If you want something wired right, sometimes you have to do it yourself.
This story started out with a new T1 connection being installed for a customer to connect to a new router. Before I arrived on location, it was understood that the T1 connection was ran by AT&T and ready to go. Upon arrival, I checked the networking closet where the new router was going. I looked for the usual biscuit jack to connect the T1 to the router. After being unable to find it, I called AT&T and had them tell me where it was. The smartjack was outside the building with no extension on the line to the inside. This required an electrician to come out and extend the T1 into the building.
After the electrician came out, we had the customer connect the cat5 cable into the biscuit jack and router’s T1 card. The electrician had already left by this time. The serial connection showed down and further tests concluded that AT&T was not seeing the router. At the other end of the T1 connection (point to point T1), it was showing looped when testing from the problem end. The electrician said that he saw “fire” on the line and refused to return saying that it was a problem with AT&T.
Later, AT&T returns back onsite to confirm that the smartjack is working and places a loopback on the biscuit jack. They concluded that there were some wiring problems with the biscuit jack and had fixed the problem. AT&T says they were able to loop back to their test equipment, but the router was still not coming up. At this point, we go to Cisco TAC and get them to send us a new T1 card to try thinking that the current one is faulty.
I receive the new T1 card and arrive onsite. I install the new T1 card and check the interface to find it still down. On the card, the alarm light was lit amber and the CD (carrier detect) was off. At this point, I’m thinking that the card isn’t the problem, but the problem is still with the wiring. The kicker here is that I put my loopback plug on the biscuit jack and verified that the other end’s router showed the circuit as up and looped. I put the original card back in the router and installed the new card into the other available slot so that there were now both T1 cards in the router.
I went outside and opened the box to the network enclosure. Instead of finding a nice simple smartjack, I find this: [more]
An RJ-48 connector from the smart jack feeds these other 4 connectors into the posts. There were two sets of 4 posts. The top set of four said T1, the bottom four said "Data". As a note, T1 only requires 4 copper wires to connect. Later examinations showed that only the T1 wires were even connected and nothing went to the bottom four pins for data (further wire tracing was necessary to determine this).
After one of our other network engineers and I began looking at the diagrams for the wiring enclosure and the biscuit jack pin outs, we could tell that something was still not quite right. I removed the biscuit jack which had been wired in as white/blue, blue, empty, white green, green, empty, brown, white/brown. We know that T1 should be using pinouts of 1,2,4,5. From the picture above, we can see that green and blue pairs are being connected to the T1 posts. The brown and orange wires aren’t doing anything here.
I cut off the biscuit jack and started putting RJ-45 plugs on the cable, and we tried switching the green and blue pairs around. This time when I plugged the cable into the router, the CD light lit up and the alarm went off. We could see the circuit as up, but it was going up and down constantly. We tried changing the wires around in different orders leaving the greens on 1 and 2, and blues on 4 and 5. After finding more diagram information on the enclosure and tracing wires, we came to a conclusion that we had it correctly set at White/green, green, empty, white/blue, blue, empty, empty, empty.
The circuit was still going up /down so I then decided to try moving the cable into the new card that had been put in the other slot on the router. This move fixed the circuit problem going up and down, and it stayed up.
The next challenge was testing pings across the line, and there were a few packets being dropped. We changed the clock rate source on the serial interface to be internal and this fixed the dropped packets.
Again, I wanted to point out that the loopback plug worked by showing the connection up, but it did not appear that the wires necessarily had to be in the correct order.