Reposting here from the 0.3.23 release thread. I have issues where one of my clients, solely relying on one of my other clients on the same LAN, get disconnected due to flood control. I don't recall seeing this in 0.3.21.
I'm having issues connecting my "internal" client (on one computer) with the one I run on another computer that's port mapped and listed on the IRC channels. The internal client does not go on IRC, isn't port mapped and the config file adds the other client on the local network using addnode. This worked well with at least 0.3.21, never really tried 0.3.22. With 0.3.23 it seems the connection gets dropped easily, and once it's been dropped it doesn't seem to come back up. Even when it's initally up (1 connection) there are very few blocks being received, and very slowly, though.
From the internal client log, although it doesn't seem to be detailed enough:
trying connection 192.168.0.123:8333 lastseen=-335573.5hrs lasttry=-363351.2hrs
sending: version (85 bytes)
disconnecting node 192.168.0.123:8333
(Both computers are Macs)
Is there a way to prioritize clients on the same local network? I'd like for my "external" client to feed blocks to the "internal" one as quickly as possible.
Might've found the culprit. This is from the "external" client's log:
sending: block (23423 bytes)
sending: inv (37 bytes)
socket send flood control disconnect (10030645 bytes)
disconnecting node 192.168.0.132:64860
If flood control is something recently added then I guess that would be where my problem comes from. It'd be great to be able to turn that off for clients on the same local network at least if not the possibility for a full whitelist.