Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 22 Oct 2002 12:59:53 +0100 (BST)
From:      Bruce Dixie <bruce@uk.clara.net>
To:        questions@freebsd.org
Subject:   Intel Pro card killing LAN
Message-ID:  <20021022123512.D92105-100000@insane.noc.clara.net>

next in thread | raw e-mail | index | archive | help
Hi

We've got a cluster of 15 news servers sending out large volumes of
traffic (80-90Mb/s per box). The boxes are of similar spec, but do differ
a bit in terms of hardware - also they are running various versions of
FreeBSD ranging from 4.3-RELEASE to 4.7-STABLE. Some of the boxes have
Intel Pro cards inside.  After working perfectly for months, one of the
boxes suddenly started spewing out garbage onto the ethernet, which
stopped the ethernet cards working on other machines with the same Intel
card.  When the switch port for the offending box was admin'ed down the
other boxes recovered straight away and carried on as if nothing had
happened. Initially we thought it was a once off, but it has since
re-occured and on the last occasion we managed to get a TCP dump of the
traffic on one of the boxes that wasn't affected. Looking at this, the
only odd traffic we could see were:

Frame 6 (60 on wire, 60 captured)
    Arrival Time: Oct 18, 2002 21:44:38.038653000
    Time delta from previous packet: 0.000632000 seconds
    Time relative to first packet: 0.002630000 seconds
    Frame Number: 6
    Packet Length: 60 bytes
    Capture Length: 60 bytes
Ethernet II
    Destination: 01:80:c2:00:00:01 (01:80:c2:00:00:01)
    Source: 00:02:b3:b4:40:3d (Intel_b4:40:3d)
    Type: Unknown (0x8808)
Data (46 bytes)

0000  00 01 01 1f 00 00 00 00 00 00 00 00 00 00 00 00   ................
0010  00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00   ................
0020  00 00 00 00 00 00 00 00 00 00 00 00 00 00         ..............

There were lots and lots of these packets. Googling for the MAC address
we've found that it's the multicast address reserved for IEEE Std. 802.3x
Full Duplex PAUSE operation. Theoretically the switch is not supposed to
pass these on to other ports but ours does (we're raising this with
Cisco).

What I'd like to know is what is causing these packets to be broadcast, as
the server has been much more loaded and nothing happens then. Also it
only happens to the one server - and we've replaced the network card and
installed the latest STABLE.

Unfortunatly the box is down and at our remote colo site so I can't give
you a dump of dmesg.

Any help would be greatly appreciated :-)

Regards

Bruce

--

Bruce Dixie


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?20021022123512.D92105-100000>