[c-nsp] BGP confusion in 12.2(18)SXE2

[Gert Doering]

Hi,

On Wed, Sep 14, 2005 at 05:40:02PM -0400, Rodney Dunn wrote:
> One I missed that I learned about. I haven't looked all that closely
> at it yet but it's in this same area.
> 
> CSCsb09852
> Externally found severe defect: Open (O)
> BGP: pathless nets not freed when updgrp members are out of sync

Dunno whether this is also biting us, but "show ip b update-group summary"
*did* display some table version differences.  

80.81.192.108   4 15169   78142   78193 16851068    0    0 2d09h          23
80.81.192.111   4  6830   78623   78178 16851626    0    0 2d09h         211
80.81.192.116   4  8404   39860   39190 16851436    0    0 2d09h         187
80.81.192.118   4 15933   39062   39151 16851626    0    0 2d09h           3

OTOH, I think this is to be expected on a router that receives a full
feed (and thus, lots of table changes) and cannot always update all 
peers in a large peer-group simultaneously (slow peer, whatever).

I have now "clear ip b * soft out" all of them, waited a couple of 
minutes (to give the BGP scanner a chance to clean up), and alas, it
didn't help - but it certainly was worth a try.

The main symptoms are still the same - "show ip b <prefix>" claiming more 
paths "available" than actually being displayed, and "show proc mem summ"
showing 219 Mb for the "BGP Router".


In the release notes for SXF, there was yet another "BGP does not
clean up things" bug - CSCsa87034 - could that be "the one"?  Or will 
that one manifest differently?


many thanks for looking into this,

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