Re: fastcgi-developers Digest, Vol 9, Issue 10

fastcgi-developers-request-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx wrote:
> Send fastcgi-developers mailing list submissions to
>       fastcgi-developers-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
>
> To subscribe or unsubscribe via the World Wide Web, visit
>       http://fastcgi.com/fastcgi-developers/listinfo.cgi/fastcgi-developers
> or, via email, send a message with subject or body 'help' to
>       fastcgi-developers-request-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
>
> You can reach the person managing the list at
>       fastcgi-developers-owner-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of fastcgi-developers digest..."
>   
> ------------------------------------------------------------------------
>
> Today's Topics:
>
>    1. fastcgi death spiral / apache cpu spikes (Scott, Brian)
>   
>
> ------------------------------------------------------------------------
>
> Subject:
> [FASTCGI] fastcgi death spiral / apache cpu spikes
> From:
> "Scott, Brian" <Brian.Scott-OAbWc/D93UTQT0dZR+AlfA@xxxxxxxxxxxxxxxx>
> Date:
> Tue, 27 Nov 2007 15:05:04 -0500
> To:
> <fastcgi-developers-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx>
>
> To:
> <fastcgi-developers-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx>
>
>
> Hello friends -
>
> I've got a open source app server (MapGuilde Server) running on Apache
> 2, fastcgi, on a windows environment.  Every now and again, running
> some reports through the application causes fastcgi / apache to spiral
> out of control and start logging error messages at an alarming rate
> (approx 12M / minute).  Log files repeat the following:
>
> "[Tue Nov 27 14:45:43 2007] [warn] FastCGI: (dynamic) server
> "C:/Program
> Files/MapGuideOpenSource/WebServerExtensions/www/mapagent/MapAgent.exe"
> has failed to remain running for 30 seconds given 3 attempts, its
> restart interval has been backed off to 600 seconds"
>
> CPU goes to close to 100% utilization and the application won't
> respond unless we restart apache, and nuke any phpcgi.exe processes
> hanging around.  I've been unable to determine a particular set of
> clicks / reports to run to generate this error and I have the feeling
> it isn't a specific set of actions so much as simply a
> misconfiguration. (?)
>
> This is a big problem for us; we've got people very interested in
> running the report, but they've got the ability to fill the server up
> with nonesense log entries if someone isn't available to toggle
> apache, not to mention we shouldn't have to bounce the app server
> every now and again to get some processes to stop running crazy.
>
> Here is the configuration section from httpd.conf in regards to fastcgi:
>
> ***************************************************************************************
>
> FastCgiConfig -idle-timeout 120 -startDelay 180 -minProcesses 4
> -maxClassProcesses 40 -multiThreshold 100 -singleThreshold 100
> -killInterval 30 -processSlack 20
>
> FastCgiServer "C:/Program
> Files/MapGuideOpenSource/WebServerExtensions/php/php-cgi.exe"
> -processes 4 -idle-timeout 120 -initial-env PHP_FCGI_MAX_REQUESTS=500
> -initial-env PHP_FCGI_CHILDREN=1
>
> **************************************************************************************
>
>
> As is likely apparent, I don't know much about fastcgi (or cgi), but
> none the less I am in the position of having to maintain this component. 
>
> Any insight is greatly appreciated.
>
> brian
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> fastcgi-developers mailing list
> fastcgi-developers-VYFNfGabXYtBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
> http://fastcgi.com/fastcgi-developers/listinfo.cgi/fastcgi-developers
>   
An easy google search found
http://www.nabble.com/Uncontrolled-Error-Logging---Non-Responsiveness-in-Apache-t4230159s16610.html
which offers as a solution:


Just in case someone else hits this, it looks as if modification to the
fastCGI configuration parameters in httpd.conf, can 'solve' this
problem.  Based on some other fastCGI problems people were experiencing,
I added this clause to my FastCGIConfig line:

 -startDelay 45

I'm not sure exactly what the problem was, but my processes were taking
a while to repaint the map, I'm thinking maybe they were taking too
long, Apache would try to modify the process and a death spiral ensued.
 Anyways, as soon as I made this entry and toggled Apache, I have been
unable to disable MapGuide by submitting requests.  

OK!

-pD

Further google searches might find more?
Fred


___________________________________
fastcgi-developers mailing list
http://fastcgi.com/fastcgi-developers/