Linux-Hams archive - April 1997: Re: Errors in rspfd ?

Re: Errors in rspfd ?


Tue, 29 Apr 1997 22:14:13 +1000 (EST)


Julian Munoz typed:
> Hello Craig, we are "observing" rspf in Madrid, look at this reporting
> packet (coming from me).
>
>
> Port ax0: AX25: EA4ABB-5->EA4RCT-5 <UI> pid=IP
> IP: len 138 44.133.28.114->44.133.28.255 ihl 20 ttl 64 prot RSPF
> RSPF: version 22 type ROUTING UPDATE fragment 1 frag total 1 sync 4 #nodes
> 3 env_id 182
[snip]
wow, that's the biggest routing update that I've seen for a while, you'll be
testing my fragmentation part of the code soon >;-)

> My situation is :
>
> 44.133.28.76 [J]
> nos.ea4rct.ampr.org--------------
> | |
> vhf port | uhf port |
> V V
> 44.133.28.122 [L] 44.133.28.114 [L]
> 44.133.28.113 [L] ----------> (ME)
> 44.133.28.96 [L] I am listening vhf port

OK so you have two ports yes? One is UHF and you can transmit on it, one
is VHF where you can only listen.

>
> [L] = Linux with rspfd 2.2 (44.133.28.96 has the very last version, so
> maybe this "errors" are solved in that version, that others haven't
> installed yet).
>
> [J] = Jnos 1.11x6 with rspf 2.1
>
> [J] Doesn't work well, because it does't not broadcast nothing, nor RRH,
> nor reporting links stat datas (looking for a solution).
JNOS has this funny thing in that it will not broadcast any packets out an
interface that it has not heard other RSPF stations already.

> Apart of the 44.133.28.113 link to itself, wich I think is an error in its
> configuration, there is two strange errors:
>
>
> 1) I report a FALSE link :
>
> . . horizon 32 ERP factor 0 cost 64 #adjacencies 1
> . . . 44.133.28.96/32
>
> This link is transient, I think it appears when I receives the RRH of this
> router, and while it is trying to ARP to it. Sometimes, that's
> 44.133.28.133/32 wich appears.
Hmm, I see. These guys should be picked up but should the fact that you
cannot hear them should not be told to the others at this point in time.

> So it looks like this link is added in the "Tentative State", while it
> should'nt (??).

No, they should be tentative. A "new" adjacency that is discovered by an
ARP or RHH will be marked as tentative until we have attempted SusPings
pings, then (and this bit is not in the current spec) it goes bad for some
time.

The bug here is that an adjacency that is marked as Tentative should not be
in a routing list.

> 2) My packets go to EA4RCT-5 instead of QST (EA4RCT-5 is the link address
> of 44.133.28.76)
>
> That's maybe because defaukt route to 44.0.0.0/8 is through it ?
Yes, you need to add a route to your broadcast address that is direct off
that interface. This is due to the assumptions that are made about an IP
network.

I've finally succumbed and there is a Web page for RSPF and RSPFd. You'll
find it at http://www.triode.net.au/~csmall/rspf There isn't much there
yet, but there will be, I hope.

- Craig

-- 
  // /\   |  | |  Craig Small VK2XLZ     @home: zdiq@roskapostia.tunk.net
 ||==||===|==|=|  [44.136.13.17]         @play: terhi.victor@logonet.com
  \\ \/   |  | |  finger xug@qcomm.com for PGP key!