Since about a week and a half ago my server is jo-joing up and down frequently apparently due to IIS 6 Application Pool crashes. The problem started (in hindsight) right after installation of the maintenance patches from two weeks ago the week before Win2003 SP2 was released and installed (didn't notice the problem until later actually).
The symptoms are that AppPools crash and crash hard - they just die and hang for a bit before another starts up and takes over. What's wonky about this is that nothing shows up in the Event Log. Yet the Visual Studio debugging service dialog pops up (I had VS installed for some previous server debugging). So I actually stepped into one of these debugging dialogs and looked at the call stack hoping that it's not one of my own components or ISAPI extensions that are the trigger. Thankfully not.
Looking at the call stack it looks like the culprit is classic ASP making a COM call followed by about 100 levels of ntdll internal function calls. There's of course no further indication of what the problem is except that it's crashing deep inside the Windows kernel.
The crashes are frequent - it was about once every 10 minutes or so. I've been trying to see a pattern by cruising my site and seeing if I can make it crash but I haven't had any luck in 'making it happen' on demand.
I have a suspicion it has something to do with a small COM component I run on most of my 'mostly static' ASP pages of the site. These pages have a counter and logging routine that go through a small efficient ATL COM component. This code has been running on the server going on 10 years now and has never been an issue but that's the only place where my ASP pages call into COM.
Has anybody else noticed failures along these lines or anything similar?
It's been quite troublesome for me because a) it's hammering my monitoring tools with site down messages and b) it's causing havoc with various applications that are expecting to stay alive all the time and are instead firing up new resources everytime the AppPool restarts. It's not cool to find various service interfaces running 50 times because they didn't get shut down properly when the AppPool crashed.
What's worse I really have no idea where to even look. I've started to move around my application pools a bit to try and isolate the problem and right now I seem to have narrowed things down a little better - the crashes now occur only once or twice a day but there are still occasional spurts when the up and down happens 10 times in a row or so. <sigh> I suspect rolling back is going to be a problem since SP 2 was installed after that original patch.
Other Posts you might also like