http://bugs.winehq.org/show_bug.cgi?id=30849
--- Comment #34 from ritual@vsritual.com 2012-06-06 18:20:43 CDT --- (In reply to comment #33)
The issue cleared up for me after this morning's maintenance. Tcpdump is telling me that Warden has been disabled for now (maybe the client patches incorporate permanent fixes, and Warden is more of a "quick fix" for blocking modified clients.)
When the battle.net connection is idle, I am getting keepalive packets at 120 second intervals. Based on that new knowledge:
- Warden scans (when Warden is enabled) are initiated at 20-second intervals
- The server will send a keepalive packet after 120 seconds of inactivity from
the client.
Thus in my original tcpdump, the packets at t=24, t=44, and t=64 were Warden scans, and the packet at t=164 was a keepalive. The server times out if the client doesn't respond to the keepalive with 30 seconds (120 + 30 = 150, which was my original observation.)
On repeated tests (before Warden was disabled), there were varying numbers of successful Warden scans before the failed one. There was probably a single signature in their list which was triggering our issue.
Troubleshooting will likely have to be put on hold until the issue recurs. In the meantime, I am building 1.5.5 on a 32-bit virtual machine, so that I have a build tree ready for deeper troubleshooting next time it happens.
I'm using POL on BackTrack 5R2 (Ubuntu 10.04 w/3.2.6-x86_64) on D3 1.0.2b. I cannot get past the retrieving hero list screen on the US Server. The CPU appears to spike to 100% when this happens.
The other realms don't appear to be patched yet so I'm unable to connect to those.