On 2011-12-27 18.21, Peter Lothberg wrote:
(if there are more than one area router on the same LAN it will
use the one with the highest node number, no metrics...)
Nah.
It will use the one with the highest number as the arbitor if they have the same router priority, unless I remember wrong. First of all it will use the one with the highest priority value.
But I might remember things wrong, as usual... :-)
Johnny
On 2011-12-27 16.36, Steve Davidson wrote:
Hey folks...
LAN costs are 4 by default (for VMS nodes anyway). If we make use of
that then all we have to touch are the Multinet circuits. I set my
Multinet circuits to 3 because they are much faster and I would prefer
to use the most "efficient" path possible. You will loose the ability
to use SET HOST/LAT, and MOM/MOP, and LAD/LAST, but that should me
minor...
Huh? I didn't follow that last part. Why would you loose the ability to use MOP and LAT when you change circuit costs for DECnet?
Johnny
-Steve
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On
Behalf Of Bob Armstrong
Sent: Tuesday, December 27, 2011 10:00 AM
To: hecnet at Update.UU.SE
Subject: [HECnet] Circuit costs
Peter Lothberg wrote:
Here is a suggestion....
Set all Multinet links to cost 5 in both ends
Set all bridged Ethernet interfaces to 10
Set all Multinet link nodes Level2 Area Routers
Well, I like it. I'll do it if you do it...
Actually that's a problem with DECnet - everybody _has_ to do their
own;
circuit costs are individually defined per node.
Bob
On 2011-12-27 15.59, Bob Armstrong wrote:
Peter Lothberg wrote:
Here is a suggestion....
Set all Multinet links to cost 5 in both ends
Set all bridged Ethernet interfaces to 10
Set all Multinet link nodes Level2 Area Routers
Well, I like it. I'll do it if you do it...
Actually that's a problem with DECnet - everybody _has_ to do their own;
circuit costs are individually defined per node.
I think it makes sense as well, as ethernet segments can't really separate costs for different next hops.
Might as well make them expensive so that ptp links are preferred when possible.
Johnny
On 2011-12-27 12.05, hvlems at zonnet.nl wrote:
Isn't the problem similar to what happens if there would have been two active bridges between two ethernet segments?
Sorry for being offline a couple of days...
Anyway, the answer to that one is no (maybe someone else already replied, I have over 200 unread after the spam filtering have run over my inbox).
If you have two active bridges between two ethernet segments one of two things happen.
1) The bridges realize this, and shut one of them down.
2) The bridges don't realize this, and starts sending packets around in a never ending loop, causing havoc.
My bridge program falls in category 2. Any sensible ethernet switch you might have at home hopefully falls into category 1.
The normal way of implementing category 1 is by having the switch implement the spanning tree protocol to detect possible loops.
Johnny
------Origineel bericht------
Van: Peter Lothberg
Afzender: owner-hecnet at Update.UU.SE
Aan: hecnet at Update.UU.SE
Cc: hecnet at Update.UU.SE
Beantwoorden: hecnet at Update.UU.SE
Onderwerp: Re: [HECnet] ROOSTA routing problem
Verzonden: 27 december 2011 18:43
Hmmm.. LEGATO can't talk to ROOSTA ("remote node not currently reachable")
even though there's a direct Multinet link to ROOSTA (and yes, the link is
UP and working).
I think the problem is that ROOSTA is in area 6, and the area router for 6
is STAR69. Trouble is, STAR69 doesn't have a path to ROOSTA and doesn't
know how to send it packets.
(It doesn't help that STAR69 still thinks that ROOSTA is 3.34 - it moved
to 6.134 a while ago - but I don't think that's causing the problem.)
Multinet links metric 5
Bridged ethernet metric 10
Set all multinet-link nodes Level2-Area-Routers
--P
.eqs. compares two strings for an exact match. It is case sensitive and thus
surprisingly strict for an OS that seems to do uppercase only...
What is regex??
Circuit names for ethernet device are DDD-N
DDCMP (serial lines), CI and DSSI devices are DD-N-M
Where D is a letter and N and M are digits.
Hans
-----Oorspronkelijk bericht-----
Van: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] Namens
Brian Hechinger
Verzonden: donderdag, december 2011 17:25
Aan: hecnet at Update.UU.SE
Onderwerp: Re: [HECnet] DCL Help needed
On 12/29/2011 11:17 AM, Steve Davidson wrote:
Steal and steal often is my motto :-)
Take whatever you need! The new version will be better commented and
not require the reboot. I will place a copy of the old version of the
data file (it matches this version) in the same directory.
I'm assuming .eqs. is rather limited (say the way = is in bash)?
Is there any sort of regex style matching available?
Short of that, is it fair to say that all circuit names in the output of
show known circuits will always be in the form of XXX-N for non-multinet
tunnels?
-brian
What bothers me are the multiple paths between locations.
------Origineel bericht------
Van: Steve Davidson
Afzender: owner-hecnet at Update.UU.SE
Aan: hecnet at Update.UU.SE
Beantwoorden: hecnet at Update.UU.SE
Onderwerp: RE: [HECnet] Re: Topology 28-Dec-2011
Verzonden: 29 december 2011 22:35
We definitely have to review the cost structure at the end of this
mapping exercise!
-Steve
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On
Behalf Of Rok Vidmar
Sent: Thursday, December 29, 2011 3:30 PM
To: hecnet at update.uu.se
Subject: Re: [HECnet] Re: Topology 28-Dec-2011
Are you looking at the links dynamically or based on what was
submitted
by each of us?
Dynamically, with ncp show adjacent nodes.
--
Regards, Rok
We definitely have to review the cost structure at the end of this
mapping exercise!
-Steve
-----Original Message-----
From: owner-hecnet at Update.UU.SE [mailto:owner-hecnet at Update.UU.SE] On
Behalf Of Rok Vidmar
Sent: Thursday, December 29, 2011 3:30 PM
To: hecnet at update.uu.se
Subject: Re: [HECnet] Re: Topology 28-Dec-2011
Are you looking at the links dynamically or based on what was
submitted
by each of us?
Dynamically, with ncp show adjacent nodes.
--
Regards, Rok
