Driver pods take a long time to run but do not require a large amount of resources. Due to this they have been seen as holding up large nodes (c6a.48xlarge) from being released since we cannot interrupt the pods.
SOLUTION
This PR sets all Driver pods to be forced to schedule on t3a.2xlarge nodes which should allow us to pack a few driver pods on them and not worry about the cost.
PROBLEM
Driver pods take a long time to run but do not require a large amount of resources. Due to this they have been seen as holding up large nodes (c6a.48xlarge) from being released since we cannot interrupt the pods.
SOLUTION
This PR sets all Driver pods to be forced to schedule on t3a.2xlarge nodes which should allow us to pack a few driver pods on them and not worry about the cost.