httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Nathan Schrenk <>
Subject Re: timeout patch
Date Fri, 26 Jan 1996 22:18:31 GMT
On Fri, 26 Jan 1996, Rob Hartill wrote:

> Nathan, is the timeout patch 
> for_Apache_1.1b0/84.timeout.patch
> going to be revised?

I was planning on revising it to include the reset_timeout() function in 
http_main.c that was suggested on the list.

> If it is, can I suggest you add a new config variable to live alongside
> the existing timeout e.g.
> Timeout 400
> BufferSendTimeout 40
> where this new variable defines the timeout for successful buffer writes.
> This number shouldn't be as big as the default overall Timeout value.

[some other stuff removed]

This sounds like an interesting feature, but I'm not sure how its going 
to be implemented without revising Apache's timeout code significantly.  
The way things are structured now, only one timeout in effect at any one 
time (per process) is possible.

Since this is a desirable feature, and the current timeout code isn't the 
best, a major timeout rewrite could do some good.  What about a system 
where a table of timeouts is kept which consists of several items: an 
identifier, the time that the timeout is set for, and pointers to a 
request_rec and conn_rec if needed.

The timeout signal handler could scan through this table whenever it 
executes and take appropriate action for all expired timeouts.  It would 
then use alarm() and reset the timeout signal to the timeout closest to 

Any thoughts?


> rob

Nathan Schrenk
Neoglyphics Media Corp.                    

View raw message