Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 5 Jun 1999 14:47:17 -0700 (PDT)
From:      Julian Elischer <julian@whistle.com>
To:        Matthew Dillon <dillon@apollo.backplane.com>
Cc:        David Schwartz <davids@webmaster.com>, Peter Wemm <peter@netplex.com.au>, Poul-Henning Kamp <phk@critter.freebsd.dk>, current@FreeBSD.ORG
Subject:   Re: RE: net.inet.tcp.always_keepalive on as default ? 
Message-ID:  <Pine.BSF.3.95.990605144517.20899C-100000@current1.whistle.com>
In-Reply-To: <199906051833.LAA15517@apollo.backplane.com>

next in thread | previous in thread | raw e-mail | index | archive | help
I think part of the solution is a new class of keepalives..
With this new class, a keepalive is sent every N second (3600?)
but if no response is heard, no action is taken. The only action that is
taken is if a NAK is recieved in response.

Most IP addresses woudl be re-used within a few days, so even if someonen
hangs up, in most cases SOMETHING will respond with a NACK withinthe next
day or two.

julian


On Sat, 5 Jun 1999, Matthew Dillon wrote:

> :	There is no logical reason for a well-designed web server to enable
> :keepalives. Of course, they don't hurt anything.
> :
> :...
> :
> :	Agreed. Telnetd is the exception, keepalives are great for it. For
> :everything else, almost, data timeouts make far more sense. And keepalives
> :will do nothing, but won't hurt anything.
> :
> :	As I have said before, any application that does not implement data
> :timeouts for all states, and does not enable keepalives is BROKEN.
> 
>     You are missing the point, big time.
> 
>     There are hundreds of programmers writing hundreds of servers, most 
>     written by third-parties.  New ones pop up every day.  Nobody
>     is going to go through and make sure all of them turn on keepalives. 
>     Nobody is going to go and try to contact all the authors involved to
>     try to get them to implement their own timeouts.  There are, in fact,
>     many servers where implementing a timeout is *inappropriate*.
> 
>     ssh, rsh, and telnet for example.  nntp is an example of a server where
>     the timeout depends on the use.  Some ISP's might want to implement a 
>     timeout, others might not.  At BEST I decided to *not* have a timeout...
>     people can stay connected and idle for hours if they want.
> 
>     Your 'solution' is no solution at all.  You aren't thinking through the
>     problem carefully enough.
> 
>     The Keepalive capability exists for a reason.  The original reasons for
>     not turning them on by default all those years ago no longer exist, and
>     the only reasons people come up with now are extremely shallow and 
>     uninformed.
> 
>     I have yet to hear a single informed opinion against turning on
>     keepalives.  
> 
>     All I hear is mob-mentality stuff: people with opinions not backed by
>     real facts, or people with opinions based on assumptions that are 
>     incorrect.
> 
> 					-Matt
> 					Matthew Dillon 
> 					<dillon@backplane.com>
> 
> 
> To Unsubscribe: send mail to majordomo@FreeBSD.org
> with "unsubscribe freebsd-current" in the body of the message
> 



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.95.990605144517.20899C-100000>