Notes from discussion with the team: - The long return time is likely due to buffer bloat - measure buffer bloat with setup Possible solutions: - use ping times to determine timeout (send pings often) and make timeout adapt to measured RTT - *only send ping when we don't receive any data AND can't send data on the socket due to full buffer* I'll open a bug for tracking that the difference of creating the PING frame and it actually going out could cause too long measured RTT and makes timeout detection harder.
Bug 1868987 Comment 16 Edit History
Note: The actual edited comment in the bug view page will always show the original commenter’s name and original timestamp.
Notes from discussion with the team: - The long return time is likely due to buffer bloat - measure buffer bloat with setup Possible solutions: - use ping times to determine timeout (send pings often) and make timeout adapt to measured RTT - *only send ping when we don't receive any data AND can't send data on the socket due to full buffer* <- our approach I'll open a bug for tracking that the difference of creating the PING frame and it actually going out could cause too long measured RTT and makes timeout detection harder.