I seem to have run into a few problems with last week’s set of Microsoft updates on my server. I picked up the updates sometime in the middle of the week and applied them all as usual. I usually do this remotely via Terminal Server. However after the updates completed, next day when I logged on to the server again I noticed that the updates were still there…

It appears that you have to shut down the IIS Admin service (IISRESET /STOP will do it) before applying the update for the .NET Framework 2.0. Without this apparently the updates fail and roll back. Something in the rollback process also screwed up my machine in such a way that the Web Server was very frequently hanging or not processing connections properly. My West Wind Web Monitor tool was going crazy for the better part of last week notifying me of errors and restarting the Web Service.

So yesterday finally after checking in with my ISP I found out about the need to shut down IIS first then apply the update and sure enough this time around the update installed just fine. In the process I suppose I also picked up this weeks updates. In either case the problems with the immediately Web Server stopped and my phone is now calming down <g>.

Go figure… but this is lame as hell. Windows is making me reboot already but no I have to shut down IIS manually? C’mon… All the more lame that something didn’t roll back properly and hosed the machine.

Incidentally I was really scratching my head last week while these failures were happening as there was absolutely no reason for them.  I didn’t really think of a system problem at first of course. When in doubt – doubt yourself first, right? <s> Nothing had changed in some time on the server so I was really trying to figure out what was going wrong. All sorts of different applications and even static urls were timing out, but it wasn’t constant – the server would run well for a few hours and then all of a sudden just get stuck. My monitoring software automatically restarts the Web Service if it gets failures, but even with these frequent restarts the failures came right back. As it turns out this must have been some sort of network stack issue, but how the heck do you even begin to troubleshoot that.

My monitoring regime captures a lot of information about applications failing and then finally with West Wind Web Monitor making sure applications are returning the right thing and in the allotted time. However, when errors pop back to West Wind Web Monitor there’s usually very little information to check on – except usually the request timed out. There could be a million reasons  - database is slow, the server is getting hammered by search engines (seems to be more frequent these days).  It’s frustrating as heck to debug remote server problems.

I’m looking forward to IIS 7 where there will be a number of new interfaces that will help at least trace requests through the entire IIS server stack! That will be a huge improvement for debugging applications including finding issues with older ISAPI applications.

Anyway, I’m just glad things are back to normal. My phone’s constant blearing was getting to be a pain <s>… 2 days and counting without failures so I think I’m back to normal.