introduce heartbeat check to fallback host support
aMahanna opened this issue · 1 comments
aMahanna commented
We are currently processing the client request to a given host, without knowing whether it is alive or not.
- If the host happens to be down, this can lead to a long wait before the
ConnectionAbortedErroris raised
(see current implementation of process_request)
We should instead heartbeat the host first, and if it is alive, process the request.
- This can be simply achieved by running a
requests.get()on the host selected by ourHostResolverwith a settimeoutparameter.
aMahanna commented
No longer applicable, not a valid solution to the the issue at hand. Will come back & search for an alternative..