Temporarily suspend specific scaling group processes to prevent Auto Scaling from interfering with maintenance tasks such as troubleshooting, program debugging, or stress testing. After the tasks are complete, resume the suspended processes to restore normal scaling behavior.
When to use process suspension
Process suspension provides process-level control over scaling behavior. Processes apply across all instances in a scaling group. If you need to control a single instance — for example, to troubleshoot or restart it without Auto Scaling terminating it — put that instance into the Standby or Protected state instead of suspending a process.
| Goal | Recommended approach |
|---|---|
| Prevent all scale-out or scale-in activity during a maintenance window | Suspend the scale-out or scale-in process |
| Prevent health checks from removing an instance you are actively debugging | Suspend the health check process |
| Stop scheduled or event-triggered tasks from firing | Suspend the scheduled task or event-triggered task process |
| Troubleshoot or restart a specific instance without it being released | Put the instance into the Standby state |
| Prevent a specific instance from being released | Put the instance into the Protected state |
Effects of suspending scaling processes
You can suspend multiple processes across multiple scaling groups simultaneously. Suspending one process can affect the behavior of other processes in the same scaling group. For example, suspending the scale-in process prevents unhealthy instances from being removed even when the health check process is still active.
The following table describes what each suspended process blocks.
| Process | Blocked when suspended |
|---|---|
| Scale-out | Auto Scaling rejects: manually adding instances; rebalancing instance distribution across zones; running scale-out rules manually or via scheduled tasks and event-triggered tasks (when the expected number of instances feature is disabled); running scale-out rules via event-triggered tasks (when the expected number of instances feature is enabled); creating instances based on the minimum instance count; creating spot instances when the supplemental spot instance feature is enabled. Exception: if the expected number of instances feature is enabled, manual execution and scheduled task execution of scale-out rules are still accepted, but the instance count does not change. The expected number of instances feature triggers scale-out only after you resume the scale-out process. |
| Scale-in | Auto Scaling rejects: manually removing instances; rebalancing instance distribution across zones; running scale-in rules manually or via scheduled tasks and event-triggered tasks (when the expected number of instances feature is disabled); running scale-in rules via event-triggered tasks (when the expected number of instances feature is enabled); removing instances based on the maximum instance count. Exception: if the expected number of instances feature is enabled, manual execution and scheduled task execution of scale-in rules are still accepted, but the instance count does not change. The expected number of instances feature triggers scale-in only after you resume the scale-in process. |
| Health check | Auto Scaling stops running health check tasks and does not remove unhealthy instances. |
| Scheduled task | Auto Scaling does not trigger scaling rules specified in scheduled tasks at the configured times. |
| Event-triggered task | Auto Scaling does not trigger scaling rules when event-triggered tasks enter the Alert state. |
Effects of resuming scaling processes
When you resume a suspended process, it immediately resumes its normal function. Be aware of changes that accumulated during the suspension — resuming a process can trigger an immediate burst of scaling activity to reconcile those changes.
Auto Scaling allows you to resume multiple processes in multiple scaling groups at the same time.
The following table describes what each process does when resumed.
| Process | Behavior when resumed |
|---|---|
| Scale-out | Resumes scale-out operations, including manually adding instances, checking the expected number of instances, and checking the minimum instance count. If the expected number of instances feature is enabled and the expected count increased during the suspension, Auto Scaling immediately triggers a scale-out to reach the new expected number. |
| Scale-in | Resumes scale-in operations, including manually removing instances, checking the expected number of instances, and checking the maximum instance count. If the expected number of instances feature is enabled and the expected count decreased during the suspension, Auto Scaling immediately triggers a scale-in to reach the new expected number. |
| Health check | Resumes health checks and immediately removes any instances identified as unhealthy. |
| Scheduled task | Triggers the execution of scaling rules specified in scheduled tasks, even before the specified points in time or during the retry interval. |
| Event-triggered task | Triggers scaling rules when event-triggered tasks enter the Alert state. |
Resuming a process after a long suspension can cause a burst of scaling activity. For example, if the expected number of instances increased while the scale-out process was suspended, Auto Scaling launches instances immediately after you resume the process to close the gap. Review the state of your scaling group before resuming to anticipate the impact.
Suspend or resume processes
-
Log on to the Auto Scaling console.
Log on to the Auto Scaling console.
In the left-side navigation pane, click Scaling Groups.
In the top navigation bar, select a region.
On the Scaling Groups page, find the target scaling group and click Edit in the Actions column.
In the Edit Scaling Group dialog box, suspend or resume processes:
Suspend: Select one or more processes from the Suspend Process drop-down list.
Resume: Remove one or more processes from the Suspend Process text box.
Click OK. If your scaling group has weak scaling strength, a confirmation message appears.
To proceed without changes, click Continue.
To adjust the configuration, click Back to Modify, update the parameters, and click OK.
ImportantWeak scaling strength means scaling activities in the group may fail. To prevent failures, click Back to Modify and increase the scaling strength before proceeding.
In the confirmation message, click Close.
Verify the result
On the Scaling Groups page, click Details in the Actions column for the scaling group. In the Scaling Group Basic Information section of the Basic Information tab, the suspended and active processes are listed.
What's next
Use the API to automate process suspension and resumption during maintenance windows: