At a glance
Active Replicas is the count of replica servers currently attached to this MySQL source and streaming changes from it. It is the simplest health check on your replication topology: if the number drops below what you expect, a replica has disconnected and your read-scaling or failover headroom has shrunk. The value is sampled in real time so a dropped replica shows up within one refresh cycle.
What it tracks
The card counts the replicas a source server can see, typically fromSHOW REPLICAS (formerly SHOW SLAVE HOSTS) on the source, or from the replica-side connection registry. A replica appears here while its IO thread holds an open connection to the source and is registered with a server ID. The expected number is whatever topology you designed: a single read replica reads 1, a three-node read pool reads 3, and so on. A reading below your baseline means a replica process has stopped, lost its network path, or been deliberately detached. The card has no alert threshold of its own (time window: RT); it is a context number you read alongside the lag and thread-health cards, which carry the alerts.
Reconcile against the source: run SHOW REPLICAS; on the source instance, or compare against the replica roster in your managed-service console (for example the Amazon RDS read-replica list or the Aurora cluster members view).