internet/awx-operator
3
0
Fork 0
mirror of https://github.com/ansible/awx-operator.git synced 2026-03-26 21:33:14 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add support for horizontal pod autoscaling (#1676)

Browse Source
This commit is contained in:
David Hageman
2024-06-03 14:59:48 -05:00
committed by GitHub
parent 6820981dd5
commit 8ead140541
5 changed files with 47 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
docs/user-guide/advanced-configuration/horizontal-pod-autoscaler.md Normal file
View File

@@ -0,0 +1,27 @@
### Horizontal Pod Autoscaler (HPA)
Horizontal Pod Autoscaler allows Kubernetes to scale the number of replicas of
deployments in response to configured metrics.
This feature conflicts with the operators ability to manage the number of static
replicas to create for each deployment.
The use of the settings below will tell the operator to not manage the replicas
field on the identified deployments even if a replicas count has been set for those
properties in the operator resource.
| Name | Description | Default |
| -----------------------| ----------------------------------------- | ------- |
| web_manage_replicas | Indicates operator should control the | true |
| | replicas count for the web deployment. | |
| | | |
| task_manage_replicas | Indicates operator should control the | true |
| | replicas count for the task deployment. | |
#### Recommended Settings for HPA
Please see the Kubernetes documentation on how to configure the horizontal pod
autoscaler.
The values for optimal HPA are cluster and need specific so general guidelines
are not available at this time.

11
docs/user-guide/advanced-configuration/scaling-the-web-and-task-pods-independently.md
View File

@@ -1,8 +1,13 @@
#### Scaling the Web and Task Pods independently
#### Scaling the Web and Task Pods independently
You can scale replicas up or down for each deployment by using the `web_replicas` or `task_replicas` respectively. You can scale all pods across both deployments by using `replicas` as well. The logic behind these CRD keys acts as such:
- If you specify the `replicas` field, the key passed will scale both the `web` and `task` replicas to the same number.
- If you specify the `replicas` field, the key passed will scale both the `web` and `task` replicas to the same number.
- If `web_replicas` or `task_replicas` is ever passed, it will override the existing `replicas` field on the specific deployment with the new key value.
These new replicas can be constrained in a similar manner to previous single deployments by appending the particular deployment name in front of the constraint used. More about those new constraints can be found in the [Assigning AWX pods to specific nodes](./assigning-awx-pods-to-specific-nodes.md) page.
These new replicas can be constrained in a similar manner to previous single deployments by appending the particular deployment name in front of the constraint used. More about those new constraints can be found in the [Assigning AWX pods to specific nodes](./assigning-awx-pods-to-specific-nodes.md) page.
##### Horizontal Pod Autoscaling
The operator is capable of working with Kubernete's HPA capabilities. See [Horizontal Pod Autoscaler](./horizontal-pod-autoscaler.md)
documentation for more information.