[OpenVPN home] [Date Prev] [Date Index] [Date Next]
[OpenVPN mailing lists] [Thread Prev] [Thread Index] [Thread Next]
Google
 
Web openvpn.net

Re: [Openvpn-users] replay-window calibration


  • Subject: Re: [Openvpn-users] replay-window calibration
  • From: Sean Patrick <spatuality@xxxxxxxx>
  • Date: Thu, 2 Sep 2004 23:50:46 -0400 (EDT)

 --- James Yonan <jim@xxxxxxxxx> wrote: 
> 
> 
> On Thu, 2 Sep 2004, Sean Patrick wrote:
> 
> > Hello,
> > 
> > Thanks for making a product which works so well in
> > many diverse environments. It's amazing how much
> time
> > can be saved just by not having to modify a lot of
> > config files for different platforms.
> > 
> > Using the suggested verb 4 setting, my
> > /var/log/messages log is showing "Replay-window
> > backtrack occurred [x]" warnings.
> > 
> > One has [9], and another has [18].
> 
> That number is printed any time the maximum
> backtrack seen so far 
> increases.
> 
> For example suppose the sender sends packets #1, 2,
> 3, 4, 5, 6, 7
> 
> Suppose the receiver receives them out of order: #1,
> 2, 7, 4, 5, 6, 3
> 
> The maximum backtrack seen in this sequence is 4,
> because we got #7 and 
> then we backtracked down to #3 before presumably
> moving on to #8.
> 
> In this case, OpenVPN would print "Replay-window
> backtrack occurred [4]".  
> Future backtracks would not be logged unless they
> exceeded the previous 
> "high water mark" of 4.
> 
> > I understand the man page states "n" means the
> sliding
> > window of size n, but does that mean my logs are
> > showing the replayed packet is 9 and 18 bytes
> long? Is
> > it the number of replay packets which occured in
> time
> > "t" (default 15 seconds)?
> 
> The default window size is 64.  That means that if
> OpenVPN sees a 
> backtrack larger than 64, it will drop the packet.
> 
> > What does the "x" mean, and how is that used in
> > relation to  calibrating the "replay-window n [t]"
> > setting in place of the "n", as the man page
> states?
> > 
> > Can anyone help explain the backtracking and
> > calibration?
> 
> If you see a message like this:
> 
> Replay-window backtrack occurred [63]
> 
> followed by packet loss, you might want to increase 
> the n parameter to something more than 64.
> 
> The t parameter usually doesn't need to be changed.
> 
> While the default replay parameters are sufficient
> for most networks, I 
> did notice a message on an IPSec list a while back
> from someone who 
> claimed that he needed a window size of 2048 when
> dealing with satellite 
> links.
> 
> The occasion where you might need to increase the
> replay parameters would
> be a case where you have a high bandwidth, high
> latency network link.
> 
> James
>  
Perfect. Thanks for the clear explanation! Looks like
I will have to look to another paramater to debug the
system when it drops file (2gb files...).

Brian

______________________________________________________________________ 
Post your free ad now! http://personals.yahoo.ca