This may have been related to the service interruptions occurring on our host. After they applied their MDS / ZombieLoad patches and rebooted the box things have been running considerably better.
Closing this out as the server code has recently gone through a major rewrite. We will need to investigate this further after it is better tested and deployed.
This may have been related to the service interruptions occurring on our host. After they applied their MDS / ZombieLoad patches and rebooted the box things have been running considerably better.