Open rajesh-blueshift opened 6 months ago
I opened a PR to send the PING every 1 hour as per the recommendation by the APNs docs. Nothing though that there may be other factors causing connection resets (e.g. APNs closing the connection or network problems) so you should still use retries.
Makes sense. Thanks for helping on this one. I appreciate it
Connection resets have stopped now. Thanks for a quick fix
Dear team,
We have deployed our push service using a2 into production, but we are frequently observing
ConnectionReset
errors on http2 connections.Our service is multi-tenant and the traffic is a steady stream of messages with idle periods and bursts in between. Although we are working around these connection resets by retrying it. I was thinking if we could do something like sending periodic keep-alive messages over the open connection.
The pushy library already does something like this: https://github.com/jchambers/pushy/pull/266
I read through the hyper documentation, and the default is to not send any keep-alives