Hi, I'm investigating some latency issues we've observed in our AS deployment, where it appears that some connections are slower than the other ones.
Would it be possible to do a test where connection pool is modified to use round-robin to select a connection from a pool for those connections that fall under minConnsPerNode, and for all connections provisioned above minConnsPerNode continue to use stack logic so that maxSocketIdle continues to work as expected? Maybe we can modify 9.0.5 client, before dynamic client configuration was added, since it would be easier to modify that.
Hi, I'm investigating some latency issues we've observed in our AS deployment, where it appears that some connections are slower than the other ones.
Would it be possible to do a test where connection pool is modified to use round-robin to select a connection from a pool for those connections that fall under
minConnsPerNode, and for all connections provisioned aboveminConnsPerNodecontinue to use stack logic so thatmaxSocketIdlecontinues to work as expected? Maybe we can modify 9.0.5 client, before dynamic client configuration was added, since it would be easier to modify that.