CF service stops at same time every week

[davidw52895080]

Hi all,

Recent issue that has started with the server.  Every weekend at roughly the same time the coldfusion server service just stops.  I have nothing in the CF logs to tell me as to why it just stops.  Windows event logs simply says :

Event error 7024 The ColdFusion 11 Application Server service terminated with the following service-specific error:

The system cannot find the file specified.

If you go in and restart the service it starts back with no issues.  I have the latest and greatest patch (installed the patch to see if that corrected the issue), but no good.

Running on Windows server 2012.

[carl type3]

Hello,

Perhaps some reason for CF11 stopping will be in logs. Check before CF start messages for error or warnings in

ColdFusion11\cfusion\logs\ coldfusion-error.log or coldfusion-out.log. Windows 2012 so perhaps using IIS. If so there might also be a clue for CF11 outage in ColdFusion11\config\wsconfig\1 isapi_redirect.log.

HTH, Carl.

[BKBK]

You should do some research on the error message "The system cannot find the file specified". It seems like a Windows error.

[Charlie Arehart]

David, did you ever resolve your issue? Did you find anything in the logs as others suggested?

If you did not, it is possible for CF to hangup in a way that would lead to nothing being logged (in CF or the Windows Event log you looked at). If requests start to hang, such as if they are waiting a response from something else (like a DB server, or a cfhttp call to another server), then there will be nothing logged. The requests just hang.

And once there are more requests running than your value for "maximum simultaneous requests" in the CF admin "request tuning" page, then any other requests would be queued. And while there is a default timeout of how long requests can wait in that queue (60 seconds), again there may be nothing logged in CF to indicate that this is happening.

And to be clear, it doesn't matter if you have a CF admin "request timeout" time set: CF requests can still hand indefinitely if calling something outside of CF (like a DB or cfhttp call, to name just a couple of examples.)

So in a case like this, you need some sort of diagnostic tool that can show you what or at least how many requests are running within CF. You can get a very high-level count of that in the metrics log (enableable since CF 10 at the bottom of the "debug output settings" page).

Or you can view the info in the CF Enterprise Server Monitor (on the Statistics>Active Requests page and its "chart" icon on the right), which would show a count of running and queued requests. But of course that monitor only works in CF Enterprise and it does require flash. But to be clear, that chart I refer to does NOT require turning on any of the "start" buttons at the top of the monitor, so it comes at zero cost or overhead, for those who may be concerned about that regarding using this CFSM.

Then there are the 3rd party CF monitors, FusionReactor (fusion-reactor.com) and SeeFusion (seefusion.com), which can also track running requests, and with virtually no overhead.

I realize that sometimes when CF hangs up, even such monitors may not respond which is why some eschew them. But note that FR does log things so that you can see the counts of running and queued requests (and much more) even when CF is up but not responding, or of course after it has crashed. And the Enterprise edition of SF can track such details in a database.

Hope that's helpful to someone reading along in this thread.