I have a catalyst 3750 with one port that is trunked with 802.1q trunk. The trunk has 2 vlans running on it. There is no un-tagged traffic running accross this connection. I see FCS errors, collisions, and runts. I have the latest firmware. On the other end of the connection, is an extreme networks alpine switch. It has 2 protocol(IP) based vlans on it. Between the switches is 2 media converters and a piece of fiber. I have replaced all items between the switches and it has not fixed the problem.
I think you might have a duplex mismatch issue here, or maybe you're just running the connection in half duplex. Because if it's a switch-to-switch connection, then the link should be full duplex; and if you are running full duplex, then you should not be seeing collisions, runts, etc.
Are you connecting the Cisco and the Extreme switches via 10/100Base-TX ports and media converters? Are either or both ports set to autodetect speed and duplex? Try setting both ports manually to 100 full. Also, make sure your media converters are configured to match the duplex setting. Whose media converters are you using?
What's the distance of the fiber? And is it multimode (MMF) or single mode fiber (SMF)?
On MMF, Fast Ethernet is only good to 412m max. when running in half duplex; that distance extends to 2,000m max. when running in full duplex. On SMF, full duplex is the preferred operating mode for most media converters; max. distances are limited by the wavelength and optical power of the fiber optic transceiver. Some media converters can be run to distances of 80-100km or more over good clean fiber.
I have both ends hard coded at 100mbs full duplex. The 3750 connects to a single mode fiber converter. We have about 9 miles of fiber then another media converter. then ethernet from the converter to the extreme switch. we had the converters replaced and had the fiber checked. The media converters are Radiant Technologies brand.
to the best of my knowledge, the 3750 switch does not support tagging of the native vlan on a dot1q trunk. also, why is the switch reporting collisions ?? Shouldn't the link be running at full duplex ?
Hard code the duplex on both sides to full duplex. When using media convertors i normally hard code the speed and duplex on both sides.
Check the speed and duplex and also make sure that the alpine switch is not tagging traffic on the native vlan ( if it supports the feature to tag dot1q native vlan traffic )
I do not know why it has collisions. It has fcs errors, collisions, allignment errors, and too small packet errors which are the most common. The duplex is hard coded on both ends. I do not wish to have the native vlan tagged. Just the 2 other vlans on it should be tagged. I have tried lots of different ideas to make the problem go away with no luck yet. We flashed the image, tried making a common vlan on both ends for the native vlan even though we do not wish to pass any un-tagged traffic, we repaced the media converters, had the fiber checked and we have 2 support calls open with cisco and one with extreme networks. I have asked several local techs also. I am open to ideas.
We keep coming back to the duplex issue here. Can you tell us the exact model number of the media converter you are using? Need to check the documentation. Are they plain old media converters, or do they have bridging capability? Are they configurable by the user (via DIP switches, jumpers, slide or toggle switches, etc.)?
I have run many 100-meg links over Transition Networks media converters, with no problems. I know they can either autodetect duplex on the copper side, or can be set manually to half or full. Maybe your Radiant ones need configuration, too. Have you used that model on any other links before?
Sounds like you're focused on the switches. But my gut feeling on this is that it's not them. I've interconnected multiple switch vendors (Cisco, Extreme, 3Com, Foundry, HP, Nortel, as well as lesser-known brands) using media converters from several vendors (but not Radiant), and the config issues on the switches are the same whether they're in the same room or separated by miles of fiber. If they work in the same room before deployment, then they should work when deployed. If they don't, the only thing that's changed is the media connection between them.
Any chance you can get the 3750 into the same room as the Alpine, so that you can use a single Fast Ethernet cross-over cable to connect them directly? This would eliminate the fiber and the media converters from the equation. Configure the ports identical to how you want them set for use over the fiber, but eliminate the fiber etc. as a variable. Then test it.
If it works _without_ the fiber and media converters, then you know where the problem is. If it doesn't work, well, then you know where the problem is, too. Either way, it's progress.
They are plain old media converters. I do not have documentation for them because we are leasing our fiber and the company we are leasing it from provided both media converters. They are model DL221a/sm-1u Radiant technologies media converters. They were both replaced by the company that we are leasing from to rule them out. I do not know if they are hard coded with their link speed and they do have dip switches on the front. Strange thing is if we hookup alpines on either end, we see no errors. we used to have a nortel switch on this end and the alpine on the other and we did not have errors in the past using the same media converters. That was until that box died. we had planned to put in the 3750 for the gig capabilies. the alpines are only 10/100 capable. The 3750 is 10/100/1000. I can sure see if i can dig up some documentation on the radiant converters. Our current game plan is to configure another alpine with a different set of IP's and make a trunked port on the cisco and hook them up in the same room and see if we get the same errors. As we are hooked up now we are definately connecting and running we just see the errors count up on the receive side of the cisco and the alpine. we are going to set that up today. Thanks so much for your help. Any progress is better then what I have had for the last 4 weeks trying to get this figured out.
Looks like the DL221A is a 10/100 bridging media converter, according to the spec sheets at www.rccfiber.com. Always runs at 100 full on the fiber side, but can do 10 or 100, and half or full, on the copper side.
The ability to go down to 10, and to also do half duplex on the copper side, is what makes this a bridging media converter versus a simple media converter. There's extra circuitry and RAM on board that allows the unit to buffer traffic between the two interfaces, in the event that the copper feed is not 100 full. Think of this device as a two-port switch, with one port copper and the other fiber.
The picture on the spec sheet (www.rccfiber.com/PDF/200%20Series/dl221_221a.pdf) shows DIP switches, along with a push-button to change the copper port from straight-through to 10/100 Ethernet cross-over. Those DIP switches probably need to be set to tell the unit how to run that copper port.
I bet the problem is, that the DL221A unit defaults to 10/100 autodetect. Once you manually set the Cisco or Extreme switch ports to 100 full, they no longer negotiate. So the Radiant converter is probably defaulting to a half duplex mode on the copper side. Which is resulting in your Ethernet errors.
Try setting the Cisco switchport back to "speed auto" and "duplex auto", then hook it up to the DL221A. Then, check the Cisco switch port to see what it negotiated the link to. If "show interface status" says it's connected at a-100, a-full, then you've isolated the problem.
If you want to then force the switch ports to 100 full, call Radiant Communications Corporation
at 800-969-3427 (they're in New Jersey, EST time zone). Ask for tech support, tell them you want to know how to set the DIP switches.
Thanks konigl! I think that fixed my problem. I contacted Radiant and they gave me the settings. The collisions and FCS errrors went away. I am still seeing too small packet errors but the cisco techs assured me that they are nothing to worry about. If the media converters would have been ours we would have known that they could have been set to different settings for the speed and duplex. We really didn't look too hard into that as we set each of our switches to 100 meg full duplex. THe other thing that led us away from finding the problem is that the last switch we had ran fine as it was setup. It was on auto sense for the speed and duplex. Thanks!!