Openwhisk Invoker turning unhealthy

I have extended invokers replica count to 3 & deployed it as a statefulset on OSP Platform.
Openwhisk-controller logs shows invoker turning unhealthy while invoking action.
When i check for resource utilization, its utilising less than half of the allocated resources. So i think resource is not an issue, but still all the invokers keep turning unhealthy/down/unresponsive for few seconds when i run the jmeter test.
I cant find why its turning unhealthy repeatedly & it spins up another instance of test-health-action pod.
Also, is there any possible way to loadbalance the invocations between the invokers externally?

Start by reproducing the issue with the reported three-invoker StatefulSet and JMeter test, then inspect the OpenWhisk-controller logs around invokers becoming unhealthy and test-health-action pods being recreated. Determine why invokers become unresponsive despite low resource utilization, and document whether external load balancing of invocations is supported.

Written by the indexing model from the issue text.