[ previous ] [ next ] [ threads ]
 From:  Melvin <melvin at sleepydragon dot net>
 To:  Jeroen Visser <monowall at forty dash two dot nl>
 Cc:  m0n0wall at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall] Re: m0n0s hanging : any progress ?
 Date:  Wed, 30 Aug 2006 08:47:32 -0400
Jeroen Visser wrote:
> On Wed, 30 Aug 2006 11:11:13 +0200, Rolf Kutz wrote
>> I second this. I only get freezes when I do
>> gnutella. Maybe some table or counter fills up.
> I second that too.
> http://m0n0.ch/wall/list/showmsg.php?id=282/20
> http://m0n0.ch/wall/list/showmsg.php?id=278/55
> --
> Jeroen Visser.
Or just perhaps, the traffic level is higher with these enabled and 
because of the traffic volume, things work differently?  I would tend to 
think that in order to definitively tag P2P as the culprit there would 
need to be a replacement by something else to represent the appropriate 
level of traffic in order to truly demonstrate that something else isn't 
failing.  That isn't to say that P2P isn't in fact the culprit, only 
that the anecdotal evidence so far doesn't seem to prove or disprove 
that.  If your normal traffic with P2P is 20 units, then killing P2P and 
letting traffic drop to 5 units doesn't prove anything. 

By way of example take the case of the car which had a tendency to just 
die at speeds over 75 mph.  Very repeatable, and absolutely no failure 
cases below that speed.  The problem turned out to be a fuel supply 
issue. The problem with using the speed as the culprit is that the only 
time the fuel requirement exceeded the available flow was at 75 mph.  
The problem was discovered after it was determined that the same failure 
could be produced by running in a lower gear and not allowing the 
transmission to shift up.  It really had nothing to do with speed, only 
fuel flow.