From: Duncan <1i5t5.duncan@cox.net>
To: linux-btrfs@vger.kernel.org
Subject: Re: btrfs balance {,meta}data to raid5 not working?
Date: Sat, 11 Apr 2015 02:14:44 +0000 (UTC) [thread overview]
Message-ID: <pan$577b$aa7780a2$54c1424a$5329065c@cox.net> (raw)
In-Reply-To: 20150410221049.GE9470@wloczykij
Piotr Szymaniak posted on Sat, 11 Apr 2015 00:10:49 +0200 as excerpted:
> Hi,
>
> I tried today to balance two drive btrfs raid1 to two drive btrfs raid5
> without luck: [snipped]
> Linux 3.19.3
> btrfs-progs v3.19.1
Two points:
1) There is (was?) a known bug with balance-conversion in (near-)current
btrfs. It was broken for a time, and I'm not sure it is fixed yet. I'm
also not sure whether it was a user-side or kernel-side issue, tho I
believe the culprit commit has been traced and posted, so the answer
should be on the back-list if nobody else replies here with more specific
info.
Which presents a problem, since fully working raid5 support is so new.
But for conversion, I /think/ you can use somewhat older versions and do
the conversion, then use current versions that better handle problems for
actual operation. If I only knew which part, userspace or kernelspace,
you have to use an old version of...
But you could try the latest 4.0-rc7+ kernel and see if it works with
that, yet.
2) You specify two drives[1] and an intended conversion to raid5.
Normally/traditionally, raid5 needs three devices to function undegraded,
altho technically, two-device raid5 is possible; it's just effectively a
slow raid1. There has been some discussion around whether btrfs should
enable two-device raid5 or not, but regardless of whether it's actually
/possible/, why would you /want/ it?
2a) If your intention is to keep it two devices, just continue using
raid1, particularly with btrfs where raid1 mode is MUCH more mature and
tested than raid5 mode.
2b) If instead your intention was to convert it to raid5 before upgrading
it to three devices, just add the third device first, then do the balance-
conversion. It'll save quite some time over effectively doing the
balance-conversion twice.
---
[1] Disks/drives/devices. In a modern world of SSDs and virtual devices,
a block device may well be neither a disk nor an actual drive. (Does SSD
refer to a solid state /device/, or a solid state /drive/; it's certainly
not a /disk/? Either way, a virtual device may not in fact be a drive of
any sort at all, while still being a device.) I guess I'm not alone
among experienced users and sysadmins of an earlier era, who find
themselves now trying to retrain themselves to use the more accurate
generic term in most contexts...
--
Duncan - List replies preferred. No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master." Richard Stallman
next prev parent reply other threads:[~2015-04-11 2:14 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-04-10 22:10 btrfs balance {,meta}data to raid5 not working? Piotr Szymaniak
2015-04-11 2:14 ` Duncan [this message]
2015-04-14 12:21 ` Piotr Szymaniak
2015-04-11 20:47 ` Philip Seeger
2015-04-11 20:48 ` Hugo Mills
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='pan$577b$aa7780a2$54c1424a$5329065c@cox.net' \
--to=1i5t5.duncan@cox.net \
--cc=linux-btrfs@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.