On Thu, Aug 02, 2001 at 12:52:15PM +0200, BERND.STURM@xxxxxxxxxxxx wrote:
> We did packet captures with tcpdump and saw that the sender wouldn´t open up
> its tcp send window wide enough and every once in a while stall with its
> sending.
> This effect didn´t occur with 2.2.16 but all the kernel settings in
> /proc/sys/net were the same !
2.4 has an additional global TCP memory limit; see the tcp_mem entry in
/usr/src/linux/Documentation/networking/ip-sysctl.txt for details.
It also has more sysctls to tune the TCP buffer management, like
tcp_app_win or tcp_adv_win_scale.
> It seems, as strange as this may sound, that the 2.4.x-kernel-series has
> some kind of congestion window memory, that still remembers the congestion
> window of a former tcp-transfer, at least it acts this way. But this would
> in no way be conforming to the current tcp-standards !
2.4 saves cwnds and some other information in the destination cache, similar
to many other stacks (e.g. Solaris). You can get rid of this information
by flushing the routing cache (echo 1 > /proc/sys/net/ipv4/route/flush)
> By the way, what´s this no_cong, lo_cong, mod_cong-stuff in
> /proc/sys/net/core ??
They define how the stack drops packets on overload.
-Andi
|