incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Filipe David Manana <fdman...@apache.org>
Subject Re: Replicating lots of large attachments
Date Wed, 10 Nov 2010 13:08:57 GMT
Hi Scott

On Sat, Oct 30, 2010 at 9:51 PM, scott <scott@meshbrains.com> wrote:
> Hopefully a simple architecture question...
>
> I have hundreds of laptop users that need to replicate hundreds of PDFs
> with several hundred pages each.
>
> For each PDF, I create a document with it as an inline attachment...
> everything is fine.
> If I later edit or add a field, maybe to many of the documents, the entire
> document (or
> documents) get replicated.

There's a new replicator being developed for some time now (Damien
started it, and I resumed his work).
This new replicator no longer does that. It only transfer attachments
that are not already in the target database.

Right now it's unstable (specially for pull replications). I expect it
to be close very close to finished within 1 or 2 weeks.
Ping me if I forget to let you know when to test it for your use case.

regargs,

>
> This discourages me from editing the meta info for each document as I know
> it's
> going to create a large replication task for such small updates.
>
> Is it better to create (for each PDF), say a standalone attachment, with
> minimal
> fields (just enough to identify it). And then create another associated
> document that
> contains more rich meta information that may change quite a bit? Smells a
> little
> kludgy.
>
> Or is there a better way to handle this?
>
> Thanks a bunch...
>



-- 
Filipe David Manana,
fdmanana@gmail.com, fdmanana@apache.org

"Reasonable men adapt themselves to the world.
 Unreasonable men adapt the world to themselves.
 That's why all progress depends on unreasonable men."

Mime
View raw message