changing CP's create time

changing CP's create time

[Ethy H. Brito]

Hi

What are the implications of changing the cp_create value for a bunch of
control points?

Motivation: I stumbled in this: 
NILFS: bad btree node (blocknr=115718887) ... 
and I decided to backup and recreate the partition since I could not find
any solution "googgling" for an answer.

(btw: any suggestions is really appreciated to solve this "bad btree" problem)

But when I rsync the data do another nilfs2 partition, snapshot by snapshot,
each "cp_create" is stamped with today's date.
Since I have a script that deletes all CP's older than 60 days, it will be a
problem to guess what do delete as all CP's will have almost the same
timestamp.

I probably doing it the wrong way but nothing clever came to my mind and I
must have this operation done by Tuesday morning when the folks returns to work.

So I decided to thought changing the cp_create values back in time to
resemble the original time of creation and hence the first question.

Again, any suggestions are welcome to save my you know what.

Regards

Ethy
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: changing CP's create time

[Peter Grandi]

> But when I rsync the data do another nilfs2 partition,
> snapshot by snapshot,

Are you using something like '--compare-dest' or '--link-dest'?

> each "cp_create" is stamped with today's date.

RSYNC is a POSIX command, checkpoints are not part of POSIX.

> Since I have a script that deletes all CP's older than 60
> days, it will be a problem to guess what do delete as all CP's
> will have almost the same timestamp. I probably doing it the
> wrong way [ ... ]

Perhaps some meditation on the difference between checkpoints
and snapshots, and the role of checkpoints in NILFS2, might be
of benefit here. Including the realization that you can turn
checkpoints into snapshots *on the target*.

Hint: NILFS2 checkpoints/snapshots are not quite the same as
"checkins" in a version control system or Btrfs snapshots.

> So I decided to thought changing the cp_create values back in
> time to resemble the original time of creation and hence the
> first question.

That could be done by adding some new 'ioctl' and modifying
'rsync' too, but looks messy to me, because checkpoints on the
source and checkpoints on the destination need not happen at the
same intervals, and checkpointing is "automatic" so far.
--
To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html