[SOLVED] PBS Installer created a RAIDZ1 - how to create a Datastore?

[zaphyre]

Hi, i am quite sure this is a really silly question but I actually banged my head against ZFS-walls (again).

I have a PBS Server freshly installed with 3 disks. I have let the installer create me a RAIDZ1 during installation. Now, I ended up with a PBS installation and under Storage -> ZFS I have one "rpool".

How do I use the space as a datastore? The manual said the installers hould have added the "rpool" automagically as a DataStore? When I add a datastore it needs a "Backing Path"?

So how do I get from the installer provided RAIDZ1 "rpool" on the three disks to a working datastore to backup to?

https://pbs:8007/docs/storage.html#datastore says to manually create a datastore - but what is my absolute path there? sorry, this is confusing to me ;-)

As per the system requirements I understand it should work with the three disks. Or do I need a dedicated disk for the PBS installation itsell?

Thanks for any help.

 

[zaphyre]

Hmmm,

proxmox-backup-manager datastore create rpool /rpool

- creates the datastore - but not sure, if this is the way?

 

[birdy]

....banged my head against ZFS-walls (again).

I hope you're ok?

You need separate disks for OS and storage. As your backup is more important I would take one for OS and 2 (mirrored) for your storage.

So do a new install, pick one disk for the OS and afterwards make a (ZFS) pool with the other two disks (and if you check the box it will be automatically be added as storage too).

 

[oversite]

Hi, i am quite sure this is a really silly question but I actually banged my head against ZFS-walls (again).

I have a PBS Server freshly installed with 3 disks. I have let the installer create me a RAIDZ1 during installation. Now, I ended up with a PBS installation and under Storage -> ZFS I have one "rpool".

How do I use the space as a datastore? The manual said the installers hould have added the "rpool" automagically as a DataStore? When I add a datastore it needs a "Backing Path"?

So how do I get from the installer provided RAIDZ1 "rpool" on the three disks to a working datastore to backup to?

https://pbs:8007/docs/storage.html#datastore says to manually create a datastore - but what is my absolute path there? sorry, this is confusing to me ;-)

As per the system requirements I understand it should work with the three disks. Or do I need a dedicated disk for the PBS installation itsell?

Thanks for any help.

Your datastore is simply a directory. If your pool is called rpool, or if you add another zfs pool on other disks, you can create a directory as /rpool/mydatastore and that is indeed your backing path.

 

[birdy]

Your datastore is simply a directory

You’re right, my mistake I did it the 2-disk way, but if you add a store you can fill in a path on your current (3-disk) pool.

 

[zaphyre]

I hope you're ok?

Hey, yeah thanks , its actually getting better every moment I work with the ZFS ;-) Thanks for your answer! In this server we have a three disk constraint (its some blade chassis) and I thought using the RAIDZ1 would maximise usable space, so i let the installer create the ZFS RAIDZ1 and hoped to later use the space for backups as well - which now - somehow worked by just

proxmox-backup-manager datastore create rpool /rpool

. I was just confused about to use the GUI or not.

 

[zaphyre]

Your datastore is simply a directory. If your pool is called rpool, or if you add another zfs pool on other disks, you can create a directory as /rpool/mydatastore and that is indeed your backing path.

Thanks! Thats helping me understaning this a lot better! So, the installer created rpool (using my three disks as RAIDZ1) and installed the PBS base system on rpool at:

root@backup10:~# zfs list -o name,mountpoint,mounted
NAME              MOUNTPOINT   MOUNTED
rpool             /rpool       yes
rpool/ROOT        /rpool/ROOT  yes
rpool/ROOT/pbs-1  /            yes

So what was missing on my site? Did I hjust had to create a directory under

/rpool

, like for example

/rpool/my-1st-datastore

and then use this path in the "Add Datastore" gui as "Backing Path" - or at the command line:

proxmox-backup-manager datastore create MY-1ST-DATASTORE /rpool/my-1st-datastore

?

Does this make sense?
Again, thanks for help!

 

[oversite]

Thanks! Thats helping me understaning this a lot better! So, the installer created rpool (using my three disks as RAIDZ1) and installed the PBS base system on rpool at:

root@backup10:~# zfs list -o name,mountpoint,mounted
NAME              MOUNTPOINT   MOUNTED
rpool             /rpool       yes
rpool/ROOT        /rpool/ROOT  yes
rpool/ROOT/pbs-1  /            yes

So what was missing on my site? Did I hjust had to create a directory under

/rpool

, like for example

/rpool/my-1st-datastore

and then use this path in the "Add Datastore" gui as "Backing Path" - or at the command line:

proxmox-backup-manager datastore create MY-1ST-DATASTORE /rpool/my-1st-datastore

?

Does this make sense?
Again, thanks for help!

Exactly, you got it

 

[zaphyre]

Exactly, you got it

Thanks @oversite !

So, to finish this, the command I already used (see above):

proxmox-backup-manager datastore create rpool /rpool

was no total disaster, it just created the datastore logic, aka its directory structure and files (.chunks, ct, vm, ...), directly under /rpool - which as far as I understand, should be okay.

 

[fabian]

you really should not use the rpool dataset itself for anything. I'd suggest creating a new dataset for your datastore:

zfs create rpool/datastore -o mountpoint=/where/you/want/your/datastore

and then creating the datastore using the GUI with the path /where/you/want/your/datastore (obviously, replace that with YOUR chosen path for both actions

the benefits of having a separate dataset for your datastore are that you can change zfs settings for it, and can snapshot it on its own.

 

[oversite]

Thanks @oversite !

So, to finish this, the command I already used (see above):

proxmox-backup-manager datastore create rpool /rpool

was no total disaster, it just created the datastore logic, aka its directory structure and files (.chunks, ct, vm, ...), directly under /rpool - which as far as I understand, should be okay.

No total disaster but you would benefit by having it under it's own directory as soon as you would like to create another datastore on the same pool.

 

[zaphyre]

Okay, understood. For the moment thats fine. I`ll have this right on the production-setup then. Thanks again.

 

[ElDark258]

Just to complement this post (sorry for revive, but imo it's worth it as I was struggling a bit with this and more people might be in the same place I was). After creating your ZFS, it might possibly not be mounted right away, in which case you will want to:

1. Make sure your ZFS points at where you want it to be mounted

zfs set mountpoint=/mnt/datastore Data​

2. Mount it, if it's your only datastore, it's totally fine to use -a like

zfs mount -a​

3. Verify it's mounted by using "zfs list -o name,mountpoint,mounted" as @zaphyre showed previously
4. Create your directories as shown before by Zaphyre as well.