Using the NorthScale client with a cluster comprised of 2 nodes. Our logs indicate very frequent "ConnectionAborted" socket errors, which mark the pools as dead. The NorthScale console shows all nodes as healthy though, and does not display any errors in the log.
When using NorthScale, shouldn't the client trust the cluster to maintain healthy pools rather than try to identify the dead ones?
Closed by f265b92 where the webclient stopped reading from the socket after a while (5 mins or so).