Still investigating "too many requests" errors from Slack, and I realized that when we build the channel info to run an action, we get user info for every member, but were using Future.sequence (aka "Future.allAtOnce"). Switched to FutureSequencer.
For channels with many members, this is likely to mean that actions take significantly longer to start running. Not sure we can avoid this unless we cache user info for a longer period of time.
Still investigating "too many requests" errors from Slack, and I realized that when we build the channel info to run an action, we get user info for every member, but were using Future.sequence (aka "Future.allAtOnce"). Switched to FutureSequencer.
For channels with many members, this is likely to mean that actions take significantly longer to start running. Not sure we can avoid this unless we cache user info for a longer period of time.