[c-nsp] MPLS TE Q

[Pete Templin]

Eric Osborne wrote:

> And if you do 'sh mpls tr tun int out $IF' and 'sh ip rsvp int' at the same
> time, you get two vastly different numbers?  From your original email, 
> it looks
> like you have 41.3M of bw allocated, which is close enough to 2*20.8M to
> suggest that either it's a simple bug or it's supposed to look like 
> that, but I
> wouldn't mind knowing what both numbers were when taken at the same time.
> If your traffic values are changing a lot you could also be looking at a
> temporary result of make-before-break, but I'd be pretty surprised if that
> would explain this big of a discrepancy.

core2-dlls#sh ip rs reservation | inc Se0/1/0
here  there  0   2     53    66.118.20.178 Se0/1/0  SE LOAD 0
here  there  0   1     301   66.118.20.178 Se0/1/0  SE LOAD 1868K
here  there  0   38    4585  66.118.20.178 Se0/1/0  SE LOAD 4331K
here  there  0   38    1265  66.118.20.178 Se0/1/0  SE LOAD 0
here  there  0   131   3225  66.118.20.178 Se0/1/0  SE LOAD 1739K
here  there  0   131   5297  66.118.20.178 Se0/1/0  SE LOAD 2223K
here  there  0   131   3709  66.118.20.178 Se0/1/0  SE LOAD 432K
here  there  0   2     36    66.118.20.178 Se0/1/0  SE LOAD 0
here  there  0   39    1186  66.118.20.178 Se0/1/0  SE LOAD 0
here  there  0   39    1276  66.118.20.178 Se0/1/0  SE LOAD 0
here  there  0   132   3284  66.118.20.178 Se0/1/0  SE LOAD 1382K
here  there  0   132   5044  66.118.20.178 Se0/1/0  SE LOAD 1446K
here  there  0   132   4292  66.118.20.178 Se0/1/0  SE LOAD 1058K
here  there  0   31    4398  66.118.20.178 Se0/1/0  SE LOAD 5238K
here  there  0   31    1209  66.118.20.178 Se0/1/0  SE LOAD 0
core2-dlls#sh ip rs int
interface    rsvp  allocated  i/f max  flow max sub max
Fa0/0/0      ena   29744K     95M      95M      0
Se0/1/0      ena   35725K     40M      40M      0
Se0/1/1      ena   0          40M      40M      0
Fa4/0/0      ena   0          95M      95M      0
Fa4/1/0      ena   0          95M      95M      0
Tu2          ena   0          0        0        0

These are taken within seconds of each other.

For what it's worth, I changed S0/1/0 to allow 10M of reservation at a 
time when I had ~15M, and TE rerouted the excess shortly thereafter, so 
I do think the "sh ip rs int" is a cosmetic bug.  The odd part is why 
I'm only seeing it on one node.

pt