Re: Browser stop: no EPIPE recieved

At 05:09 AM 12/01/04, Peter Åstrand wrote:

>I'm trying to detect when the user presses "stop" in the browser, and
>abort an operation in my FastCGI program that takes a long time. It
>doesn't work. Despite several hours of debugging, I cannot find the
>problem.

I have a stack of emails in which we basically worked out that the FCGI process will never see a SIGPIPE because Apache will (usually? always?) buffer up outgoing data from the FCGI.  So there's almost no window from the FCGI POV to receive a SIGPIPE.  This remains a "known problem" in my application that will require support intervention to address (or a design mod. :) 

The alternative to allow time for SIGPIPE, that the FCGI execution is somehow paced by the download progress, can ruin scaling if a bunch of hits come in over slow connections: you have a bunch of FCGI sessions in memory plodding along, using up the quota, etc.  Any updates are welcome.

>Any ideas?
>http://www.fastcgi.com/archives/fastcgi-developers/2004-February/003268.html
>mentiones Apaches content-length filter. Could this be the cause of my
>problem?

That was me, so I'll say "yes". :)

HTH,
TW

PS I am on Apache 2.0.47 and FCGI works well otherwise.

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