🚀 go-pugleaf

frank@iris.ansto.gov.au (Frank Crawford) writes:
................
>>Somewhere along this long return route messages are inclined to go astray.
>>At the moment we can get no response from munnari.oz.  The last time this
>>happened it was shown to be a fault with the return path
>>munnari->basser->perch->ditmela.  At the moment the return path is supposed to
>>be munnari->basser->ditmela.

>Yep currently most of your messages are ending up at ANSTO, where I have to
>manually reroute them to pta (which has a link to yarra, which is in
>Melbourne).  I can't work it out either!  Part of the problem seems to lay
>with metro, because messages bounce between basser and metro, before being
>rerouted (manually?) via atom.  I have no idea why munnari wants to send
>things via basser.

David Keegel at munnari has changed the routing so that messages from munnari
to yarra go direct to ditmela.  This made fetchfile from munnari work for me.
But fetchfiles from archie.oz and yallara at RMIT still don't work.  Something
is still screwed.

Here are some extracts:


From David Keegel:

]Problem summary: ACSnet mail to yarra.oz was getting lost; cpg is down
]stream from yarra.  Munnari's route was munnari!basser!ditmela!yarra,
]which has worked in the past.  Basser's route was (and is) NOT basser
]!ditmela!yarra but basser!metro!atom!pta!yarra (or similar).
]
]Since I last looked at this (and guessed that the problem was caused
]by a state change from metro, atom or pta and that messages were
]probably stuck somewhere between basser and pta), the state info has
]changed again.
]
]Now munnari is using munnari!ditmela!yarra as the route.  It seems
]that the basser-ditmela link has been declared "dead".  I have
]verified that files sent to bohra from munnari go into the queue for
]ditmela and disappear soon after.  Anthony, if you don't receive two
]files called "date", please let me know.
]

and from a bounce message.
*******************************************************************************
FILE SENT TO "als@bohra.cpg.oz" RETURNED FROM "pta.oz"

Failure explanation follows :-
Routing loop detected to node "atom"
*******************************************************************************


>As of today, I've rerouted about 100 messages from metro, some for bohra,
>but most for monu6.  At a guess, someone came back after Easter and found
>all these looping messages and has now rerouted them to me!


We need an "fsck"-like tool for route maps.

Could someone at the relevant nodes check their queues for lost messages?
You can't miss the several MBytes of files I fetched from yallara this week and
haven't got yet.  



>>I wonder how long until ACSnet fragments really badly, or will some kind 
>>soul look into it?

>This doesn't seem too different to the past.  It is just that some major
>nodes are a little slower to respond to ACSnet problems these days.

That's how it happens.
-- 
Anthony Shipman                 "You've got to be taught before it's too late,
C. P. Software                   Before you are six or seven or eight,
19 Cato St., East Hawthorn,      To hate all the people your relatives hate,
Melbourne, Australia, 3121       You've got to be carefully taught."  R&H