Re: [Linux-decnet-user] DECnet routing

[Larry B. <ba...@us...>]

There is a previous thread from earlier this year dealing with DECnet =20=

routing and SuSE 10.0, subject "Decnet-Linux under SUSE 10.0", that =20
may be relevant.  Here's the instructions I used on my SuSE 9.3 =20
system to properly configure the DECnet interface for multicast and =20
the removal of the routing hack in /etc/init.d/decnet (SuSE names the =20=

ifcfg-eth-XXX files by MAC address, which must be changed for DECnet; =20=

other Linux's just name the files ifcfg-eth0, etc., so the rename =20
step is unnecessary):

Fixup the installation so dnetd can find fal:

# ln -s /usr/sbin/fal /usr/local/sbin/fal


Edit the DECnet startup/shutdown script, /etc/init.d/decnet:

#daemons=3D"dnetd phoned"
daemons=3D"dnetd"

Also, remove the DECnet routing hack, which is fixed by enabling the =20
allmulticast interface option, below:

       # FIXME dirty hack for SUSE 9.1 kernel.
       # Add all known nodes to neighbor list
       # comments to st...@op...

       for NODE in $(grep ^node /etc/decnet.conf| awk '{print $2}')
       do
         /sbin/ip -f dnet neigh add $NODE dev $CCT
       done

       # end of dirty hack

  Configure the Ethernet MAC address and enable the allmulticast =20
interface option in /etc/sysconfig/network/ifcfg-eth0:

# cd /etc/sysconfig/network

# mv ifcfg-eth-id-00:07:e9:16:b5:94 ifcfg-eth0

# vi ifcfg-eth0


# DECnet host address 54.145

LLADDR=3DAA:00:04:00:91:D8

# DECnet needs =93allmulticast on=94

LINK_OPTIONS=3D=94allmulticast on=94


Larry Baker
US Geological Survey
650-329-5608
ba...@us...

On Oct 10, 2006, at 12:12 PM, Steven Whitehouse wrote:

> Hi,
>
> On Fri, Oct 06, 2006 at 01:26:04PM -0400, Alan Kemmerer wrote:
>> Hi,
>> 	My last message was premature. The problem is not the packet =
size.
>> It is in the routing code.
>> I need to get the routing code working correctly for our network.
>> Since it is segmented by L1 routes
>> What I need to do is have the endnode populate its neighbors from =20
>> the hello
>> messages.
>> If the destination is not on the local segment, send the packet to =20=

>> the L1
>> Router
>> Or if the packets source is the L1 Router send the reply to the L1 =20=

>> router
>> and not the destination.
>>
>> Below is the tcpdump of a ping from 1.43(VMS) to 1.99(linux)
>> from linux.
>>
>> 1.99 is on a segment by itself with the router 1.405
>>
>> 1.43 is on a different segment with router 1.405
>>
> Would it be possible for you to do the tcpdump again, but with the -e
> flag so that we get the link layer headers too? Also if you could send
> the content of /proc/net/decnet_neigh it would be very helpful. If the
> router doesn't appear there, then that would explain the problem.
>
> Since the endnode hello you included doesn't include the router it
> would appear that this is the reason for the routing problem.
>
> Since the router hello is multicasted, the things to check are
> that Linux is listening on the correct address and that your network
> card is set up to receive multicast. The command
>
> /sbin/ip link ls
>
> will tell you that and
>
> /sbin/ip maddr ls
>
> well tell you if its listening to the correct multicast address. You
> could also try /sbin/ip link set eth0 promisc on    just to see if it
> makes any difference. It will take a few seconds for Linux to pick
> up the router in any case,
>
> Steve.
>
> ----------------------------------------------------------------------=20=

> ---
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to =20
> share your
> opinions on IT & business topics through brief surveys -- and earn =20
> cash
> http://www.techsay.com/default.php?=20
> page=3Djoin.php&p=3Dsourceforge&CID=3DDEVDEV
> _______________________________________________
> Linux-decnet-user mailing list
> Lin...@li...
> https://lists.sourceforge.net/lists/listinfo/linux-decnet-user