Adjust timeouts and improve logging #53
Conversation
| @@ -145,6 +141,7 @@ def _poll_atproxy() -> None: | |||
| f"{query_url}/{request_to_handle.id}", | |||
| json=fulfillment, | |||
| auth=basic_auth, | |||
| timeout=(1, 2), | |||
There was a problem hiding this comment.
Those values seem low compared to others.
There was a problem hiding this comment.
This is intentional and hopefully correct :) Both calls to atproxy handler endpoints should have very little overhead -- just accessing shared memory and returning the result. The first one (query) is given a longer timeout (8 seconds) because it intentionally blocks for 5 seconds optimistically hoping that a request might show up on the queue. The fulfillment query (this one) merely writes to shared memory and returns, so therefore should never reasonably take longer than this (I think).
I'm not 100% sure this is correct though because there's obviously something I don't understand because #28 continues to be a problem, especially on my local machine where I can reproduce it pretty much every time.
In the effort to address #28, this PR improves logging and makes some timeouts more explicit and/or configurable, and shortens default timeouts to fail more quickly.