[Mailman-Users] Importing archives again

[Mark Sapiro]

Brad Knowles wrote:

>At 10:53 AM -0600 4/19/07, Ivan Van Laningham wrote:
>
>>  There's a FAQ entry on importing archives.  However, if the
>>  instructions are followed, the existing list.mbox would be overwritten
>>  with an old archive.
>
>No, I don't think so.  It just reads from the specified mbox file, or 
>from the default mbox file if you don't give it a different one.  The 
>"cooked" text & HTML indexes will be rebuilt during the process, but 
>the "raw" mbox file should be read by the process, but otherwise 
>untouched.


I believe Ivan is concerned about the first step in the FAQ which says
"If your archive is in mbox format copy it to
archives/private/<list>.mbox/<list>.mbox".

That is written in the context of importing an archive into a brand new
list which has no posts yet. If the list has been running for a while
and has a non-empty archives/private/<list>.mbox/<list>.mbox file, you
don't want to overwrite it with the old archive .mbox file.


>>                        That surely cannot be right.  Can I put it
>>  someplace else in order to import it?  Shouldn't I lock the list while
>>  running bin/arch?


You have lots of choices. You don't need to lock the list. bin/arch
locks the archive.

Your choices depend on whether or not you care about message numbers in
the archive and whether they should be preserved (they are part of the
message URL), and how much trouble you want to go to the 'next time'.
Here's a rough summary of your choices. You may want to briefly stop
mailman if you are merging into the existing .mbox to prevent it from
being updated while you are manipulating it.

1) append the old archive .mbox to the existing archive .mbox and run
bin/arch --wipe <listname>

2) prepend the old archive .mbox to the existing archive .mbox and run
bin/arch --wipe <listname>

3) just run bin/arch <listname> /path/to/old/.mbox

Choice 1) preserves the message numbers for the current archive but
gives a non-chronological .mbox file.

Choice 2) gives a chronological .mbox file, but alters message numbers
for the current archive.

Choice 3) is the simplest, but it complicates future rebuilding and
makes it very difficult to preserve message numbers in a future
rebuild.

Also note, it is a good idea to run bin/cleanarch to at least check all
the .mbox files.


>>  I've got seven years worth of archives to do for as many lists, so I'd
>>  really like to get this right.  I'd be happy to edit the FAQ once I
>>  know the proper steps.  (And if I missed the answers in my admittedly
>>  cursory search of the mailman-users archives, they should still be
>>  inserted into the FAQ.)
>
>So far as I know, the FAQ is pretty complete on this topic.  If 
>there's anything missing, please let us know.


The FAQ could be updated for the 'merging archives' case. Please do if
you feel you could help.

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan