Relay Server Failing with Mad Pings

Ok well sum of RTT is probably good enough for overall conn.

Maybe that mixed with a query of port 22070 for sessionLimitBps and globalLimitBps of the remote relay for a weigted average or some other grade to decide the relay to use. Or maybe just based on the number of clients already connected to it vs connected to other relays.

I may look into the code for this.

At the risk of repeating previous conversationsā€¦ An ideal solution needs to be resilient to misconfigured and malicious relays. This means that it canā€™t do things like check the relayā€™s rate limits: they might be wrong. A better option might be to open connections to several relays (but only at the point that data needs to be transferred), and let the faster ones transfer more traffic, bittorrent style.

99% of the time, devices either arenā€™t connected to anything over a relay, or are but arenā€™t transferring data. We donā€™t really care how much spare bandwidth the relay has in these cases, so long as communication is possible. Itā€™s only when we want to start transferring large indexes or files that speed starts to matter.

1 Like

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.