[Mailman-Users] archiving partial duplicates

Dragon dragon at crimson-dragon.com
Thu Apr 6 21:49:27 CEST 2006

Con Wieland sent the message below at 11:50 4/6/2006:
>I'm having trouble with one of my archives. This archive is about 5
>years old and is fairly large. The problem is I recently had to
>remove a message and reindex. When I checked the archives for the
>current month a number of partial messages showed up with no from
>line and no date. These actually turned out to be chunks from some
>very old messages. I am confused though because the original message
>shows up fine and the current month's archive shows an incomplete
>portion of the old message. I'm assuming it's a corrupt message but I
>don't understand why it's showing up in both places. I have a couple
>dozen of these. I don't know if this existed before my last arch
>rebuild.  I have gone through several of these and am not able to see
>anything unusual. I have also checked the messages preceding and
>after the messages in question to no avail.
>Any idea's on what to look for would be appreciated.
---------------- End original message. ---------------------

I am just kinda guessing here but there is a caveat somewhere in the 
documentation about pipermail or the arch script that it has trouble 
digesting large MBOX files at a single gulp. It may be that this is 
the cause of the problem, it may not.

It is also stated somewhere in the documentation for either pipermail 
or the arch script that you can specify a start and end index for a 
chunk of messages in the MBOX file. What you could then do is rebuild 
the archive in chunks, say 500 posts at a time to see what happens.

So if you decided 500 was a good number to index, for the first chunk 
you would do:

         bin/arch --wipe --start=1 --end=500 listname

For subsequent chunks you would do (adjusting the start and end 
indexes of course...):

         bin/arch --start=501 --end=1000 listname

Hope that helps.


  Venimus, Saltavimus, Bibimus (et naribus canium capti sumus)

More information about the Mailman-Users mailing list