Heyho!
I''m currently (once more) trying to experiment with btrfs, especially
since
the ENOSPC handling and "delete snapshots/subvolumes" make it look
extremely
appealing now.
Some questions:
* how do I list available snapshots/subvolumes?
* assuming a disk has failed: just to be sure:
"btrfs-vol -r missing" will remove the missing device. Not sure:
will it
also replicate the missing copies of my (raid1) data and metadata? Given that
it takes ages (ok, just running a few minutes now ;-) to run on this system,
I''d assume so. Or is a separate "btrfs-vol -b" necessary?
* "subvolume" and "snapshot" is technically the same,
except that a subvolume
is created empty and a snapshot is a cow copy of an existing tree. Handling
these two is the same once they''re created?
Ok, that''s what I can think of right now. Eagerly waiting for
"set default
subvolume to be mounted" to be written... :-)
I just did a few tests with a 4 x 450G raid 1 btrfs volume: adding some 5G of
small files, remove a disk (didn''t pull it out, but shut down the
system
first), remove failing disk, re-add a disk. So far it seems to work just
fine. Just a curiosity: it seems to distribute all data between only two of
the disks, the other two stay empty. Is that desired behaviour?
Wouldn''t it
be more efficient to distribute data across all disks?
(everything with 2.6.32-rc8. Filesystem was created with btrfs-tools 0.19.)
cheers
-- vbi
(((
P.S.: since recovery after disk failure is a tense moment where I don''t
want
to have to hunt around for the answer ...:
--- a/btrfs-vol.c
+++ b/btrfs-vol.c
@@ -53,6 +53,8 @@ static void print_usage(void)
fprintf(stderr, "\t-a device add one device\n");
fprintf(stderr, "\t-b balance chunks across all devices\n");
fprintf(stderr, "\t-r device remove one device\n");
+ fprintf(stderr, "\t use ''missing'' as device
name "
+ "to remove a failed device\n");
exit(1);
}
)))
--
featured link: Debian Bookmark Collection - http://bookmarks.debian.net/