This PR seeks to avoid unnecessarily creating duplicate Redis clients per published channel. We discovered the issue at scale at EchelonFit. Warning this PR does need to be extensively tested - which is why it's marked experimental. Thanks to @rjschineller for discovering the issue.
This PR seeks to avoid unnecessarily creating duplicate Redis clients per published channel. We discovered the issue at scale at EchelonFit. Warning this PR does need to be extensively tested - which is why it's marked experimental. Thanks to @rjschineller for discovering the issue.