tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Remy Maucherat <>
Subject Re: [VOTE] Split soTimeout to soTimeout and keepAliveTimeout
Date Mon, 02 Oct 2006 13:21:49 GMT
Mladen Turk wrote:
> Hi all,
> I would like to propose a simple vote on the thing I consider
> as very important.

I see that.

> The things we have right now for dealing with Keep-Alive is
> dependent only on the soTimeout.
> I propose that we split that to the real soTimeout which is the
> timeout between two consecutive read() on http request and the
> additional keepAliveTimeout that will be used for determining the
> timeout between two requests.
> Few reason why we should need those timeouts separately
> configured is:
> 1. Why not, if its possible without breaking current config ?
> 2. If the timeout between two request is lower then soTimeout it
>    will allow to have much higher number of slow clients
> 3. If the timeout between two requests is higher then soTimeout
>    it will allow to deal with slow clients sending one byte at the
>    the time with the unacceptable rate.

Only scenario 3 is not available currently, and it seems useless (in 
particular for the connector, which can never afford having a 
long timeout while waiting for the next request). If you really want a 
property named "keealiveTimeout", then "disableUploadTimeout" can be 
refactored into it.

> So, I'll just throw a vote here:
> [ ] I'm for that proposal
> [ ] I'm against that proposal
> [X] I don't care
> Of course, there is always an option to have an veto.


> As many of you already know, the code itself was already
> in the SVN, but it was reverted.

No, I had made some adjustments to it, and then you reverted it, along 
with a couple of my other patches. I do not think the defaults you 
implemented were sensible either (especially for AJP). Apparently is not 


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

View raw message