publish for backup about datalad HOT 2 CLOSED

Yaroslav Halchenko wrote:

@joeyh - what would be your thoughts on how to wrap it up? tags?

Yes, you can already do this with tags.

git annex metadata -t precious foo/ bar/

git annex copy --to externaldrive --metadata tag=precious .

It might be better though, to configure precious data to have a high
numcopies value. This would prevent git-annex dropping a precious file
unless it can verify enough other copies of it exist.

Doing it this way, you'd write a .gitattributes files that matches the
precious files and sets their number of copies to some reasonably high
value. (Depending on how low a chance you want to have of losing the
data, vs the expense to keep additional copies of it.)

*.raw annex.numcopies=5

And then you coud ask git-annex to copy files to external drives until
the files' various numcopies values were satisfied, by using the --auto
parameter.

git annex copy --to externaldrive --auto .

This also has the benefit that you could have, say, 10 drives, and
git-annex will copy each precious file to only 5 of them, and then stop
making more copies. So as you add drives and files over time, you don't
need to worry about manually telling git-annex what files to put where.

With external drives and a high numcopies setting, it can be annoying to
need to plug in lots of drives just so git-annex drop can verify that the
required number of copies have been made. So you might decide to mark
some/all of the external drives as trusted.

see shy jo

from datalad.

I think it is largely superseded by #925 which suggests to provide convenient user interface to define common schemas for siblings

from datalad.