Re: [Openh323gk-developer] Status port - printactivecalls cannotlist more than 327 calls
H.323 Gatekeeper for VoIP and videconferencing
Brought to you by:
willamowius
From: Jan W. <ja...@wi...> - 2006-10-03 20:43:54
|
Hi, I looked into this today with GnuGk on a Linux server and status clients on Linux and Windows. I noticed that output gets stuck a lot earlier if you are using LARGE_FDSET, but it also happens with the PWlib socket implementation. You can actually see the TCP timeout error in a level 4 trace Unfortunately the number of bytes sent before the error happens seems to be reported inacurately (by both implementations). That is probably the reason why the FlushData method isn't working and commented out since Chih-Wei wrote it in 2003. So, simply resending the missing data doesn't work. Splitting large messages into 10K chunks isn't enough either. We still run into the same timeout after about the same amount of data. The only solution I found so far is to sleep 250 msec after each 10KB chunk. This isn't really great, but the impact is limited since most messages are much shorter. I'll put this into the CVS shortly, but if somebody has a better idea, i'm all ears. Regards, Jan -- Jan Willamowius, ja...@wi..., http://www.gnugk.org/ |