| View previous topic :: View next topic |
| Author |
Message |
sdauth
l33t
Joined: 19 Sep 2018
Posts: 658
Location: Ásgarðr
|
 Posted: Fri Nov 22, 2024 11:23 pm Post subject: [SOLVED] btrfs raid5/6 status ? |
 |
|
Hello,
Anyone using btrfs raid5/6 ? (I plan to do a raid6 with 6x hdd, currently on a 4x hdd btrfs raid10, without issue)
I read it is considered dangerous online but maybe it is not now ? (Almost 2025)
Or should I forget about it and go for mdadm/lvm + xfs/ext4 instead ?
Thanks.
Last edited by sdauth on Sun Nov 24, 2024 5:06 pm; edited 1 time in total |
|
| Back to top |
|
 |
CaptainBlood
Advocate
Joined: 24 Jan 2010
Posts: 3909
|
| Posted: Sat Nov 23, 2024 2:17 am Post subject: |
|
|
IIRC it's far from being declared stable.
And chances are it won't be for a long while...
btrfs.readthedocs.io
Thks 4 ur attention, interest & support.
_________________
USE="-* ..." in /etc/portage/make.conf here, i.e. a countermeasure to portage implicit braces, belt & diaper paradigm
LT: "I've been doing a passable imitation of the Fontana di Trevi, except my medium is mucus. Sooo much mucus. " |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3746
Location: Rasi, Finland
|
| Posted: Sat Nov 23, 2024 5:49 am Post subject: |
|
|
You should then go with btrfs-raid1 with three copies.
That'll guarantee you raid6 like failure protection.
Although lvm+xfs isn't a bad choice either.
_________________
..: Zucca :..
| My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
sdauth
l33t
Joined: 19 Sep 2018
Posts: 658
Location: Ásgarðr
|
| Posted: Sat Nov 23, 2024 12:16 pm Post subject: |
|
|
| CaptainBlood wrote: | IIRC it's far from being declared stable.
And chances are it won't be for a long while...
btrfs.readthedocs.io |
I read too fast!
I saw "Scrub + RAID56 mostly OK mostly OK" but missed the part below  (Block group profiles / RAID56 : unstable)
Well.. 
| Zucca wrote: | You should then go with btrfs-raid1 with three copies.
That'll guarantee you raid6 like failure protection.
Although lvm+xfs isn't a bad choice either. |
I've thought about that but I really need the extra tb with a little safety net. At least until I can get a 12tb hdd to make a full mirror of that array. (hopefully sooner than later)
In general I always make backup, including for raid just in case of catastrophic failure (I lost a raid10 5 years ago so I now I know  ) but anyway this new array will only have archived stuff so it can wait a little bit.
I think I'll stay reasonable and go for mdadm + lvm/xfs then. (Hopefully dedup is as good as with btrfs!) |
|
| Back to top |
|
|
CaptainBlood
Advocate
Joined: 24 Jan 2010
Posts: 3909
|
| Posted: Sat Nov 23, 2024 12:47 pm Post subject: |
|
|
What is/are the feature(s) you want to benefit from?
Side note. there is a link in the btrfs doc that dives in a little more.
Maybe u got there already... 
Thks 4 ur attention, interest & support.
_________________
USE="-* ..." in /etc/portage/make.conf here, i.e. a countermeasure to portage implicit braces, belt & diaper paradigm
LT: "I've been doing a passable imitation of the Fontana di Trevi, except my medium is mucus. Sooo much mucus. " |
|
| Back to top |
|
|
sdauth
l33t
Joined: 19 Sep 2018
Posts: 658
Location: Ásgarðr
|
| Posted: Sat Nov 23, 2024 1:09 pm Post subject: |
|
|
| CaptainBlood wrote: | | What is/are the feature(s) you want to benefit from? |
Right now, the data sits on multiple hdd. (external too, so painful to access / navigate and slow) The goal is to consolidate that on a raid6 to benefit from higher read speed and by using btrfs, having the possibility to dedup this data. I think I could at least dedup 500GB (at some point in the past, I ran dedup on this particular set of data and was able to save hundred of gb), potentially more. But apparently it is possible to have dedup with xfs too. Although I will have to read a little bit on that since I never used this filesystem yet. Apparently, checksumming is possible with xfs too, but only metadata. But in my case it is not crucial anyway. |
|
| Back to top |
|
|
pingtoo
Veteran
Joined: 10 Sep 2021
Posts: 1288
Location: Richmond Hill, Canada
|
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3746
Location: Rasi, Finland
|
| Posted: Sat Nov 23, 2024 6:17 pm Post subject: |
|
|
| sdauth wrote: | | I really need the extra tb with a little safety net. |
btrfs-raid1 with three copies guarantees that for each file you have, it'll store two more copies on two separate disks. So it'll give you around the same amount of space as RAID6. I don't know how many copies does btrfs-raid1 supports nowdays, but I think it'll go to four, or even more. So four means you can lose three disks without data loss.
The original btrfs-raid1 (which is the default, I think) distributes each file twice on different disks, so that's like RAID5.
btrfs "raid" works on file-level, not on physical disk level.
I'd still be using btrfs mainly if it weren't so slow at scrubbing and if it could host swap files on multidisk volume.
_________________
..: Zucca :..
| My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
sdauth
l33t
Joined: 19 Sep 2018
Posts: 658
Location: Ásgarðr
|
|
| Back to top |
|
|
ecko
Tux's lil' helper
Joined: 04 Jul 2010
Posts: 111
|
| Posted: Sat Nov 23, 2024 9:13 pm Post subject: |
|
|
I think your values are correct.
With raid1, it will use 3 drives to store and 3 drives as a mirror. The usable space is therefore only 3 drives (9 Tb). That means a space utilization of 50% as in the table in the link you quoted. Since there is a full mirror of each drive, a raid1 array is tolerant to 1 fault.
With raid1c3, it would use 2 drives to store, 2 drives as a mirror, and 2 drives as another mirror. giving 6 Tb of usable space (33% of the total), and resilience to 2 faults.
Raid5/6 distribute the data and copies on all the drives, and manages to have a space utilization of 75%, which gets 12 Tb out of the total 18. |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3746
Location: Rasi, Finland
|
| Posted: Sun Nov 24, 2024 7:20 am Post subject: |
|
|
Yes. I've totally counted the numbers wrong. It's been a while since I used btrfs. Sorry for the confusion.
If you want the speed and space efficiency, then use mdraid (or LVM).
_________________
..: Zucca :..
| My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
sdauth
l33t
Joined: 19 Sep 2018
Posts: 658
Location: Ásgarðr
|
| Posted: Sun Nov 24, 2024 5:06 pm Post subject: |
|
|
@Zucca: No problem 
I finally end up setting up a mdadm raid6 with luks on /dev/md0 & a single xfs partition on the mapper, it took a while to fully resync
Maybe I could have use lvm but it seemed a bit overkill after second thought. I don't intend to do any kind of resize or put root on it (an old ssd is good enough for that)
Now moving the files on it.. |
|
| Back to top |
|
|
Zucca
Moderator
Joined: 14 Jun 2007
Posts: 3746
Location: Rasi, Finland
|
| Posted: Sun Nov 24, 2024 5:26 pm Post subject: |
|
|
XFS is a good choice for big partitions.
Since it supports reflinking nowdays, you can create snapshot -like copies.
And I think deduplication is also on the menu.
_________________
..: Zucca :..
| My gentoo installs: | init=/sbin/openrc-init
-systemd -logind -elogind seatd |
| Quote: | | I am NaN! I am a man! |
|
|
| Back to top |
|
|
|
Other Things Gentoo
