The MPLS WG Archive

[Eric Gray <eric.gray@sandburst.com>]

Vincent,

    Think about it this way: a (set of) label(s) on an inSegment map
reliably to a specific label stack manipulation (pop, swap, push,
shake, rattle and/or roll) for any given LSP.

--
Eric Gray

You wrote:

> What if you replace the top label and then push another label?  Against
> which label would you store the outSegment counts?  Or would you store them
> against both labels?
>
> Thanks,
> -V
>
> -----Original Message-----
> From: owner-mpls@UU.NET [mailto:owner-mpls@UU.NET]On Behalf Of Vipul
> Sharma
> Sent: Thursday, June 21, 2001 8:03 PM
> To: naderv@msn.com
> Cc: mpls@UU.NET
> Subject: RE: LSR-MIB Question
>
> Shahram,
>         I had run into similar question a while ago...Attached is the info. which
> will help you for 1.
> My feeling is that since there is no error conditions maybe they need to be
> collected in mplsInSegmentDiscards. But that is just what I beleive maybe
> the gurus have a differing opinion.
> For 2, I believe that we are reqd. to keep 2 sets of stats one for the
> swapped label and one for the pushed label
>         Thanx,
> Vip
> vsharma@ieee.org
>
> Attached message
> =================================================================
>
> >         In the MPLS MIBs (MPLS TE , MPLS LSR MIBs)I see stats for errored
> > packets.
> >For e.g. in the MPLS LSR MIB there is mplsInSegmentErrors which is "The
> >number of errored packets received on this segment."
> >         My question is what type of errors make these packets errored
> packets
> >because most errored packets are dropped at the physical interface iteself
> >by the chips (my understanding). So does this stats put a req. to run the
> >phy. interface device in promiscous mode?
>
>          Basically any case where a labeled packet is received and
> cannot be forwarded by the MPLS software. There are many cases:
> out of bandwidth in the switching fabric, illegal labels in the
> packet, etc...
>
>          --Tom
>
> >Vip
> >
>
> -----Original Message-----
> From: owner-mpls@UU.NET [mailto:owner-mpls@UU.NET]On Behalf Of S. Davari
> Sent: Monday, June 11, 2001 6:42 PM
> To: mpls@UU.NET
> Subject: LSR-MIB Question
>
> Hi All,
>
> A few questions regarding LSR-MIB:
>
> 1) If TTL expires in an MPLS packet is the packet
> counted in
> mplsInSegmentErrors" or not?
>
> 2) If we receive a labeled packet and swap the label
> and then push a new label
> do we keep 2 sets of statistics on the outgoing
> segment, one for the swapped
> label and one for the pushed label or only one?
>
> BTW, I also noticed that in LSR-MIB a swap consists of
> a pop on incoming segment followed
> by a push on outgoing segment, this is odd !! The
> odder thing is that the incoming segment has
> a parameter called InSegmentNpop in which "N"
> identifies the number of pops to be performed on the
> insegment. How does one know in advance how many pops
> to perform? If you know that, then it means
> that all those labels that you pop are not used at
> all.
>
> Thanks,
> -Shahram
>
> _______________________________________________________
> Do You Yahoo!?
> Get your free @yahoo.ca address at http://mail.yahoo.ca