Difference between raid1 zfs and ext4 option on installation wizard?

[padi]

Hello,

I'm trying to install proxmox on a second server and I have the followings:

1'st server : 1 x 250G ssd
2'nd server 2 x 2T hdd

on the first server I proceed with default options - ext4
for the second one I select the zfs-RAID1 option for target hdd's and I let 1G free

now the question:
on first server I have:

$lvs
LV VG Attr LSize Pool Origin Data% Meta% Move Log Cpy%Sync Convert
base-300-disk-0 pve Vri---tz-k 16.00g data
data pve twi-aotz-- 140.45g 10.98 1.65
root pve -wi-ao---- 55.75g
swap pve -wi-ao---- 8.00g
vm-100-disk-0 pve Vwi-aotz-- 8.00g data 27.70

$cat /etc/pve/storage.cfg
dir: local
path /var/lib/vz
content vztmpl,backup,iso

lvmthin: local-lvm
thinpool data
vgname pve
content rootdir,images


and on second one I don't have any output for lvs nor vgdisplay, so no volume group pve.

$cat /etc/pve/storage.cfg
dir: local
path /var/lib/vz
content iso,vztmpl,backup

zfspool: local-zfs
pool rpool/data
sparse
content images,rootdir

on the second server I have ( for both disks ) a free 900G space so I create sda4 partition. is there any chances to create la pve vg on that? and after a thinpool? and after a lvm-thin storage?

how can I have the same storage configuration* if I install the proxmox with RAID1 zfs option from installation wizard?
with 'same storage configuration' I mean having local and local-lmv but on a raid1 hdd's on second server

 

[Stefan_R]

In the GUI, select your node and navigate to "Disks" -> "LVM-Thin". There you can create a thin-pool via the button at the top.

Keep in mind that your nodes are not clustered (storage.cfg is different between them), so I'm not quite sure what your use-case for this would be.

 

[padi]

No unused disks ...

 

[Stefan_R]

...which means exactly what it sounds like Make sure your disks are blank (e.g. 'wipefs -a', ensure there's *really* no data on there, it will be lost!) and try again.

 

[padi]

can I do on a single partition? or I need a new physical disk? because both hdd's are in RAID1 ( set in install wizard ) and proxmox is installed on /dev/sda(b)3

I've tried
root@pmx:~# wipefs -a /dev/sdb4
root@pmx:~# wipefs /dev/sdb
DEVICE OFFSET TYPE UUID LABEL
sdb 0x200 gpt
sdb 0x1d1c1115e00 gpt
sdb 0x1fe PMBR
root@pmx:~# wipefs /dev/sdb4

 

[padi]

@Stefan_R I think I will let the node as it is and set thinpool only for others

but, if select the ext4 from the target hdd option menu I will get the thinpool but if the hdd with proxmox will fail the node will fail
if I select the RAID1 from the target hdd option menu I will be able to change de hdd if will fail.

correct?

thanks,
Adrian

 

[aaron]

but, if select the ext4 from the target hdd option menu I will get the thinpool but if the hdd with proxmox will fail the node will fail
if I select the RAID1 from the target hdd option menu I will be able to change de hdd if will fail.

Yes, if you select ext4 in the installer, Proxmox VE will not create any RAID. If you need RAID in this case, you will need a hardware RAID.

If you select ZFS with RAID1, the installer will set up a disk mirror for the root disk and if one disk fails, you should be able to boot from the second one.

 

[padi]

ok, thanks

and with ZFS and RAID1 option with 2 hdd's I cannot have this storage type:
lvmthin: local-lvm
thinpool data
vgname pve
content rootdir,images

right?

 

[aaron]

correct. You will have a ZFS pool. I prefer ZFS as it has a lot of very nice features. But there are a lot of articles on the internet about that.

One thing that makes it nice in regards to PVE is how the free space is used up between the different dataset. See, when you have LVM, you have fixed sizes for the root LV (which contains the root file system and the `local` storage) and the data LV. If you are in the situation that one of the LVs is getting to small while the other has plenty of space, it is tedious to shift that free space around.
ZFS on the other hand does not have such hard borders and the free space is available for all datasets in the pool.

 

[padi]

yep, but on the first node I had only one ssd, so I proceed with the default ext4 and for the second and 3'rd I have 2 x ssd (ZFS+RAID1) and I saw that I cannot have the same config.
maybe I will buy another ssd and reinstall the first node can I reinstall the node from which the cluster was created?

thanks again for clarifications

 

[aaron]

can I reinstall the node from which the cluster was created?

PVE does not care which node created the cluster initially. Just make sure that you remove the node form the cluster cleanly: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_remove_a_cluster_node

 

[padi]

yep, thanks again

have a good day,
Adrian