Error 500.13 (Server too busy) and Event 4689

[jn00s]

I am running a windows 2003 server with close to 150 websites. The majority of the sites, which are all running the same very complex classic ASP portal application using ASPImage/ASPMail/ASPUpload, are receiving the vast majority of the traffic and all are working fine. However all of my smaller websites that do not use this application or any of those 3rd party COM DLL's, and only get a tiny bit of traffic across just a few simple ASP scripts, all fall into a state of perpetual 500.13 errors after about an hour. I have had to put IISRESET onto an hourly scheduled task to keep them from being down too much, but this adversely affects all my users because they're getting Service Unavailable errors ever hour. The main clue I've found so far is that the Application log is completely full of Event 4689 errors that are happening at a rate of sometimes several per second. The text of this eventlog entry is always the same:

The run-time environment has detected an inconsistency in its internal state. This indicates a potential instability in the process that could be caused by the custom components running in the COM+ application, the components they make use of, or other factors. Error in d:\nt\com\complus\src\comsvcs\threads\stathread.cpp(285), hr = 8007000e: CSTAThread: CoGetApartmentID failed
 

I've been scouring the web for weeks and the only solution I found so far was to completely uninstall/reinstall IIS and DCOM, and I'm not really convinced that this step would solve the problem as both of my webservers do the same thing.

One other thing that seems like it's probably relevant (but may not be at all), is the way in which I set up my websites. I use an ASP script to build or rebuild my websites in IIS. It's based on code put out by microsoft in their IIS admin scripts, I just simplified it into functions that I could call from within my scripts. Originally it was built to work with Windows 2000 Server so it would just set the app to isolated on each site and that worked great, and I never had any 500.13 erros. For Win2003 I modified the script so that it creates an app pool for each website as it builds or rebuilds the website. If the app pool already exists, it seems to still work fine: either because another duplicate pool is not created, or because the first one is deleted when the script deletes the existing website before replacing it (I'm not sure which). HOWEVER in my Component Services control panel, I see that duplicate COM+ applications are being added with each successive rebuild of each site. Again I'm not sure if this really SHOULD affect anything, and I didnt even add the part of the script that created separate app pools until AFTER this problem started happening and I wanted to isolate out the sites causing the problem.

[jn00s]

More info I forgot to mention:

NONE of the sites that are having problems are built or rebuilt by my scripts, only the group of sites that is not having problems gets built or rebuilt in this manner. The ones that are having problems I set up by hand using the IIS snap-in and I manually set them each up with their own app pools, yet they all seem to fail at the same time.

Also this event continues to happen even during and after an IISRESET, while all
of the sites are still running fine.

[jn00s]

bump... still have not found any solution to this problem... any help would be greatly appreciated!

[markrawlingson]

There's a similar post here

That ended up having a solution.

Wow,I think I fixed it :

It's the GOD DAMN COM+ System Application service. It was turned off.
Turn that back on and ym comp will be okay.

These resources..

http://support.microsoft.com/defaul...kb;EN-US;290513

http://support.microsoft.com/defaul...&NoWebContent=1

http://www.google.com/search?hl=en&...G=Google+Search

Were also given and may help you.

Other than that... if it's just a few specific sites there has got to be a common denomitator between those sites.

Would it be possible to backup and move one of the affected sites to a completely different location so that you can just place some simple code like a for.. next loop to display "hello" on a page 20 times or so - and see if it crashes?

Do each of the affected sites share a specific application pool, or do they have their own?

Also, you mentioned that you are running 150 web sites. How many session/application level variables are in use and how many hits in total do the sites aquire per day?

Does the site crash x times an hour, every hour, no matter what time of day?

And you are DEFINATELY closing all open objects on all scripts within all of the websites before the page stops execution?

Do these pages use ANY objects at all? Like an recordset object or connection object?

Answer those questions anywho.. but it sounds to me like the websites that are experiencing the problems are trying to create an object and simply don't have enough resources to do so.

[jn00s]

I finally found the fix to this problem, solution posted here:

Error 500.13 (Server too busy) with many sites on one machine : The Official Microsoft IIS Site