Cisco 7200 Duplex Issue

seanx

2[H]4U
Joined
Aug 1, 2001
Messages
3,191
So Ive got a 7200 with 2x 1 port fast ethernet cards in it that Ive been playing with.

Whats happening is when I plug the router into the switch the switch is setting the port to half duplex even though I have the port on the router set to full.

If I force the port on the switch to full duplex I get lots of late collisions on the port.

This is happening with both cards and with different cables.

I was wondering it there was some thing I was missing to force the port to full duplex other then specifying "duplex full" on the interface.

Thanks a bunch.
 
Well Ive let the switch auto negoiate but the router I cant seem to remove the duplex full or the duplex half line, it just doesnt go away when I do "no duplex full" or half.

Ive tried a couple different switches, none of them are cisco switches. Im useing an HP procurve.
 
by design,
when you end is hard-coded (say, 100full) and the other side is set to auto-negotiate, it will connect at 100/half duplex.

it can determine speed but not duplex...so it always falls back to the lowest common denominator (half duplex).

this is cause for issue in many environments (speed/duplex mismatches).... make sure BOTH ends are hard-coded

anytime you see an interface at 100/half, chances are / it's a good heads-up that one end is set to auto.
 
Yeah, exactly.

And the reason why it falls back to half duplex is because it doesn't receive anything back during the auto-negotiation process, so it thinks the link is one-way.
 
Well that is what I have been doing, forcing the switch into full mode, as well as the router and that is when I get all of the late collisions.

Also the "duplex auto" command is not valid.
 
Im going to try a different ios on it, currently it has some huge 44mb ios on there. Im sure I dont need everything that it has.
 
just leave everything in auto speed and duplex.... it'll make your life easier...
 
What IOS and NPE are you using?

EDGE(config)#int fa0/1
EDGE(config-if)#duplex auto

EDGE#sh ver
Cisco IOS Software, 7200 Software (C7200-ADVIPSERVICESK9-M), Version 12.4(22)T4, RELEASE SOFTWARE (fc2)

Looks valid to me...
 
LB-7200(config)#int fa4/0
LB-7200(config-if)#duplex auto
^
% Invalid input detected at '^' marker.

LB-7200#sh ver
Cisco IOS Software, 7200 Software (C7200-ADVENTERPRISEK9_SNA-M), Version 12.4(22)T, RELEASE SOFTWARE (fc1)

It has the NPE-225.

I just tried loading another ios on it and I get the same thing

Router#sh ver
Cisco IOS Software, 7200 Software (C7200-ADVIPSERVICESK9_LI-M), Version 12.4(11)T, RELEASE SOFTWARE (fc2)


Router(config-if)#duplex ?
full Force full duplex operation
half Force half-duplex operation

Router(config-if)#duplex auto
^
% Invalid input detected at '^' marker.

Router(config-if)#


And the interfaces still get detected as half on the switch even if the router is set to "duplex full"
 
Does the "duplex auto" command work on your fa0/0 and fa0/1 ports? You and I are using the same NPE and a similar IOS. I wonder why there's a difference.
 
Well my I/O card doesnt have any built in fast ethernet ports so I dont have any fa0/0 or fa0/1 interfaces. The router has 2 seperate single port FE cards and a FDDI-MM card but I dont have any fiber cables to use that.

This is hurting my head, I just want to get it to work.
 
I figured it out, finally!
Thanks to this post:
http://www.velocityreviews.com/forums/t31023-collisions-with-full-duplex.html

I had my cards in ports 3 and 4, I staggered them and now they are in 3 and 6 and no more errors after I forced all the ports to full duplex!

Thank you all for the suggestions, I was banging my head off the wall.

Edit going to quote the post that I found:
Well folks, it ended up NOT being the cable. The interface in slot 2
of the 7200 (Fa2/0) was apparently sharing the same backplane bus as
the interface in slot 1 (Fa1/0). As such, during times of heavy
network traffic, the bus was overwhelmed causing the errors. Moving
the module in slot 2 to slot 3 fixed the problem.
 
Last edited:
you clearly do not work in the field

duplex mismatching on auto is very 1990s

I have had no issues with modern gear for many many years now. The only time I see a "problem" is when some putz sets a server to full despite telling them to leave it in auto.
 
Well I will say that with out manually placing the switch into full it was not negotiating correctly so it is something to watch out for.
 
duplex mismatching on auto is very 1990s

I have had no issues with modern gear for many many years now. The only time I see a "problem" is when some putz sets a server to full despite telling them to leave it in auto.

well it must still 1995 because duplex mismatching is still an issue in modern ethernet networks.

let me rephrase - you clearly do not understand industry best practices.

you do not put your business critical servers/apps/whatever and take a gamble on auto-negotiate working flawlessly for the life of the server/network eq.

hard-code where you can for obvious reasons. it's one less thing to fubar.
 
I was bit by this issue recently. I was provisioning a customer with a metropolitan fiber connection (100Mbps) into our colo. At first they were getting horrible errors and video feed drops, so I went out and checked the switch (2948) that I installed, found port errors, and for some reason, errors went away when their modern unmanaged gig-e switch was set to 100/half. The next night, they had horrible performance (15Mbps). So I went out again, checked every switch, every port, the fiber, had a ticket open with the fiber guys (the previous day we replaced the whole fiber run from the provider to our gear due to a nasty light loss). Finally, my boss and I found the issue. About a month previous, I moved their stuff off of our big muti-vsys firewall onto a dedicated SSG. This worked fine when they had less bandwidth. We had the switch that this SSG was hooked up to set to 100/Full and both ports were logging serious errors (CRC's and I think runts). We switched the ports to Auto and the problems went away, they immediately went to full speed on their circuit. I also switched the internal switch ports to auto and all was well. Incidentally, on auto the ports were running 100/Full....

So after all that, duplex and speed issues on ports can drive you absolutely crazy to this very day. My boss says it's been like that for 30 years. I've been on the server and small business side since 95, ISP only recently.
 
this is kind of off topic (im just learning CCNA), but doesn't late collisions usually mean a weak connection (like a cable being run too far)?
 
this is kind of off topic (im just learning CCNA), but doesn't late collisions usually mean a weak connection (like a cable being run too far)?

Not necessarily a weak connection. You could have a bad NIC, but properly performing cables. Bad cables, too many hubs/repeaters, and, as you mentioned, cables being run too far can cause late collisions as well.
 
well it must still 1995 because duplex mismatching is still an issue in modern ethernet networks.

let me rephrase - you clearly do not understand industry best practices.

you do not put your business critical servers/apps/whatever and take a gamble on auto-negotiate working flawlessly for the life of the server/network eq.

hard-code where you can for obvious reasons. it's one less thing to fubar.

This was true for Fast Ethernet, but not Gigabit. You let Gigabit autonegotiate.

Anyway, although this thread has been solved, some of the older I/O boards for the 7200 routers did not autonegotiate at all. It wouldn't take the command "duplex auto". You had to hardcode it no matter what. It has nothing to do with the IOS revision.
 
Back
Top