Hi, I dont know if this question is appropriate for this mailing list. If not will someone please guide me to the appropriate list ? Thanks. I am working on tcp modifications. while testing the tcp behavior in the unmodified kernel (2.4.20 and 2.4.24) i noticed that it is not even coming out of slowstart, and the only window reduction is through congestion window validation. more digging using web100 showed that it was mostly receive limited and SndLimTimeRwin is close to the duration of the ftp transfer. (I tried with both ftp and sftp clients). I went over several auto tuning guides and have currently set the following sysctl variables /proc/sys/net/ipv4/tcp_wmem 4096 65536 8388608 /proc/sys/net/ipv4/tcp_rmem 4096 87380 8388608 /proc/sys/net/core/wmem_max 8388608 /proc/sys/net/core/rmem_max 8388608 And I am using a regular(not gigE) lan - one of my NICS is just 10Mbps - and still, there are no congestion drops ! I have tried these experiments before to clearly see the window adaptation in the congestion avoidance phase ! I am using the same kernel image as before... so its not due to any code changes....I even tried using the nistnet emulator between the sender and receiver to see if the small rtt was causing the problem... but it doesnt look like that is the problem. I have not changed any hardware since the last set of experiments. I also built the same kernel on another machine (on the same test network) and tried.... i still get similar results !!!! Can someone please tell me what other things could lead to a receiver limited window ??? Can a NIC driver get corrupted and result in something like this ? I am at my wits end...Any help is greatly appreciated.... Thanks in advance, Sumitha - : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
