On 12/3/2010 07:56, Michael Van Canneyt wrote:
On Fri, 3 Dec 2010, Michael Schnell wrote:
On 12/03/2010 01:32 AM, waldo kitty wrote:
using apache as an example, if the maximum number of servers (threads) is
reached, then requests are queued for X amount of time in the hopes that a
server comes free to handle the request... if a server does not come free
within X amount of time, then that request is dropped in the bitbucket...
i think that the queue (bucket?) also has a limit on the number of requests
that it will hold and if this is reached, then in FIFO fashion, the first in
is dumped since it is also the oldest...
Do I understand correctly that Apache performs the queuing ?
Apache performs it's own queuing, independent of FastCGI, CGI or whatever.
yes...
So The FCGI interface would need to provide it with additional information
(e.g. if any or how many threads are still available) _before_ it opens a
connection to the FCGI "Server" (as this would go directly to the thread and
not to some kind of thread-scheduler).
No. You probably misunderstood what Waldo wanted to say.
i think so, too...
What he said pertains to the inner workings of Apache, and is independent of how
a particular request is handled. It pertains to how Apache handles new
connections on its HTTP socket. No more, no less.
exactly... i offered how apache does it as an example so that those doing the
FCGI stuff could see how a mainstream app does this and maybe they would not see
any problems implementing it...
--
_______________________________________________
Lazarus mailing list
[email protected]
http://lists.lazarus.freepascal.org/mailman/listinfo/lazarus
