Studying a SQL Server Connection Failure

[Shaggy Hiker]

Unfortunately, I don't have enough data to say definitively that it IS intermittent. There are two problems. The first is that the testing itself is intermittent, the second is that too many variables change between each test to be certain that they are comparable. For example, this WAS working the last time I tried it, but a whole lot happened between then and now...including the months of August, September, October, and November, but also I changed pretty nearly ALL the hardware on the desktop.

This issue of too many variable changing exists between any pair of failed and working test. The earlier failures were using a different server, and sometimes got resolved by simply rebooting. When it worked, it would work for a long time. For example, when it was last working and I was looking at network traffic using WireShark, it worked for days. I moved on to other things and there were no issues.

It hasn't been good testing, either. Partly, I mostly only sought solutions when it failed, which might be once a year. One should change only one variable at a time, and I wasn't being good about that. In this case, I tried to be methodical. I'd make a change, run the test, make a new change, run the test again, and so forth. Then, once it was working, I'd undo a step and see whether that changed things. That's how I decided that both applications (SQL Server and SQL Server Browser) had to be allowed through. I wasn't always so methodical, or I ended up chasing other issues at the same time. For instance, I was having an issue on one server and not a different one, but at the same time, I realized that I ultimately had to stop using the first server for other reasons, so I was highly motivated to switch over to the new server. That meant that I switched hardware without fully understanding why one worked and the other did not.

It may well be a red herring. It isn't the total solution, either, because of ping still isn't working. I also haven't compared network traffic now to the saved traffic patterns I have from the earlier success. I still have to do that to see if it is the same. Lots of chaff in those WireShark dumps, though, so it's a bit slow to get through, and I won't be able to get back to it until Sunday, at this rate.

I'd say there is something NIC related going on, probably due in some way to the new hardware being used. I would say I have learned something, but not quite enough, yet.