Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 4 Sep 2016 14:11:37 -0400
From:      Paul Kraus <paul@kraus-haus.org>
To:        FreeBSD FS <freebsd-fs@freebsd.org>
Subject:   Re: phantom snapshots
Message-ID:  <8B542627-1FBA-469D-BAAE-60C9BB615326@kraus-haus.org>
In-Reply-To: <bbb97a59-df1c-4ffc-f41b-953b0018c903@kateley.com>
References:  <58816ff0-3ab2-cbae-3d50-c4d5e89d9773@kateley.com> <57C9D1FB.6040904@b1t.name> <bbb97a59-df1c-4ffc-f41b-953b0018c903@kateley.com>

next in thread | previous in thread | raw e-mail | index | archive | help

> On Sep 2, 2016, at 3:41 PM, Linda Kateley <lkateley@kateley.com> =
wrote:
>=20
> How did you remove the files in the snapshot? I keep trying but it is =
read-only? Can I remount it read-write?

Snapshots are by design read only copies of the data.

When you create a ZFS snapshot you are (functionally) copying the =
Uberblock for the dataset and preventing any slabs used by it from being =
added to the free list. This is partly why ZFS snapshots have little to =
no performance penalty.

When you clone a snapshot you are effectively making a read-write copy =
of it.

So to delete files from a snapshot you destroy the snapshot itself. If =
the snapshot has already been destroyed, then you _may_ be running into =
the delayed snapshot destruction implemented a bunch of years ago for =
some really good reasons. If a system rebooted in the midst of a =
snapshot destruction, the reboot would not complete until the snapshot =
destruction was completely done. For large snapshots this could be hours =
(or even days).

So the zfs destroy <snapshot> command returns once the snapshot has been =
scheduled to be destroyed, not after the snapshot has been completely =
destroyed. How big are these datasets and snapshots ?




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?8B542627-1FBA-469D-BAAE-60C9BB615326>