[4082] in linux-net channel archive

home help back first fref pref prev next nref lref last post

Re: shell/tcp server failling(looping)

daemon@ATHENA.MIT.EDU (Rob McKee)
Sat Aug 17 12:50:12 1996

Date: 	Sat, 17 Aug 1996 08:49:42 -0700 (PDT)
From: Rob McKee <Imcom@imcom.com>
To: Klaus Lichtenwalder <klaus@gaston.m.isar.de>
cc: linux-net@vger.rutgers.edu
In-Reply-To: <Pine.LNX.3.94.960817150013.739C-100000@gaston.m.isar.de>

Thanxs Klaus;

	Now I have to read through inetd source to find out when it
starts reaccepting requests at the port.  Since this this is a production
(IE: public accessable) machine I have to be lenient to a degree. I might
have to hack inetd to restart accepting requests after a specified time.

maybe a wait.<count>.<restart> type interface.

	wait
		(wait, <default>, <disable [at least that's what the
		 default seems to be.]>)
	wait.300
		(wait, 300 connects max min, <disable [at least
		 that's what the default seems to be.]>)
	wait.300.300
 		(wait, 300 connects max a minute, if declared dead -- 
		 restart in 300 seconds)
	wait..300 
		(wait, <default>, 300 seconds..)


PS: I'm not subscribed to linux-net just linux-kernel


--
#         Rob McKee                      ImCom.Com       207.33.73.2
#         SysAdmin                       1630 N. Main St. Suite #444
#         Imagine Communications         Walnut Creek, CA 94596-4609
#
#         (510) 939-1691 BBS                    (510) 939-1684 Voice
#         The best defense is not to have a hay seed for a president


On Sat, 17 Aug 1996, Klaus Lichtenwalder wrote:

> On Sat, 17 Aug 1996, Rob McKee wrote:
> 
> Reading the manpage of inetd in the new NetKit 0.7 explains how to do
> it: after [no]wait you may append a `.' and the number of how many
> invocations per minute you will allow. Default is 40.
> 
> Klaus
> > 
> > I believe it does..  reading news via nntp and using the movement key to
> > fly by spamming definately triggers it.  I have to killall -15 inetd and
> > restartit.
> > 
> > > 
> > > SunOS's inetd protected itself from runaway TCP clients bombing it
> > >  with connect requests by declaring a service dead if it got over
> > >  40 requests / minute. A later patch from Sun (talking 4.1.x, heh)
> > >  allowed invocation of inetd -r nn ss where nn was number of connect
> > >  requests and ss was time frame. Don't know if Linux inetd has same
> > >  mechanism of self-defense, your script might well trigger it.
> > > 
> > > --alessandro
> > > 
> > 
> 
> ________________________________________________________________________
>  Klaus Lichtenwalder, Dipl. Inform., PGP Key:  email to key@Four11.com
>   Lichtenwalder@ACM.org, http://www.wp.com/Klaus, fax: +49-89-98292755
>  Check out Oregon vs. Schwartz:  http://www.lightlink.com/spacenka/fors
> 


home help back first fref pref prev next nref lref last post