Hi, is it possible to have the connection pool pooling connections by host instead of db? i.e. it will change db upon reuse.
I'm asking because I have a lot of shards (databases) on every MySQL host and if I create a connection pool size of N, and the host contains M shards, it will create N*M connections to MySQL. Now that I use ProxySQL to do connection pooling, the actual number of connections to MySQL is lowered, but the number of connections to ProxySQL is still N*M, causing high CPU on ProxySQL due to system call poll and sleep.
Thanks!