Date: Fri, 13 Nov 1998 00:24:13 -0600 (CST) From: Chris Dillon <cdillon@wolves.k12.mo.us> To: Dave Bodenstab <imdave@mcs.net> Cc: questions@FreeBSD.ORG Subject: Re: Are collisions normal on a local net Message-ID: <Pine.BSF.4.05.9811122353210.16362-100000@duey.hs.wolves.k12.mo.us> In-Reply-To: <199811130354.VAA11666@base486.home.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 12 Nov 1998, Dave Bodenstab wrote: > If I rcp a big file between the two ne2000 boxes, the collision led on > the hub flashed occasionally, but if I do the same between one of the > ne2000 boxes and the intel etherexpress, the collision led is on > much more often. > > Is this normal or do I have something wrong with my setup? Could it be > something with the fxp driver or hardware trying to go at 100M instead of > 10M? I notice that the fxp driver can be ifconfig'ed with a media type > and options... I tried -media 10baseT/UTP but it didn't seem to affect > anything. I noticed something similar on a small 10mbit network I set up for someone. I have an Intel EtherExpress Pro/100B in a P200 box running a FreeBSD 2.2.6-BETA (a few days before 2.2.6 came out... has an uptime of about 210 days now, been running ever since I set it up <grin>) and Samba that talks to various Win95 clients ranging from 486DX/33s to 486DX4/100s all with 3COM EtherLink III NICs in them. When transferring data between any of the clients and the server, collision rates are absolutely horrible, and traffic flow oscillates between a fair amount of real data passing (500-700kb/sec) to so many collisions that no data passes at all (according to netstat -w1). The collision indicator on the hub is lit solid when any transfers are occurring. I recently added a K6-233 client with a Pro/100B in it and now get a consistent 800-950KB/sec between it and the server with not nearly as many collisions as with the other clients. The only explanation I can think of at this time is there is some kind of odd timing difference between the Pro/100B and some other types of NICs, just bad enough to cause storms of collisions. On a similar note, I have Pro/100B's in all of our servers at work, and never experience collision problems like this. Of course, we're in a heavily switched environment. :-) Well, I do have one setup in a (underpriveledged, technolgy-wise) classroom with a Pro/100B in a P133 server running (ugh) NT talking to clients with cheap NE2000 clones and Intel EtherExpress ISA NICs, and get collision storms with some of the clients, but not others (some do/don't among identical clients, even). Go figure -- gotta love CSMA/CD. -- Chris Dillon - cdillon@wolves.k12.mo.us - cdillon@inter-linc.net /* FreeBSD: The fastest and most stable server OS on the planet. For Intel x86 and compatibles (SPARC and Alpha under development) ( http://www.freebsd.org ) */ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.05.9811122353210.16362-100000>