Closed pborzenkov closed 4 months ago
Make sure the count is only cleared once the client get a successful request or encounters a non-IO error. Otherwise it never finishes.
Also, since the upper layer usually retries on its own, limit the numer of retries to just 10.
This should be enough in case of a simple server restart, but will provide faster reaction times if the host is down completely.
Make sure the count is only cleared once the client get a successful request or encounters a non-IO error. Otherwise it never finishes.
Also, since the upper layer usually retries on its own, limit the numer of retries to just 10.
This should be enough in case of a simple server restart, but will provide faster reaction times if the host is down completely.