Serverless clusters can be suspended automatically when idle and started again when a connection arrives. This page explains how to configure the automatic start behavior and how to start a suspended cluster manually when you need it available immediately.
How automatic start works
When Enable No-activity Suspension is set to Enable, the cluster monitors its default cluster endpoint and custom endpoints for incoming connections. If no connection arrives within the Detection Period for No-activity Suspension, the cluster enters the suspended state. When a connection arrives at one of those endpoints, the cluster starts immediately.
Startup sequence on resume:
The primary node starts first so the cluster can accept connections.
Read-only nodes are restored based on the Minimum Number of Read-only Nodes value.
Behavior on suspension: All read-only nodes are released when the cluster is suspended — regardless of the Minimum Number of Read-only Nodes setting — to reduce resource usage and prevent high availability (HA) compatibility issues.
Automatic start is only triggered by connections to the default cluster endpoint and custom endpoints. Connections through the primary endpoint do not trigger automatic start. For details on endpoint types, see Manage an endpoint.
Limitations
| Limitation | Detail |
|---|---|
| Endpoint scope | Automatic start only monitors the default cluster endpoint and custom endpoints — not the primary endpoint |
| Cluster specifications | Automatic start cannot be enabled on a serverless cluster with defined specifications |
Enable or disable automatic start
Log on to the PolarDB console.
In the top navigation bar, select the region where your cluster is deployed.
Find the cluster and click its ID.
On the Basic Information page, go to the Database Nodes section and click Serverless Configuration.
In the Configure Serverless-related Parameters dialog box, set Enable No-activity Suspension to Enable or Disable.
If you set Enable No-activity Suspension to Enable, set the Detection Period for No-activity Suspension.
Parameter Range Description Detection period for no-activity suspension 5 minutes–24 hours The idle period after which the cluster enters the suspended state Click OK.
Start a suspended cluster manually
If the cluster is in the suspended state and you need it available immediately — without waiting for an incoming connection — start it manually.
Log on to the PolarDB console.
In the top navigation bar, select the region where your cluster is deployed.
On the Clusters page, find the cluster in the Suspended state.
On the Basic Information page, go to the Database Nodes section and click Start Cluster.
In the Start Cluster message, click OK.
Troubleshooting
Automatic start is not working as expected
Check the following:
Endpoint type: Connections to the primary endpoint do not trigger automatic start. Make sure your application connects through the default cluster endpoint or a custom endpoint.
Cluster specifications: Automatic start cannot be enabled on a serverless cluster with defined specifications. Confirm that your cluster is running in fully serverless mode.