In our organization we have an KubeHpaMaxedOut alert that fires if a pod's number of replicas is equal to maxReplicas for more than 15 minutes. This alert is pretty much constantly firing for our bentos, because we typically don't explicitly set the runner's autoscaling parameters and so they get the default values of 1 for both minReplicas and maxReplicas.
I would like to propose to change the default value of maxReplicas to 2 for runner pods. What do you think?
In our organization we have an KubeHpaMaxedOut alert that fires if a pod's number of replicas is equal to
maxReplicasfor more than 15 minutes. This alert is pretty much constantly firing for our bentos, because we typically don't explicitly set the runner's autoscaling parameters and so they get the default values of 1 for bothminReplicasandmaxReplicas.I would like to propose to change the default value of
maxReplicasto 2 for runner pods. What do you think?