On Wed, 9 Dec 2009, Markus Kovero wrote:
From what I've noticed, if one destroys dataset that is say 50-70TB and reboots
before destroy is finished, it can take up to several _days_ before it's back
up again.
So, nowadays I'm doing rm -fr BEFORE issuing zfs destroy whenever possible.
It stands to reason that if deduplication is done via reference
counting then whenever a deduplicated block is freed its duplication
count needs to be reduced and it needs to be done atomically. Blocks
such as full-length zeroed blocks (common for zfs logical volumes) are
likely to be quite heavily duplicated. That may be where this
bottleneck is coming from.
Bob
--
Bob Friesenhahn
bfrie...@simple.dallas.tx.us, http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer, http://www.GraphicsMagick.org/
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
