[nsp] ESR 10008 MPLS Tunnel woe

[James Galliford]

I believe that this delay is also apparent on the uBR10012 platform.
The delay in between pkts in/out of an interface as opposed to pkt
counters incrementing is quite different.  

For example:(Two intervals taken over a 5 second period)

  2 minute input rate 89066000 bits/sec, 19143 packets/sec
  2 minute output rate 50564000 bits/sec, 18453 packets/sec
     511284413 packets input, 218571828 bytes, 0 no buffer
     267755998 packets output, 3508356507 bytes, 0 underruns
 
  2 minute input rate 92288000 bits/sec, 19911 packets/sec
  2 minute output rate 52696000 bits/sec, 19205 packets/sec
     511284514 packets input, 218583115 bytes, 0 no buffer
     267756027 packets output, 3508363178 bytes, 0 underruns

I've often found this strange, but since the counters seem ok while
polling them with SNMP, I disregard their inaccuracy on the CLI.

-----Original Message-----
From: Eric Osborne [mailto:eosborne at cisco.com] 
Sent: Saturday, April 24, 2004 7:20 PM
To: nic at gblx.net; 'Cisco-Nsp'
Subject: RE: [nsp] ESR 10008 MPLS Tunnel woe




 > -----Original Message-----
 > From: cisco-nsp-bounces at puck.nether.net 
 > [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Nic McCartney
> Sent: Saturday, April 24, 2004 4:47 PM  > To: Cisco-Nsp  > Subject:
[nsp] ESR 10008 MPLS Tunnel woe  > 
 > 
 > Dear all,
 > 
 > Here is my network topology :-
 > 
 >  [7513a]<--pos-->[ESR10k]<---ATM--->[7513b]<--pos-->[7513c]
 > 12.0(25)S1       12.0(25)S1         12.0(24)S     12.0(25)S1
 > 
 > I have a pair of tunnels, one ESR10K > 7513b and t'other 
 > 7513b > ESR10K. When I send traffic from 7513c to 7513a then 
 > I see the traffic use the tunnel I have set up between 7513b 
 > and ESR10K (interestingly when I send a high number of pings 
 > it takes a full 7 seconds for the Tunnel stats to show that 
 > anything is passing).


7 seconds on the 7513, or on the 10k?  Or both?  Often this delay is
because the LCs (which keep the packet counters in a
distributed-forwarding system like the 7500) only periodically send
packet counter updates to the RP.  I know this is true on the GSR; the
stats period is like 5 or 10 seconds. Should be about the same on the
7500, but not the 10k.

 > 
 > When I send traffic from 7513a -> 7513c then I see my pings 
 > go OK but the traffic does not appear to be going over the 
 > tunnel ESR10K->7513b. By this I mean the stats do not show 
 > anything. The tunnel stats are all zeros. Funny thing is 
 > this, if I am on ESR10K and ping to 7513c then the Tunnel 
 > stats *do* increment. If I do sh ip rou <dest_ip> on ESR10K 
 > then it shows the Tunnel as the chosen route. It is indeed a 
 > most perplexing mystery. Anybody seen this before?
 >

Since it's only a one-hop tunnel (at least from the picture), it sounds
tricky to show whether this is a counters thing or whether forwarding
isn't right.  What do CEF and the 10k HW forwarding think about the
forwarding entry for the dest ip addr?  Are there any counters
incrementing on the ATM interface on either headend router?  



eric
 
 > Regards,
 > 
 > Nic
 > 
 > _______________________________________________
 > cisco-nsp mailing list  cisco-nsp at puck.nether.net 
 >  > https://puck.nether.net/mailman/listinfo/cisco- > nsp
 > archive 
 > 
 > at http://puck.nether.net/pipermail/cisco-nsp/
 > 


_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/