tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Filip Hanik - Dev Lists <>
Subject Re: svn commit: r451157 - in /tomcat/tc6.0.x/trunk/java/org/apache: catalina/connector/ coyote/ajp/ coyote/http11/ tomcat/util/net/
Date Fri, 29 Sep 2006 14:43:38 GMT
Remy Maucherat wrote:
> Filip Hanik - Dev Lists wrote:
>> Remy Maucherat wrote:
>>> wrote:
>>>> Author: mturk
>>>> Date: Fri Sep 29 01:23:37 2006
>>>> New Revision: 451157
>>>> URL:
>>>> Log:
>>>> Introduce keepAliveTimeout to be able to separate
>>>> the Keep-Alive and Socket timeout.
>>>> The patch enables to have infinite socket timeouts
>>>> while still having correct keep alive.
>>> As I said before, I see no point in that.
>> maybe I'm being dumb here, so please enlighten me, but how are they 
>> (keepalivetimeout vs connection timeout) different on a HTTP connection?
>> wouldn't the smallest of these two values always close the connection?
>> connection timeout on HTTP is after all only measured when the worker 
>> thread is not using the connection. on, it would be on a 
>> read() from the input stream, and on APR and NIO it would be time 
>> spent in the poller or a read()
> From what I understand, keepAliveTimeout is supposed to be the timeout 
> when doing HTTP keepalive (= keepalive between requests) while the 
> other one is supposed to be the timeout while processing requests (= 
> time spent blocking on a read when processing a PUT or POST). Since 
> there's already the disableUploadTimeout flag (which inflates the 
> timeout while requests are being processed), I could not find any 
> actual benefit with this change.
:) yeah, and with comet, the keep alive shouldn't be applied ever if 
that is the scenario, so inside the poll timeout checking it needs to 
make that distinction.
I think I'll leave this out of the NIO unless you think we should have 
it there.


To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message