Mar 26

Things the previous staff did before you

I have gone into many different locations to find really odd things people have done with network gear that really surprise me. Some times there are valid reasons why things are done the way they are, other times, I think it’s just poor choices.

One facility I went into didn’t have the racks mounted to the floor. The person used Velcro to secure the rack to the rack next to it.

One of my first tasks at a new company was to find a particular switch. The entire networking staff had left and everybody was new and they couldn’t find this specific switch. After using the ARP and MAC table, I found the port it was connected to; then started tracing the cable. After pulling up multiple floor tiles in the data center, I found it. It was on the floor under the raised floor in the data center.

I was sent to a facility to troubleshoot some wireless issues. I was told that all of the AP’s inside the facility were Cisco 1252’s. I found multiple models of old Cisco AP’s with multiple external Omni directional antenna’s. After asking some questions, I found out that all of the old wireless equipment was Motorola Symble equipment…. Nobody knew how the old Cisco AP’s got there????

I went into a my wife’s office building and they had a network rack on wheels. The rack had a clear plastic door on the front so I could see all of the network gear in it. I thought this was pretty nice to show off their gear to the people. The best part was that there was no cable management and cables went everywhere. I wouldn’t want to be the guy that had to clean up that mess.

VLAN Naming, I try to name the VLAN based off of the purpose of the VLAN. In one business I worked for, the previous staff named the VLAN by color, so we had the PINK, BLACK, and BLUE VLAN. We never did figure out what the Pink VLAN was for.

What odd, strange or stupid things have you run across and had to fix?

Mar 12

Test points when troubleshooting a fiber optic connection on the LAN

Last week I talked about some of the basic tools that you would need once you started using fiber optic connections on your LAN. This week I am going explain how to test an optical connection from a server to the network switch.

There are many different scenarios and methods of fiber connections in the LAN. This diagram assumes the server is connected to a patch panel in the same rack the server is, then structured cabling to a patch panel close to the switch, then connects to the switch. This design takes three different duplex jumpers. When I use the term Duplex Jumper, I am describing two strands of fiber combined to be a single jumper. I state this because you can purchase a single strand jumper. In this scenario you could use two separate single strand jumpers, one in each direction. In total, there are six strands of fiber, twelve connectors, and two optical modules that need to be tested.

The symptom in this scenario is that the server will not connect to the switch. All ports have been enabled, but still unable to get a connection.
Here are the steps I would take to troubleshoot this connection issue.

1. Swap the two fiber jumpers at the switch (or server). One of them should be sending light to the switch. It’s possible that it’s connected to the transmitter instead of the receiver. The connectors on the duplex jumper are usually connected together. If they are connected, you may need to gently remove them from the connector and then swap the connectors.
2. If swapping the fibers did not resolve the issue, I would get my light meter to look for the fiber stand with the light.
a. It really does not matter what side of the link you start at, in this scenario, I’m going to start at the server, Test Point #1.
b. Pull the fiber out of the server and look for light on either jumper. If you have a light level within the specifications of your optic that optical module may be bad.
c. Repeat this same process at the switch (Test Point #6). If you have good light levels at the switch, one of your optical modules may have a bad receiver.

3. If the light levels are low, or not there at all you need to start eliminating segments of the fiber run. For this scenario, we have good light at the server from the switch, but low light at the switch from the server
a. I like to start at the source. Start at Test Point #1, I would use a known good fiber jumper to connect my light meter to the optical transmitter and get a light level reading. If there is low light, then your optical module may be bad. If you have a good light level, continue testing.
b. Reconnect the original jumper to the server. Disconnect the jumper from the patch panel and test the light from there (Test Point #2). If the light is low, use your fiber scope to see if either side of the jumper is damaged or dirty. If the connector is dirty, use your fiber cleaner to clean it and then use the scope again to make sure it’s clean. If the connectors are clean with no damage to them, replace the jumper.
c. If the light levels are good, repeat this process all of the way through Test Point #6. At each Test Point, make sure all connectors are firmly connected and clean.

At some point you are going to find something disconnected, loose, dirty, broken or not even connected to the proper device. After you locate and fix the failure check the server connection to the switch, it’s possible there are multiple failures. Keep testing until you resolve all of the failures until your connection comes up.

Clearly your connection may be different than the one presented, but the same methodology should be used to track down the failure. I have had new optics and jumpers that don’t work and the vendors have always been good about replacing the failed jumper with a new one. This method has worked well for me over the years that I have been working with fiber connections; I hope it works well for you too.

Does this match your troubleshooting steps?
Did you find this to be helpful?

** To find the optical level that is acceptable for your optic you need to find the document with the specifications for your specific model of optic. I expect every model to be different, so I look up each model.**