When there's too many connection_timeouts the monitor starts killing nodes without testing. Probably related with the GenStage consumer recursively trying to fetch the data from the available nodes. On restarting the server, the monitor refill the nodes list and they seem fine.
When there's too many connection_timeouts the monitor starts killing nodes without testing. Probably related with the GenStage consumer recursively trying to fetch the data from the available nodes. On restarting the server, the monitor refill the nodes list and they seem fine.