster high availability

Select a documentation version:

This page describes how to monitor the high availability topology, node health, and replication lag of your AlloyDB Omni cluster using the Overview page in the AlloyDB Omni web user interface.

Before you begin

Before you monitor cluster high availability in the web UI, make sure that you have completed the following prerequisites:

  1. Install the web UI.
  2. Connect to the web UI and register the primary and standby or read pool instances for your cluster.

Monitor cluster topology and replication lag

You can use the Overview page in the web UI to verify that standby and read pool nodes are actively streaming from the primary instance and to check for replication latency without running manual pg_stat_replication queries on individual nodes.

To monitor your cluster high availability status and replication metrics, complete the following steps:

  1. Connect to the web UI and select your cluster.
  2. In the navigation menu, click Overview.
  3. In the Cluster overview section, review the overall health metadata for your cluster:
    • Cluster ID: The identifier of the selected cluster.
    • Database version: The PostgreSQL engine version running on the cluster (for example, 18.1).
    • Status: The operational health state of the cluster (for example, Ready).
  4. In the Instance details section, inspect the CPU Usage chart for the past hour and review the instance topology and replication table, which displays the following columns for the primary instance and any standby or read pool instances:
    • Instance name: The role or name of the instance in the cluster topology (for example, Primary, standby, or read pool node).
    • Sync State: The replication synchronization mode of the standby or read pool instance.
    • Sender State: The current WAL sender state on the primary instance streaming to the replica.
    • Replay Lag: The time delay or lag in replaying WAL records on the standby or read pool instance.
    • Write Lag: The lag in writing WAL records to disk on the standby or read pool instance.
    • Flush Lag: The lag in flushing WAL records to durable storage on the standby or read pool instance.

What's next