CatOS Problem

john heasley heas at shrubbery.net
Sat Oct 9 23:49:31 UTC 2004


Thu, Oct 07, 2004 at 01:29:13PM -0800, Christopher E. Brown:
> john heasley wrote:
> >Wed, Oct 06, 2004 at 05:12:20PM +0200, Soren Hansen:
> >
> >>Well, on all my CatOS boxes I only had the output from "write term"
> >>in CVS. With the update, I now have the output from "write term
> >>all". No changes in CatOS recently. I'll probably change the
> >>intervals for retrieving end user access switch configurations now,
> >>as I keep getting mails when users come and go, because of the
> >>spanning tree port cost changes.
> >
> >
> >In theory, you should not run spanning tree on access ports, only
> >trunk ports; ie: should which connect to another switch.  And, if
> >there are no forwarding loops in your switch topology, dont run it at
> >all.
> 
> 
> Even if spanning tree is shut down things like port costs are still 
> tracked by CatOS.  And they change on up/down or rate change.
> 
> An example of something we see many times daily (and YES, spanning tree 
> is shutdown on this switch.
> 
> 
> Index: configs/xxxxxx
> ===================================================================
> retrieving revision 1.842
> diff -u -4 -r1.13 xxxxxxxx
> @@ -1419,10 +1419,10 @@
>   set spantree portinstancepri 8/45 0 mst
>   set spantree portinstancepri 8/46 0 mst
>   set spantree portinstancepri 8/47 0 mst
>   set spantree portinstancepri 8/48 0 mst
> - set spantree portcost    8/1,8/3-4,8/10,8/31-48  200000 mst
> - set spantree portcost    8/2,8/5-9,8/11-30  2000000 mst
> + set spantree portcost    8/1,8/3-4,8/31-48  200000 mst
> + set spantree portcost    8/2,8/5-30  2000000 mst
>   set spantree portinstancecost 8/1  cost 199999 mst
>   set spantree portinstancecost 8/2  cost 1999999 mst
>   set spantree portinstancecost 8/3  cost 199999 mst
>   set spantree portinstancecost 8/4  cost 199999 mst
> @@ -1430,9 +1430,9 @@

bizarre.  whats the point?  is there some other process that makes
use of these values?

makes you wish juniper made switches.



More information about the Rancid-discuss mailing list