Back in the late 80's I was working on an X.25 gateway. We had it installed at a Wall Street firm. It had a problem where it would accumulate thousands of CRC e

Back in the late 80's I was working on an X.25 gateway. We had it installed at a Wall Street firm. It had a problem where it would accumulate thousands of CRC errors in just a few minutes. We knew the problem was on the customer premises because we had loaded the gateway down using the same test suite. We ran them through everything. Their point was when they plugged the scope on both sides of the connection and loaded the link with a loop back test, they had no problem. I get on a plane. We have the scopes hooked up on both sides and the guys in the data center are on the phone talking. We load up the test suite and no errors occurred. Scratching heads, we watched it not fail for a half hour. The consensus was we needed to regroup and come back at the problem from a different angle. The data center guys agreed. Then, suddenly on our scope, we got like 500 CRC errors... I said to the DC guys "what did you just do? We're getting CRC errors" "Nothing" "You didn't touch anything?" "No we just unplugged the scope from the patch panel" "Try plugging it back in please" And the CRC errors stopped. Faulty patch panel diagnosed. The firm probably paid $25k in consulting fees for that diagnosis. But for me, the lesson that my tools might alter what I am observing was priceless.