[PROXMOX CLUSTER] Add NFS resource to Proxmox from a NAS for backup

[avaaru]

I've encountered a synchronization and availability issue when using NFS network storage within a Proxmox VE cluster.

- Environment: Proxmox VE Cluster with 2 nodes (node 01, node 02).
- Storage: Synology NAS shared via NFS.
- Configuration: The connection to the NAS is made through a public IP address.

- Symptom 1: When adding the resource to /etc/pve/storage.cfg, I get the following error:
`create storage failed: storage 'NAS_Sysnalogy_LEGANES' is not online`
The command I'm using to add it is:
pvesm add nfs NAS_Sysnalogy --server IP_NAS --export /volume2/RUTE_NAS --content backup,images,rootdir,iso --options vers=3,soft,timeo=300,proto=tcp


- Symptom 2: The problem isn't related to NAS connectivity or NFS permissions, as a manual mount test (`mount -t nfs IP:/volume2/ROUTE_NAS /mnt/ NAS_Sysnalogy`) works correctly.
If I add it manually, it's added as a DIR within /etc/pve/storage.cfg, and it causes the backups made by Node 1 to be invisible on Node 2, and vice versa, even though it's the same shared resource.


Why can't I add it from Proxmox but I can with the mount command?
I understand that NFS is working correctly and it's not a port or NAS problem since I can add it using mount.

 

[avaaru]

HELP

 

[Moayad]

Hi,

Do you mount the the correct export path? How about if you get rig the version or options flags?

Could you please also run the `showmount -e IP_NAS` to confirm the exact export?

 

[avaaru]

Hello.

Thanks for your reply

If the path is the same as the one I configured in the mount command and I select the NFS 3 protocol, which is the one I have configured on my Synology



If I run the command showmount -e IP_NAS, it just stays like that, as if it's thinking, and doesn't end up returning anything.



However, if I run this mount inside my fstab:



It mounts the folder correctly.



Do you need any more information?

Hi,

Do you mount the the correct export path? How about if you get rig the version or options flags?

Could you please also run the `showmount -e IP_NAS` to confirm the exact export?

 

[Moayad]

Thank you for the output!

If I run the command showmount -e IP_NAS, it just stays like that, as if it's thinking, and doesn't end up returning anything.

Could you please also run the following commands on your PVE node?

pvesm nfsscan IP_NAS
rpcinfo -p IP_NAS

And see the output if have any timeout message?

 

[avaaru]

After waiting a few minutes, it shows me the following errors:

 

[Moayad]

Thank you for the output!

This show that the NAS isn't answering, please check your NAS or firewall, you can try nmap port 111, 2049 to see if they are allowed from your Proxmox VE node.

 

[avaaru]

They are open according to the nmap results. Furthermore, if they weren't open, they wouldn't mount with mount from Proxmox, right?



I don't understand why mount doesn't reject the connection but proxmox's NFS does.

 

[Moayad]

I don't understand why mount doesn't reject the connection but proxmox's NFS does.

The manual mount uses NFSv3 or NFSv4?

UPDATE: i.e., could you try `mount -o vers=3 ....`

 

[avaaru]

My command does not specify the version: IP:/volume2/Backup-Proxmox /mnt/NAS_Sysnalogy nfs defaults 0 0

I modified the command for version 3 and it still works. I've attached a screenshot of the mount and ls commands showing that the content is visible.

IP:/volume2/Backup-Proxmox /mnt/NAS_Sysnalogy nfs nfsvers=3 0 0

 

[bbgeek17]

There are two operations that are happening in the background when adding NFS storage pool in PVE:
- health probe
- mount

When you are doing manual mount, the health probe is skipped.
In PVE case the health probe is a "showmount" command: https://github.com/proxmox/pve-storage/blob/master/src/PVE/Storage/NFSPlugin.pm#L195

You've confirmed via multiple checks that there is an RPC configuration mismatch. Your showmount hangs, rpcinfo times out, etc.
I would inquiry from Synology support what you need to do to enable those services.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[avaaru]

But I've seen videos online showing people with Synology NAS devices, configured with the same settings as mine, getting it to work with Proxmox. I understand it shouldn't be a problem with the NAS, right?
Could I run some tests using the GitHub link you provided?
Thank you so much for all the time you're investing in my problem.

There are two operations that are happening in the background when adding NFS storage pool in PVE:
- health probe
- mount

When you are doing manual mount, the health probe is skipped.
In PVE case the health probe is a "showmount" command: https://github.com/proxmox/pve-storage/blob/master/src/PVE/Storage/NFSPlugin.pm#L195

You've confirmed via multiple checks that there is an RPC configuration mismatch. Your showmount hangs, rpcinfo times out, etc.
I would inquiry from Synology support what you need to do to enable those services.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[VictorSTS]

Configuration: The connection to the NAS is made through a public IP address

It's unclear if you are using some VPN or direct connection using public IPs (hope not, NFS has no encryption), but maybe there's some firewall and/or NAT rule that doesn't allow RPC traffic properly? Maybe your synology uses some port range for RPC and those aren't the same on the FW?

 

[avaaru]

It's unclear if you are using some VPN or direct connection using public IPs (hope not, NFS has no encryption), but maybe there's some firewall and/or NAT rule that doesn't allow RPC traffic properly? Maybe your synology uses some port range for RPC and those aren't the same on the FW?

In my case, it's through a public IP address, since I have no other way to connect the NAS for backups to my cloud server. What other method would you recommend? I understand that SMB would be less secure.

 

[bbgeek17]

RPC is done over UDP. Routing UDP over the internet is ... tricky. Either create a VPN-LAN, or find another way.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[VictorSTS]

Can't really recommend anything specific without infrastructure details, but I would definitely use some VPN and tunnel NFS traffic inside it, both for obvious security reasons and ease of management on the WAN side (you'll only need to expose the VPN service port to the internet). Now that you ask, NFS over WAN performance is terrible. An I really hope you have a QDevice for that 2 node cluster so you can keep quorum when one host is down

 

[avaaru]

La RPC se realiza mediante UDP. Enrutar UDP por internet es... complicado. Se debe crear una VPN-LAN o buscar otra solución.

---

Blockbridge: Almacenamiento compartido NVME de latencia ultrabaja para Proxmox - https://www.blockbridge.com/proxmox

That could be it. I don't have any open UDP protocols; they're all closed. Is that why I can't perform the RPC phase?

 

[avaaru]

Podría ser eso. No tengo ningún protocolo UDP abierto; todos están cerrados. ¿Es por eso que no puedo realizar la fase RPC?

I have a dedicated server on a cloud computing platform. This dedicated server only has a public network card, and I don't have the option to add a private network card. I also don't have the option on the cloud computing platform to create a VPN tunnel between the server and my firewall. The NAS would only be used to back up the machines I have running on Proxmox.