The birth of a urban legend: Forcing network ports (to full duplex). : Remco's Sun blog

In the beginning of TP ethernet getting ports from different vendors to play nice together was sometimes a challenge, those days are over....... I guess in order to battle this problem some companies require the ethernet ports to be software wise "forced" to the highest speed and "full duplex", this sometimes seems to work but too often causes wierd connection problems. Now that most of these interoperatibility problems have gone the "old" rule is still continuing to cause, hard to find problems, every now and then. Nowadays, and for quite some time, autonegotiation works (and if not have the vendors fix it). I guess its now not more than an urban legend, IT style.

Comments:

Ah, but all legends are seeded with a grain of truth. In my case, none of our sun boxes with hme or eri interfaces will autonegotiate with our cisco catalyst switches. We have no problems with any other combination (other servers -> catalyst or sun box to another brand of switch) And this is with solaris 10 and a rather recent cisco ios release to boot. (the ce and bge gig interfaces work just fine)

Posted by john on November 28, 2005 at 02:19 PM CET #

Like most urban legends, this one may have had some truth in it in some past millenium. But it is about time it was laid to rest. The only autonegotiation failures I've seen recently are where the cisco end has been manually forced to 100 full in the mistaken belief that it might help. Doing so is guaranteed to make autonegotiation fail.

Posted by Peter Tribble on November 28, 2005 at 03:59 PM CET #

I agree with john. I dont know what kind of datacenter you run but we still have plenty of atleast eri/qfe interfaces around. But its slowly fading away..

Posted by Kenneth on November 28, 2005 at 04:28 PM CET #

Try this: Take a 420R with an hme interface on it and attach it to a cisco 6509 w/ a 48 port 10/100 blade (not the 10/100/1000 blades) and set both sides to auto. Watch the cisco negotiate to 100 full and the sun box to pick 100 half. (again, I have not seen this issue with the 10/100/1000 cards) There is the issue of forgetting that a port is forced to 100 full and a newer box picking 100 half (this is per the spec). The good news for me is that I hope to have all of my 420's out of here in the next year and replace them with x4100's and v490's. For an easy way to find mismatches, just watch the backup logs. Netbackup (and i'm sure other apps will too) will report on the backup rate. With our tape drives, we can expect 10MB/s. If we see a server reporting 1MB/s or less, we now know to look for duplex mismatches first since 90% of the time that is the problem .

Posted by john on November 28, 2005 at 07:37 PM CET #

Not an urban legend, never was. It's an ongoing, observed *fact*.

Autonegotiation has never worked for us on Cisco 29xx and 6509 switches, connecting Sun gear ranging from AXi's with hme's through v210's with bge to brand new x4100's with e1000g's.

Maybe you can afford to drop six figures a year to replace all your networking equipment, but we sure can't.

Posted by Anthony on December 10, 2007 at 11:34 PM CET #