I am running ping.exe from a Windows 2003 Server machine through FreeBSD
4.8 running ipfw, to another Windows server.  With plr=0, I get no ping
failures at all, and with ipfw at its defaults I get no successes, so
I'm pretty sure the network is good.

I configured ipfw with various packet loss rates (call it X) in one
direction, _zero loss_ in the other direction.  Then I ran ping for
several hundred cycles.  This should be sending an ICMP echo request
(with X prob. of loss), and an echo reply (which should always get
through).  Therefore, I would expect to see a ping failure probability
of X.

I tried this at plr's from 0.002 up to 0.47, and did the math:  ping is
failing exactly (allowing for statistical variation) 1.500 times more
than expected.  This is too unlikely to be chance, so I figure the nice
round fraction of additional failures is due to something I don't yet
understand.  Can anyone explain this result?  Thanks.

No luck searching the Internet or the FreeBSD/search sites.
FreeBSD rd_bsd 4.8-RELEASE FreeBSD 4.8-RELEASE #0: Thu Oct  2 23:31:46
CDT 2003
    [EMAIL PROTECTED]:/usr/obj/usr/src/sys/MYKERNEL  i386

Grant D. Schultz
Senior Software Engineer

5828 Reeds Road
Mission, KS 66202

freebsd-questions@freebsd.org mailing list
To unsubscribe, send any mail to "[EMAIL PROTECTED]"

Reply via email to