The MPLS WG Archive

Cell Relay Retreat>MPLS WG Archive>month:2001-Jul> msg00087



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

[Fwd: I-D ACTION:draft-pan-lsp-ping-00.txt]

  • From: Robert Raszuk <raszuk@cisco.com>
  • Date: Tue, 10 Jul 2001 07:48:26 -0700
  • CC: Ken Nagami <ken.nagami@toshiba.co.jp>, Ping Pan <pingpan@juniper.net>, George Swallow <swallow@cisco.com>, mpls-list <mpls@UU.NET>, mpls-ops <mpls-ops@mplsrc.com>, Nischal Sheth <nsheth@juniper.net>, Dave Cooper <dcooper@gblx.net>
  • Organization: Signature: http://www.employees.org/~raszuk/sig/

Hi Eric,

> So the most you can learn from this is that the failure of the original ICMP
> ping is  due to  a problem  in the return  path for  data packets.   Is that
> correct?

Yes and in the same time prove that the LSP under test is fine.
 
> But if all you want to know is whether the reverse data path is working, why
> not just do ordinary ICMP ping in the reverse direction?

I am sure the same test procedure will be executed in the reverse
direction as well again to verify unidirectional LSP state not the both
direction state.

I think that is the author's intention.

R.


> 
> What am I missing?

> Eric Rosen wrote:
> 
> I'm still confused about what this actually tells you.
> 
> From the failure  of the ICMP pings, you already know  that either the LSP's
> data plane is failing, or else  there is no operational return path for data
> packets.  However, there  is apparently an operational path  for the forward
> and reverse control traffic.
> 
> If you send the LSP-ping down the LSP itself, then it "ought to" fail if and
> only if the LSP's data plane  is failing.  So presumably, if it succeeds you
> know that the failure is in the reverse data path.
> 
> If you  do not send the  LSP-ping down the  LSP itself, then it  will likely
> succeed, but you don't learn anything.
> 
> So the most you can learn from this is that the failure of the original ICMP
> ping is  due to  a problem  in the return  path for  data packets.   Is that
> correct?
> 
> But if all you want to know is whether the reverse data path is working, why
> not just do ordinary ICMP ping in the reverse direction?
> 
> What am I missing?