On Thursday 20 August 2015 08:21:22 Nikolaus Rath wrote:

> > What would be better to do for creating "snapshots" in s3ql? I can use the
> > s3qlcp command to copy everything in a new directory or I can use  rsync
> > to a new directory?
> 
> You can use either, but s3qlcp is orders of magnitude faster.


> 
> > Since s3qlcp exist I would say use the s3qlcp command, but what would be
> > the difference, since deduplication is on anyway?
> 
> rsync will still read and write every file. S3QL will then calculate a
> checksum, and throw all the written data away. s3qlcp will not read or
> write anything.

Good point indeed. Didn't think about that part.
> 
> > What I am thing of going to do is:
> > - create a date labeled btrfs snapshot
> > - rsync that snapshot to s3ql.
> 
> Here rsync will copy everything (because it doesn't know about
> de-duplication).
> 
> > or would it be better to:
> > - create a date labeled btrfs snapshot
> > - create a s3qlsnapshot with s3qlcp
> > - rsync the btrfs snapshot into a main dir
> 
> Here rsync will only copy whatever has changed since the previous
> snapshot, so this will be much faster.

Thanks for the explanation! The latter it will be than to go forward 

Cheers,
Sjoerd


-- 
You received this message because you are subscribed to the Google Groups 
"s3ql" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to