For the net_queue_trace thing, I'd like to find out if there is a correlation between your server hitches and the "threaded queued packet sent" spew. If you never see the spew, that helps narrow things down.
-NoQueuedPacketThread is actually different than net_queued_packet_thread 0. The former doesn't create a thread at all, whereas the latter creates a thread but leaves it idle. The bug we're reproing on our dual-core Linux boxes here is that a supposedly idle thread is not in fact idle.. It's eating lots of CPU that it shouldn't be. We're investigating that behavior right now. -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of karumba Sent: Thursday, July 05, 2007 2:35 PM To: [email protected] Subject: Re: [hlds_linux] Help! Alfred? ...Anyone?! thanks mike. i think thats what most of us wanted to hear. of course we will try to help. plz find some additional questions inline. > 1. Set net_queue_trace to 581304 and notice if/when the engine starts > spewing "xxxx - threaded queued packet sent" messages. with "notice if/when the engine starts spewing..." you mean after what time after enabling? > 3. Add -NoQueuedPacketThread to your command line. The engine will say > "Found -NoQueuedPacketThread" so you'll know it's doing the right thing. > Note that if you use this on the command line, then net_queue_trace > and net_queue_flush_interval won't be used. is this the same as "net_queue_flush_interval 0" + "net_queued_packet_thread 0"? this is a workaround & not intended for debugging, right? _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: http://list.valvesoftware.com/mailman/listinfo/hlds_linux
