I was having trouble identifying exactly what users were doing when our application bombed on them. There were a number of written accounts of the steps. They all seemed related. So I tried asking for a reliable set of steps that always results in the problem.The response I got back was that the problem was happening at random. This was not getting me anywhere. So I called up an administrator at the site. I asked him if he could log into the application and make it happen. He did. But he had trouble communicating what exactly he was doing.
So I went over the situation again with him. I tried to slow him down on some steps that were unclear. Then the administrator recommended we use NetMeeting so I could see for myself what he was doing. Why didn't I think of that? At first I could not dial him up. But I gave him my IP address and he could dial me.
The administrator said he was set up so I could see full 32-bit color. But the screen refresh on my side of the NetMeeting just could not keep up with all the painting. So he turned down the color depth. It helped that I had him on a land line phone. I had to keep asking him to slow down until I could follow what he was doing.
Seeing is believing. Turns out that once I was watching over his shoulder, we could find a minimal set of steps to duplicate the problem. Once I had these, I was able to locate the source of the problem in an hour. Some meta data tables had incorrect data in their training environment. I gave the DBAs a set of SQL statements to regenerate these tables. Problem solved with an assist from Microsoft NetMeeting.
