Hello team,

The customer would like to use a Tekton pipeline-based automation. On OpenShift there is clusterautoscaler enabled. The customer is experiencing an issue that the pods are not running properly when a larger load is deployed on the cluster.

Suggestion: Setting on the clusterautoscaler to specify eg. in percentage the threshold to fire up new nodes.  

Proposed title of this feature request
Custom setting for percentage-based setting on cluster autoscaler

What are the nature and description of the request?
The customer would like to use a Tekton pipeline-based automation. On OpenShift there is clusterautoscaler enabled. 
The customer is experiencing an issue that the pods are not running properly when a larger load is deployed on the cluster. The autoscaler seems to be a little bit slow and we need to improve this behavior. Percentage and/or load-based prescaling.

Why does the customer need this? (List the business requirements here)
These tests are part of the development process.

How would the customer like to achieve this? (List the functional requirements here)
There is Helm chart-based Pipelineruns definition. In these generated manifests the request and the limits are the same. So the pods are running in guaranteed mode. These parallel pipeline runs generate circa ~ 400 pods. We have a limited autoscaling configuration with 0 -8 machines (each VMs has the same type. 16cpu && 64GB ram)

For each functional requirement listed in question 4, specify how Red Hat and the customer can test to confirm the requirement is successfully implemented.
OCP version is OCP 4.9.33, We have a limited autoscaling configuration with 0 -8 machines (each VMs has the same type. 16cpu && 64GB ram). This is a dedicated region which also part of the worker MCP. We are working in dedicated NS where we defined the node selector. You need to create a minimum of 10 parallel pipeline runs definition, which pulls the config central RWX storage, and it also generates I/O intensive tasks. Each pipeline runs have a dedicated pod which do the requested preparation and each PR=(pipelineruns), start 35 UI check pods that have 8 containers. (Tekton pipeline)