I run a set of MvM servers on Debian Wheezy x64, and have been experiencing a problem with net buffer overflows partway into long rounds of 30+ waves after the SteamPipe update. I recently discovered via SPUF that other server owners running comparable popfiles have also been having this problem, which gives random users net buffer overflows during the round, and keeps them from re-entering the server.
Unfortunately, despite using every available remedy (increasing rates on both client and server, trying to cap currency via plugins, etc), we're all still receiving this issue, so I'd like to hear if there's any way to help the devs debug this. There are no unusual log entries to report, even with -debug, as the users disconnected just show up as "Disconnect by user". I'd be happy to provide Wireshark packet captures, but I suspect that the packets involved are encrypted and wouldn't do much good without the relevant keys. The client's log merely shows these two entries: Disconnect: Buffer overflow in net message. Host_EndGame: Buffer overflow in net message This is a huge problem for our servers, and on my end, many users are beginning to leave, as the only way to "fix" it is to restart the server process or change map, both of which cause a disappointing loss of progress into a long mission. The popfiles we use are here on my github, if needed: https://github.com/N3X15/failtrain-30wave-popfiles The engineers do spawn and do their thing on our end, since we use an add-on to spawn the relevant hint entities, but it should work fine beyond that on any other server. Thanks in advance. _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: https://list.valvesoftware.com/cgi-bin/mailman/listinfo/hlds_linux
