Shut down an ECS instance's OS and services; billing impact varies by billing method and stop mode.
Stopping an instance interrupts all services. Ensure your workloads can tolerate the interruption before proceeding.
Billing impact
Billing method | Impact after stop |
Subscription | Billing continues. Expiration date is not affected. |
Pay-as-you-go (standard mode) | Billing continues until the instance is released. See Release an instance. |
Pay-as-you-go (economical mode) | Billing stops for vCPUs, memory, images, and pay-by-bandwidth public bandwidth of the static public IP address (if any). Billing continues for system disks, data disks, and pay-by-bandwidth public bandwidth of the EIP (if any). |
Spot instance | The instance may not restart if resources are insufficient or the bid price falls below market price; it is then reclaimed. See Spot instance overview. |
Stop an instance in the ECS console
Go to ECS console - Instances.
In the top navigation bar, select a region and resource group.
Click the instance ID to open its details page.
Click All Actions in the upper-right corner, then search for and click Stop.
Configure the stop parameters and click Confirm.
Parameter | Option | Description |
Stopped By | Stop | Gracefully shuts down the instance through the OS. |
Force Stop | Immediately powers off the instance, equivalent to pulling the power cord. May cause data loss if the OS has not finished writing to disk. | |
Stop Mode | Standard Mode | Retains all resources. Billing continues. |
Economical Mode (formerly known as No Fees for Stopped Instances) (pay-as-you-go VPC instances only) | Releases vCPUs, memory, and the public IP address. The private IP address is retained. See Economical mode risks. |
The instance enters the Stopped state after the operation completes.
Economical mode risks
Economical mode releases compute resources to stop billing, which introduces two risks:
Restart failure. The instance may fail to restart due to insufficient inventory or overdue payments. If inventory is insufficient, try again later or change the instance type. For overdue payments, add funds to your account. See Economical mode risks and Overdue payments.
Public IP address change. If the instance has a static public IP address, it is released on stop. A new, potentially different, public IP address is assigned on restart.
Stop an instance through API
Use the following API operations to stop instances. See Integration overview.
Operation | Description |
Stops a single ECS instance in the Running state. | |
Stops one or more ECS instances in the Running state. |
FAQ
What if an instance is stuck in the Stopping state?
An instance can remain in the Stopping state for two reasons:
Normal stop in progress. The OS is releasing processes, installing patches, or responding slowly due to limited resources. If processes cannot release, a forced restart is automatically triggered after approximately 10 to 20 minutes.
Abnormal stop. An OS error prevents the stop from starting.
To diagnose the issue:
View instance screenshots to check OS stop progress. See View system logs and screenshots.
If screenshots are inconclusive, connect through VNC to observe the OS directly. See Connect to an instance using VNC.
Take action based on what you find:
Normal stop: Wait for the OS to finish.
Abnormal stop: Troubleshoot the OS issue through VNC, or force stop the instance.
A force stop is equivalent to a physical power-off and may cause data loss if data has not been fully written to disk.