Why proxmox creates /dev/md/proxmox:md0 AFTER i created a sw RAID-5 on /dev/md/md0 ?

J

jmar83

Active Member
Nov 26, 2015
87
0
26
42
CH
www.cannabislegal.xyz
Hi there

As title says: Why proxmox creates /dev/md/proxmox:md0 AFTER i created a sw RAID-5 on /dev/md/md0 ?

Then, i fomatted it with "btrfs" filesystem (makefs.btrfs) and created a mountpoint on /mnt/raid5

1636644841700.png
Thank you very much for your feedbacks. :-)
 
MD-based software raid is not and will not be a part of PVE. This is always stated as the topic mdadm-realted stuff is coming up.
This is pure Debian and the name you see here is your hostname. You can easily check this in a Debian 11 Bullseye environment:

Code: 
root@test-for-jmar83 ~ > mdadm --create /dev/md0 --level 1 --raid-devices=2 /dev/sdb /dev/sdc
mdadm: Note: this array has metadata at the start and
    may not be suitable as a boot device.  If you plan to
    store '/boot' on this device please ensure that
    your boot-loader understands md/v1.x metadata, or use
    --metadata=0.90
Continue creating array? yes
mdadm: Defaulting to version 1.2 metadata
mdadm: array /dev/md0 started.

root@test-for-jmar83 ~ > cd /dev/md/

root@test-for-jmar83 /dev/md > l
insgesamt 0
lrwxrwxrwx 1 root root 8 11. Nov 18:28 test-for-jmar83:0 -> ../md0

Why are you using mdadm as basis of btrfs? Why don't you just use ZFS it direclty use a similar method called raidz1 and is fully supported on PVE. You can just install from the official iso to a working raidz1 ZFS pool.
 
Thank you for your feedback.

I don't have any experiences with ZFS - so i better stay with the "old school" linux software raid by "mdadm"...

I never saw that /dev/md/[HOSTNAME]:md0 thing before. And i used "mdadm" on many many distributations. But not (yet) on Debian 11 "bullseye"... maybe that naming standard is new in Debian 11 "bullseye"...?
 
 
AND: Also important i.m.o.: After i formatted the raid 5 array with btrfs and used is as a vm storage, it was very very slot... and in the middle of the installation it freezed. The shutdown of the vm needed many many minutes...

Then i formatted the raid 5 array with ext4 - no performance problems and no freezing anymore.

Why it's is so slow and it freezes with btrfs??

I think btrfs now (2021) is stable - or int's it yet? (Maybe it's dangerout to use it - and better to avoid in general?? :eek: )
 
Like LnxBil already said....if you want all that nice modern features like copy-on-write, snapshots, block level compression, deduplication, bit rot protection and so on why not just try to learn ZFS? It got all the btrfs features, is more mature/stable than btrfs and also got raid integrated so you don't need HW/SW raid anymore.

Isn't that hard to learn it. Just make sure you choose the right ashift and volblocksize for your pool layout and don'T fill your pool up more than 80%. For the remaining stuff the defaults should work fine.
 
Last edited:
OK thx, but i [repectively the company i work for] don't have the time to learn all that new things. maybe one day.

But currently, the EXT4 mdadm sw raid-5 works welll :-)
 
jmar83 said:
OK thx, but i [repectively the company i work for] don't have the time to learn all that new things. maybe one day.
Click to expand...
Sounds counterintuitive to me. You're willing to learn btrfs but not zfs? Btw brtfs is a reimplementation of the zfs ideas by oracle, which then just bought SUN - the company behind zfs. btrfs has also all the cool features, but you do not install it on top of mdadm. Also, btrfs has not a fully trusted raid5 implementation, but mirroring and striping is fully supported and works well.
 
OH - maybe i understood the concept of BTRFS wrongly?

Until now, i used BTRFS like every other file system (like EXT[n]...) so i didn't done someting "special" - just created the RAID array, then formatted it with btrfs.make... then i mounted it and used configured the mount point as storage in Proxmox Web GUI. Then i saw that this construct is very very slowly AND instable...

" Also, btrfs has not a fully trusted raid5 implementation"

Now i understand a lil bit more - thank you very very much!!
 
You must log in or register to reply here.
Top Bottom
Back