[pybsddb] Snapshotting / replicating bdb

[Jon Dyte]

On Monday 25 May 2009, Jesus Cea wrote:
> Cameron Simpson wrote:
> > I would point out, in case it wasn't made clear in an earlier post, that
> > like copying filesystems, copying database files shoud be surrounded by
> > some sort of sync function:
> >
> >   db.freeze()
> >   ...copy/snapshot the db...
> >   db.unfreeze()
> 
> This is a good advice in general. But current databases have procedures
> to do a hot-backup in a safe way, simply because you can not stop
> database operations for the hours it take to copy a multi-terabyte
> database environment.
> 
> Working with BerkeleyDB, you have a couple of options:
> 
> 1. Working with snapshots+cloning (ZFS), I simply clone the database and
> then do a database recovery over the clone. Of course the BDB
> environment must be transactional. This is very fast. And safe (if the
> database is transaction protected).
> 

Jesus 

Just to be clear are you saying you do the ZFS clone within a berkely db 
transaction, or that providing the database is transactional, the ZFS clone 
can be performed any time and the recovery procedures will guarantee the 
consistency of the data. I think you mean the latter but I just wanted to be 
sure.

Thanks


Jon