Auto Scaling:Suspend and resume scaling group processes

Suspend scaling group processes

Suspending a scaling group process means that you proactively halt a specific automated function of the group to perform other operations. For example, if you are debugging an application on an instance, you can suspend the scale-in process to prevent the instance from being unexpectedly terminated.

When you suspend scaling group processes, note the following:

• You can suspend one or more processes for a scaling group. Be aware that suspending one process may affect others.

  For example, if you suspend the scale-in process, health checks may still mark an instance as unhealthy, but Auto Scaling cannot remove it from the scaling group.

• Suspending processes provides process-level control. If you need instance-level control, consider using the Standby and Protected states.

  For example, you can put an instance into the Standby state before you troubleshoot or restart it, or put an instance into the Protected state to prevent it from being terminated. For more information, see Manually adjust instance states and Put an instance into the Protected state.

• The effects vary by process, as described in the following table:

  Scaling group process	Effect of suspension
  Scale-out	Suspending the scale-out process has the following effects: Manual requests to add instances are rejected. Requests to rebalance distribution are rejected. scale-out rules triggered manually, by a scheduled task, or by an event-triggered task are not executed if the expected number of instances feature is disabled. scale-out rules triggered by an event-triggered task are not executed if the expected number of instances feature is enabled. If the expected number of instances feature is enabled, scale-out rules triggered manually or by a scheduled task can still change the expected number of instances, but this does not change the actual number of instances. After the scale-out process is resumed, the check task for the expected number of instances can trigger a scale-out. The minimum instance count check task cannot automatically create instances. If the spot instance compensation feature is enabled, automatic creation of spot instances is rejected.
  Scale-in	Suspending the scale-in process has the following effects: Manual requests to remove instances are rejected. Requests to rebalance distribution are rejected. scale-in rules triggered manually, by a scheduled task, or by an event-triggered task are not executed if the expected number of instances feature is disabled. scale-in rules triggered by an event-triggered task are not executed if the expected number of instances feature is enabled. If the expected number of instances feature is enabled, scale-in rules triggered manually or by a scheduled task can still change the expected number of instances, but this does not change the actual number of instances. After the scale-in process is resumed, the check task for the expected number of instances can trigger a scale-in. The maximum instance count check task cannot automatically remove instances.
  Health Check	The health check for instances in the scaling group is suspended. Even if an instance becomes unhealthy, the system does not automatically remove it.
  Scheduled Tasks	When a scheduled task reaches its execution time, its associated scaling rule is not triggered.
  Event-triggered Tasks	When an event-triggered task enters the Alert state, its associated scaling rule is not triggered.

Resume scaling group processes

Resuming a scaling group process reactivates a suspended process, allowing the scaling group to execute it as defined. For example, after you resume the health check process, any instance that is considered unhealthy is automatically removed from the scaling group.

When you resume scaling group processes, note the following:

• After you resume a process, the scaling group may immediately respond to changes that occurred during the suspension period.

  For example, if the expected number of instances was increased while the scale-out process was suspended, no scale-out action was triggered. After the scale-out process is resumed, the check task for the expected number of instances runs and triggers a scale-out to reflect the increase.

• The effects of resumption vary by process, as described in the following table:

  Scaling group process	Effect of resumption
  Scale-out	The scaling group resumes scale-out actions, such as manually adding instances and checking for the expected and minimum number of instances. If the expected number of instances feature is enabled and this number increased during the suspension, the corresponding check triggers a scale-out to reflect the increase.
  Scale-in	The scaling group resumes scale-in actions, such as manually removing instances and checking for the expected and maximum number of instances. If the expected number of instances feature is enabled and this number decreased during the suspension, the corresponding check triggers a scale-in to reflect the decrease.
  Health Check	The scaling group resumes health checks and automatically removes unhealthy instances.
  Scheduled Tasks	If a scheduled task has not yet run or is still within its retry period, resuming the process triggers its associated scaling rule.
  Event-triggered Tasks	The associated scaling rule is again triggered when the task enters the Alert state.

Procedure

1. Log on to the Auto Scaling console.

2. In the navigation pane on the left, click Scaling Groups.

3. In the top navigation bar, select a region.

4. Find the desired scaling group and click Modify in the Actions column.

5. In the Edit Scaling Group dialog box, suspend or resume processes.

   • To suspend processes: Select one or more processes from the Suspend Process drop-down list.

   • To resume processes: Remove one or more processes from the Suspend Process drop-down list.

6. Click OK.

   If the scaling group has a low or medium-low elasticity score, a confirmation dialog box appears. You can perform one of the following operations based on the prompts:

   Important

   A low or medium-low elasticity score for a scaling group may cause scaling activities to fail. We recommend that you follow the prompts to modify the settings that cause the low score.

   • If you do not need to modify the configuration, click Continue.

   • If you want to modify the configuration immediately, click Back to Modify. After you modify the configuration, click OK again.

7. In the success confirmation dialog box, click OK.

Verify the result

On the Scaling Groups page, find the desired scaling group and click View Details in the Actions column. On the Basic Information tab, you can view the suspended and active processes for the scaling group in the Scaling Group Basic Information section.

Related documents

• Suspend specific processes in a scaling group: SuspendProcesses

• Resume suspended processes in a scaling group: ResumeProcesses