Manually change the status of instances

Instance states

Instances in a scaling group can be in the following states: Adding, Pending Add, In Service, Standby, Protected, Removing, Pending Remove, and Stopped. For more information, see Lifecycle states of instances in a scaling group.

The following table summarizes the manual state transitions available.

Instance category	From	Operation	To	API
ECS instances or elastic container instances (ECIs)	In Service	Put into Standby	Standby	EnterStandby
ECS instances or ECIs	Standby	Move out of Standby	In Service	ExitStandby
ECS instances or ECIs	In Service	Put into Protected	Protected	SetInstancesProtection
ECS instances or ECIs	Protected	Move out of Protected	In Service	SetInstancesProtection
ECS instances only	In Service, Standby, or Protected	Put into Stopped	Stopped	None
ECS instances or ECIs	In Service, Standby, or Protected	Remove or delete	Stateless	RemoveInstances or DetachInstances
ECS instances or ECIs	In Service, Standby, or Protected	Remove from scaling group (keep running)	Independent	DetachInstances
Managed instances	In Service	Remove from scaling group	Independent	None

An instance is temporarily set to Removing while being put into the Standby state or while being deleted or removed from a scaling group.

Put instances into the Standby state

Put an ECS instance or ECI into the Standby state when you need to update, troubleshoot, or reconfigure it without having Auto Scaling terminate it during a scale-in or health check.

How it works

The instance enters the Standby state and stays there until you manually move it out.
If a Server Load Balancer (SLB) instance is attached to the scaling group, the instance's backend server weight is set to 0, stopping new traffic from being routed to it.
Auto Scaling stops managing the instance lifecycle — health check status is not updated when the instance is stopped or restarted.
During scale-ins, Auto Scaling does not remove Standby instances.
Perform your maintenance: stop, restart, change the instance type, change the OS (system disk), reinitialize cloud disks, or migrate from classic network to virtual private cloud (VPC).
Move the instance out of the Standby state. Its SLB backend server weight is reset to 50, and Auto Scaling resumes health check monitoring.

To release a Standby instance, first remove it from the scaling group. If the scaling group manages the instance lifecycle, the instance is released after removal. Otherwise, it continues running independently. If you delete the scaling group, Standby instances are automatically removed and exit the Standby state.

Put instances into the Standby state

Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
Find the target scaling group and click View Details in the Actions column.
Click the Instance Management tab.
Select the instance list based on how the instances were added:
- For automatically created instances, find the Auto Created list.
- For manually added instances, find the Manually Added list.
Select the instances to put into the Standby state:
- For a single instance, find the instance and click Asynchronously Switch to Standby in the Actions column.
- For multiple instances, select them and click Switch to Standby State at the bottom of the page.

In the Switch to Standby State dialog box, choose the execution mode:

Mode	When to use	How to proceed
Synchronous	Small number of instances	Click OK.
Asynchronous	Large number of instances — the console returns immediately and processes the request in the background	Select Asynchronously Switch to Standby, then click OK. To check the result, go to the Scaling Rules and Activities tab and click Scaling Activities.

Move instances out of the Standby state

Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
Find the target scaling group and click View Details in the Actions column.
Click the Instance Management tab.
Select the instance list based on how the instances were added:
- For automatically created instances, find the Auto Created list.
- For manually added instances, find the Manually Added list.
Select the instances to move out of the Standby state:
- For a single instance, find the instance and click Asynchronously Remove from Standby in the Actions column.
- For multiple instances, select them and click Remove from Standby at the bottom of the page.

In the Remove from Standby dialog box, choose the execution mode:

Mode	When to use	How to proceed
Synchronous	Small number of instances	Click OK.
Asynchronous	Large number of instances	Select OK, then click OK. To check the result, go to the Scaling Rules and Activities tab and click Scaling Activities.

Put instances into the Protected state

Put an ECS instance or ECI into the Protected state to prevent Auto Scaling from removing it during scale-ins, while keeping it actively serving traffic.

How it works

In the Protected state:

The instance continues to serve traffic. Its SLB backend server weight remains unchanged.
Auto Scaling does not remove the instance during scale-ins.
Health check status is not updated when the instance is stopped or restarted.
The instance stays in the Protected state until you manually move it out.

What's blocked:

Automatic removal by Auto Scaling during scale-ins

What's still allowed:

Manual operations you initiate directly, such as stopping or restarting the instance
Removing the instance from the scaling group manually

To release a Protected instance, first remove it from the scaling group manually.

Put instances into the Protected state

Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
Find the target scaling group and click View Details in the Actions column.
Click the Instance Management tab.
Select the instance list based on how the instances were added:
- For automatically created instances, find the Auto Created list.
- For manually added instances, find the Manually Added list.
Select the instances to protect:
- For a single instance, find the instance and click Switch to Protected State in the Actions column.
- For multiple instances, select them and click Switch to Protected State at the bottom of the page.
Click OK.

Move instances out of the Protected state

Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
Find the target scaling group and click View Details in the Actions column.
Click the Instance Management tab.
Select the instance list based on how the instances were added:
- For automatically created instances, find the Auto Created list.
- For manually added instances, find the Manually Added list.
Select the instances to move out of the Protected state:
- For a single instance, find the instance and click Remove from Protected State in the Actions column.
- For multiple instances, select them and click Remove from Protected State at the bottom of the page.
Click OK.

Put instances into the Stopped state

Put an automatically created ECS instance into the Stopped state to stop it from serving traffic and avoid charges for released resources such as vCPUs, memory, and public IP addresses. Charges still apply for retained resources such as cloud disks and elastic IP addresses (EIPs).

For pay-as-you-go ECS instances, you do not need to manually enable the economical mode for the instance in the ECS console when you put the instance into the Stopped state.

During scale-outs, Auto Scaling preferentially starts Stopped instances before creating new ones.

Prerequisites

The ECS instances reside in a VPC.
The scaling group's Instance Reclaim Mode is set to Economical Mode.
The instances appear in the Auto Created list — only instances created by Auto Scaling are eligible.

Put instances into the Stopped state

Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
Find the target scaling group and click View Details in the Actions column.
Click the Instance Management tab.
Find the Auto Created list.
Find the target ECS instance and choose > Stop Instance in the Actions column.
Click OK.

Remove managed instances from a scaling group

If you do not use scaling groups to manage the lifecycles of managed instances, you must manually remove managed instances from scaling groups. You can remove managed instances from only scaling groups of the ECS type. Removed instances are not stopped or released.

Remove managed instances

Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
Find the target scaling group and click View Details in the Actions column.
Click the Instance Management tab.
Find the Managed Instance list.
Select the instances to remove:
- For a single instance, find the instance and click Remove from Scaling Group in the Actions column.
- For multiple instances, select them and click Remove from Scaling Group at the bottom of the list.
In the dialog box, configure the Modify Expected Number of Instances option and click OK:
- Selected (default): The expected number of instances decreases after you remove the instances from the scaling group.
- Cleared: The expected number of instances does not decrease after you remove the instances from the scaling group.