|
From: | Sam Jansen |
Subject: | Re: [lwip-users] TCP transfer grinds to a halt... |
Date: | Wed, 20 Apr 2005 12:54:30 +1200 |
User-agent: | Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.3) Gecko/20040913 |
Eric wrote:
Well that looks bad. In the trace you attached, it looks like there is a whole heap of packet loss going on. To get a good idea of this you can try and measure a trace on both the sender and receiver sides. tcptrace (www.tcptrace.org) gives you a good graphical view on this.After about six meg of data transfer my TCP HTTP client grinds to a halt. I see some retransmits and dup'd acks before this happens, but they don't seem to cause a problem (although I'd understandably like to get rid of the retransmits as well). My source base is 1.1 updated with the latest commits. If you're interested, please look at my attached file.
In that trace, it looks like many of the data packets are being lost (then subsequently retransmitted) while as far as I can tell the ack packets are fine (there are often duplicate acks).
If you can, take a trace at the other side of the connection and try and verify if the packets are indeed being lost.
--Sam Jansen address@hidden Wand Network Research Group http://www.wand.net.nz/~stj2
[Prev in Thread] | Current Thread | [Next in Thread] |