Enable Autoscaler
Autoscaling is a crucial feature in cloud computing and Kubernetes deployment. It helps organizations:
- Optimize resource usage
- Reduce operational costs
- Ensure high availability of applications
With Taikun CloudWorks’s autoscaling, you can set rules to automatically scale workers based on conditions like CPU utilization. For example, when CPU usage exceeds a certain threshold, new workers will be added to the cluster.
Note
The Min and Max values you configure define the lower and upper limits of the node count. The autoscaler scales nodes up or down based on unschedulable pods and cluster workload. Once the number of servers is above the minimum, it will not scale below it. For example, with Min=3 and Max=10, if the cluster scales up to 5 nodes and the workload decreases, the autoscaler will reduce nodes as needed but will always retain at least 3 nodes.
Important Changes and Operation
Autoscaler Group Name
As of this release (October 2025), the autoscaling group is permanently named taikunca (Taikun Cluster Autoscaler). If you were using it in previous versions, please disable and then re-enable the autoscaler to ensure correct functionality with the new configuration.
Viewing Autoscaler Logs
To see what the autoscaler is doing (scaling up, scaling down, or waiting), you can check its logs:
Failure Handling
If the autoscaler detects a server in a failed state, it will stop its operations and wait until the issue is resolved.
If a deploy fails (e.g., due to an OpenStack quota limit being exceeded), the autoscaler will wait and then retry the deployment. It will continue retrying for up to 3 hours.
Scaling Down Issues
If your autoscaler is not scaling down, check your pods. If you are using EmptyDir volumes, you must add the following annotation to the relevant pods to allow the autoscaler to safely remove the node:
Once a node is no longer needed, the autoscaler waits for 5–10 minutes before deciding to remove it.
Further Reading
For more detailed information on the Cluster Autoscaler, please refer to the official Kubernetes project FAQ: Kubernetes project FAQ
Enabling Autoscaler During Project Creation
Manager | Partner
The first way to enable autoscaling is during the project creation process.
Enable Autoscaler
Define Autoscaling Parameters
1. Minimum Workers Count - Define the minimum number of workers in the cluster.
2. Maximum Workers Count - Define the maximum number of workers in the cluster.
3. Disk Size (GB) - Set the disk size for the worker nodes.
4. Flavor - Choose a server flavor for the autoscaling instances.
Enabling Autoscaler in an Existing Project
User | Manager | Partner
You can also enable autoscaling after the project has been created.
Enable Autoscaler