Seems a lot of clients will timeout, before sending anything.
I.e. after completing TCP handshake, they will sit idle, hogging resources, and kiryuu will send HTTP 408 after 5000ms (for now).
We can decrease this timeout to something like 1000ms? Earlier thought of 100ms , but it is easily possible the RTT is > 100ms, e.g. here it is ~300ms:
Additionally, we can add a post-request hook in actix, where it will report HTTP 408 to some external service (TBD where). Could be compilation option. These may be malicious and can be firewalled off to save resources.
Seems a lot of clients will timeout, before sending anything.
I.e. after completing TCP handshake, they will sit idle, hogging resources, and kiryuu will send HTTP 408 after 5000ms (for now).
We can decrease this timeout to something like 1000ms? Earlier thought of 100ms , but it is easily possible the RTT is > 100ms, e.g. here it is ~300ms:
Additionally, we can add a post-request hook in actix, where it will report HTTP 408 to some external service (TBD where). Could be compilation option. These may be malicious and can be firewalled off to save resources.