Safely Shut Down and Restart an Oracle RAC Cluster Without Service Disruption

Background

Directly executing shutdown -h now on the operating system to stop a RAC (Real Application Clusters) database can cause many services to remain in an UNKNOWN state after the OS restarts, leading to instability. A more controlled, "gentle" shutdown and startup sequence is recommended.

Step 1: Verify Service Status

Use the crs_stat command to check whether the RAC node services are running normally.

Step 2: Shut Down Cluster Services

Execute srvctl (service control) commands to stop services in the following order:

Close the database instances.

Shut down the ASM (Automatic Storage Management) instances.

Stop the node services.

Shutdown sequence (illustrated):

关闭数据库(实例) → 关闭ASM实例 → 关闭节点服务

01 – Shut Down Database

02 – Shut Down ASM Instances on Each Node

03 – Shut Down Services on Each Node

Step 3: Start Cluster Services

Use srvctl to start services in the reverse order:

Start node services.

Start ASM instances.

Start database instances.

Startup sequence (illustrated):

启动节点服务 → 启动ASM实例 → 启动数据库实例

01 – Start Services on Each Node

02 – Start ASM Instances

03 – Start Database and Instances

Additional Step: Manual Failover Service Start

If any failover services were not automatically started, start them manually using the appropriate srvctl commands.