ColdFusion randomly stops rendering

[Aegis_Kleais]

So, this is an issue we've been dealing with indefinitely.  CF10 on a Win2008 environment just stops working.  I remote into the server, the ColdFusion services are using 0% CPU, usually eating about 600-850MB of memory, and no pages are being delivered.  Simply a white page.  Browser (Chrome 30) times out eventually reporting an 'ERR_UNEXPECTED' error.

Restarting World Wide Web Service has no effect.  Restarting IIS sites/server (nodes) has no effect.  ONLY restarting the ColdFusion Application Server service results in services getting restored.  And it takes normal amount of time to STOP the service, but starting it up takes a long time (though it eventually starts, and ColdFusion is rendered again)

I don't have troubleshooting skills necessary to know what log files to watch or a background in networking know-how, but I'd really love to start taking steps to finding out what the heck is doing this.  So if you guys can suggest any steps, I'll gladly take them.  The last time this happened for us was around 10:00AM-10:30AM on January 9, 2014 EST.

[vishu_13]

Hi

Test the below steps during development

Go to C:\ColdFusion10\cfusion\runtime\conf and take a back up of server.xml. Open server.xml and enable the built in port : 8500 like this

	<!-- internal webserver start -->
	<Connector executor="tomcatThreadPool" maxThreads="50"
	port="8500" protocol="org.apache.coyote.http11.Http11Protocol"
	connectionTimeout="20000"
	redirectPort="8445" />
	<!-- internal webserver end -->

Restart CF 10 service and then place few test cfm files in CF 10 root (C:\ColdFusion10\cfusion\wwwroot)

Open URL : http://localhost:8500/a.cfm  and observe that CF is rendering the pages or not. If yes, then CF 10 is working fine.

Now you need to test when a request is coming to webserver, is your webserver forwarding the request to CF 10 or not. You can create a new test website in your webserver and configure that website with CF 10. Place few .cfm files and then observe the behavior. While you launch the webserver connector tool to add this test website, click on Advanced and check "Enable Verbose logging for connector"

NOTE: Enabling verbose is not all recommended in production environment.

This way you will be able to capture logs and troubleshoot accordingly. You can also check the Failed Request Tracing logs of IIS to find the root cause of this issue.

If you still encounter the issue, you can share the stack trace here.

Also, if this is a dev environment and you think that you have messed up some settings without backup then go ahead and give a shot to reinstall CF 10

HTH 

Thanks

VJ

[Aegis_Kleais]

Vishu.  Thanks for the recommendations.  This is a production environment, but it's an Intranet site.  Though I'd love for the site to tout a 100% uptime guarantee, that's not critical at the moment, especially if I can go so far as to troubleshoot and determine the cause of this issue.

As for what you recommended, you're suggesting that when the issue occurs again, attempt to view these sample CF files on the :8500 port to see if the issue is IIS-related vs. CF-related, no?  If it renders, CF is working fine, and this might be an issue with IIS instead, but if CF is not executing on the :8500 when the problem presents itself, then the problem may be more CF-related than IIS.

[vishu_13]

What if you create a new site in IIS and that starts working fine? If yes, then the difference between the existing site and the new site will be of the " Application Pool". Try using a new App pool for one of your existing sites and observe the behavior.

Though I'm sure that you must have figured out the root cause and fixed it as I myself has learnt alot from you

Cheers!!!

Thanks

VJ