[c-nsp] BFD timers in SXH3a - can't go below 1s?

[Gert Doering]

Hi,

I'm a bit confused.  I have enabled BFD on a few links between SXF* and
SXH3a machines, and on other links between SXH3a and SXH3a boxes.  SXF
box is 7604+Sup720, SXH3a boxes are 6504+Sup720-10G.

All interfaces are configured with

 bfd interval 50 min_rx 50 multiplier 10

(which is fairly conservative, as BFD goes, but 500ms second error
detection time is perfectly fine for us, and we're aware that this
platform is not really happy with faster BFD)


On the SXF-SXH3a link, I see:

SXF#sh bfd neigh det
OurAddr       NeighAddr     LD/RD RH  Holdown(mult)  State     Int
195.30.xx.xx  195.30.xx.yy   1/1  1   452  (10)      Up        Gi2/8
  
Local Diag: 0, Demand mode: 0, Poll bit: 0
MinTxInt: 50000, MinRxInt: 50000, Multiplier: 10
Received MinRxInt: 50000, Received Multiplier: 10
Holdown (hits): 500(0), Hello (hits): 50(11438)
Rx Count: 12105, Rx Interval (ms) min/max/avg: 40/56/46 last: 48 ms ago
Tx Count: 11441, Tx Interval (ms) min/max/avg: 40/68/49 last: 8 ms ago
Registered protocols: EIGRP
Uptime: 00:09:24
Last packet: Version: 0            - Diagnostic: 0
             I Hear You bit: 1     - Demand bit: 0
             Poll bit: 0           - Final bit: 0
             Multiplier: 10         - Length: 24
             My Discr.: 1          - Your Discr.: 1
             Min tx interval: 50000    - Min rx interval: 50000
             Min Echo interval: 50000


-> so it seems to be doing 50ms just fine.


On the SXH3a-SXH3a link, I see:

SXH3a#sh bfd neigh det

OurAddr       NeighAddr     LD/RD  RH/RS  Holddown(mult)  State     Int
195.30.XX.AA  195.30.XX.BB   1/1    Up        0    (10)   Up        Gi2/8
       
Session state is UP and using echo function with 50 ms interval.
Local Diag: 0, Demand mode: 0, Poll bit: 0
MinTxInt: 1000000, MinRxInt: 1000000, Multiplier: 10
Received MinRxInt: 1000000, Received Multiplier: 10
Holddown (hits): 0(0), Hello (hits): 1000(780297)
Rx Count: 780119, Rx Interval (ms) min/max/avg: 624/1128/877 last: 188 ms ago
Tx Count: 780298, Tx Interval (ms) min/max/avg: 756/1000/880 last: 332 ms ago
Elapsed time watermarks: -1 0 (last: 0)
Registered protocols: EIGRP
Uptime: 1w0d
Last packet: Version: 1            - Diagnostic: 0
             State bit: Up         - Demand bit: 0
             Poll bit: 0           - Final bit: 0
             Multiplier: 10         - Length: 24
             My Discr.: 1          - Your Discr.: 1
             Min tx interval: 1000000    - Min rx interval: 1000000
             Min Echo interval: 50000

... which looks to me like "it's sending packets in 1-second intervals"
(and not 50ms), so the error detection time would be "10 seconds" - which
is certainly not what we are aiming for.


Has anyone of you seen this before?  Where is this coming from?  I checked
the interface configuration again and again, and I can't even configure
"1 second" intervals (max. is 999 ms).

This is happening on ports on 6408A-GBIC cards as well as on 6704-10GE(CFC)
cards.  The 50ms example on the SXF box is also on a 6408A-GBIC card.


(Even more fascinating: why is it 50ms on SXF-SXH3a?  Shouldn't SXH3a
dictate "min_rx: 1000000" there as well?  It doesn't...)

gert
-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert at net.informatik.tu-muenchen.de