[SOLVED] Cant add NFS Storage

[lmonasterio]

Hi everyone. I cant add a NFS storage (172.16.1.62) to my pve (172.16.1.115). I made the NFS Server using linux, and the strange thing is i've mounted (via CLI in the pve node) a shared folder to that server, and its working perfectly. I can create and delete files. With that in mind i can say its not about permissions or firewall rules.
I can share some screens

On my pve-node:




But in the GUI says "scanning" and nothing...

 

[gurubert]

It tries to lookup the NFS exports via RPC (showmount -e 172.16.1.62). If you are running a NFSv4 only NFS server this may not even exist any more.
You should be able to enter the Export path manually.

 

[lmonasterio]

Thanks @gurubert for your quick reply and time! Im using NFSv4 and i put the path manually, but nothing happened:

 

[bbgeek17]

You mentioned that you are running NFS service on Linux server, meaning its self-installed. There are potentially a few issues, for example:
1) you did not install all components (ie rpcbind)
2) you dont have all the ports open on the firewall for discovery.
Discovery is different from direct mount

You have few troubleshooting avenues:
a) try to add the storage using "pvesm" from the shell. It may provide you a better error output than GUI
b) check /var/log/messages and journalctl for errors and timeouts
c) manually edit /etc/pve/storage.cfg to add the NFS storage. You will need to restart some services after that
"systemctl restart pvestatd pveproxy pvedaemon"

This will likely not workaround the underlying issue and you will still see hangs and healthcheck issues.

---

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

 

[lmonasterio]

Thanks @bbgeek17 for your reply and attention. I really appreciate.
Yes, its a self-installed NFS server over Debian. The weird thing is i can access it an use it if i mount the "exports" drives via CLI on my pve node. That is working
"1) you did not install all components (ie rpcbind)": rpcbind and relates services are up and running, both server and client (pve)
"2) you dont have all the ports open on the firewall for discovery. Discovery is different from direct mount": im adding the path manually in the export option (/mnt/webundav) since discovery brings me no output

"a) try to add the storage using "pvesm" from the shell. It may provide you a better error output than GUI": it says not online

when i can use it mounted in pve (i.e. df -h and create/delete files):


"b) check /var/log/messages and journalctl for errors and timeouts": brings me the same output

 

[bbgeek17]

you have not specified v4 in your pvesm add. Clearly v3 health check is not working for you. This is what the code does:

sub check_connection {
   ....
        $cmd = ['/usr/sbin/rpcinfo', '-T', $transport, $ip, 'nfs', '4'];
    } else {
        $cmd = ['/sbin/showmount', '--no-headers', '--exports', $server];
    }

Can you run these commands manually successfully?

---

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

 

[lmonasterio]

you have not specified v4 in your pvesm add. Clearly v3 health check is not working for you. This is what the code does:

sub check_connection {
   ....
        $cmd = ['/usr/sbin/rpcinfo', '-T', $transport, $ip, 'nfs', '4'];
    } else {
        $cmd = ['/sbin/showmount', '--no-headers', '--exports', $server];
    }

Can you run these commands manually successfully?

---

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

Ive got timeout:


And i add the version number manually on storage.conf:



When i mount it manually via CLI in the node, it works, but not as storage...

 

[bbgeek17]

Ive got timeout:

Well, that is the problem. As was mentioned several times in the thread - manually directly mounting is not the same as operating within the design of the PVE storage subsystem. PVE probes the storage every minute or so using above rpc commands. Your storage is not responding to the built-in probes. You have to solve that if you want to use NFS as storage object within PVE.
The same probe is done by UI both for listing the exports and for allowing you to press "ok" - PVE has to determine that storage is online. Yours is not as far PVE is concerned.

---

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

 

[lmonasterio]

Well, that is the problem. As was mentioned several times in the thread - manually directly mounting is not the same as operating within the design of the PVE storage subsystem. PVE probes the storage every minute or so using above rpc commands. Your storage is not responding to the built-in probes. You have to solve that if you want to use NFS as storage object within PVE.
The same probe is done by UI both for listing the exports and for allowing you to press "ok" - PVE has to determine that storage is online. Yours is not as far PVE is concerned.

---

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

Ok. I didnt know or realized that "manually directly mounting is not the same as operating within the design of the PVE storage subsystem". Sorry.

"You have to solve that if you want to use NFS as storage object within PVE". Any idea how can i do that?

 

[bbgeek17]

"You have to solve that if you want to use NFS as storage object within PVE". Any idea how can i do that?

can you run these commands directly on your NFS server against itself? Against 127.0.0.1?

---

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

 

[lmonasterio]

can you run these commands directly on your NFS server against itself? Against 127.0.0.1?

---

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

 

[bbgeek17]

Looks like a firewall, smells like a firewall, blocks like a firewall - must be a firewall.

---

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

 

[lmonasterio]

Looks like a firewall, smells like a firewall, blocks like a firewall - must be a firewall.

---

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

I thought that, but im testing it with firewall disabled...

 

[bbgeek17]

This "can you run these commands directly on your NFS server against itself? Against 127.0.0.1?" was meant as two tests:
against local 172.xxx. and 127.
Either RPC is not listening on appropriate interfaces or you have other issues in your network.
We have reduced the complexity of troubleshooting from Linux NFS/PVE to Linux NFS/Linux Client. There are many NFS troubleshooting guides on the net.
Good luck.

---

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

 

[lmonasterio]

This "can you run these commands directly on your NFS server against itself? Against 127.0.0.1?" was meant as two tests:
against local 172.xxx. and 127.
Either RPC is not listening on appropriate interfaces or you have other issues in your network.
We have reduced the complexity of troubleshooting from Linux NFS/PVE to Linux NFS/Linux Client. There are many NFS troubleshooting guides on the net.
Good luck.

---

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

Yes, thanks @bbgeek17. I agree that perhaps it is not a proxmox-related-issue. Thanks for your time

 

[lmonasterio]

Its a network (switch) problem with NFS traffic. I close the case.