[SOLVED] OpenVZ Container on NFS Share error with exit code 48

[kerios213]

I cannot create an OpenVZ Container within my NFS Store.
I have created several KVM's within the store without any problems.

The error code upon creation of the container using a template downloaded from the store

Creating container private area (/var/lib/vz/template/cache/centos-6-standard_6.3-1_i386.tar.gz)
chmod: changing permissions of `/mnt/pve/proxmoxNFS/private/113.tmp': Operation not permitted
tar: ./usr/lib/libdns.so.81.4.1: Cannot open: Operation not permitted
tar: ./usr/lib/libpth.so.20: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/lib/libtiff.so.3.9.4: Cannot open: Operation not permitted
tar: ./usr/lib/libnsssysinit.so: Cannot open: Operation not permitted
tar: ./usr/lib/libisccfg.so.82.0.1: Cannot open: Operation not permitted
tar: ./usr/lib/libtiffxx.so.3: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/lib/libffi.so.5.0.6: Cannot open: Operation not permitted

I am using Freenas 8.3.
How can i solve this issue ?
Thanks

 

[mir]

Is the Freenas storage based on ZFS?

If ZFS simply add this to your option line in /etc/pve/storage.cfg
noacl

 

[kerios213]

Yes,
No above exit code 48 error but new error:

TASK ERROR: can't lock container 102

I am trying to fix.
But after restart the Proxmox Cluster i saw the error exit code 48 again

tar: ./usr/bin/snmpinform: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/bin/strings: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/bin/xxd: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/bin/syslinux2ansi: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/bin/db5.1_deadlock: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/bin/head: Cannot change ownership to uid 0, gid 0: Operation not permitted
tar: ./usr/bin/groff: Cannot change ownership to uid 0, gid 0: Operation not permitted

The config in /etc/pve/storage.cfg still left :

nfs: vm-nfs-01
path /mnt/pve/vm-nfs-01
server 192.168.40.148
noacl
export /mnt/vol1/images-nfs
options vers=3
content images,rootdir
maxfiles 1

 

[mir]

You need this change:
from -> options vers=3
to -> options vers=3,noacl

And to be certain the changes has taken effect you will need to reboot the proxmox host.

 

[kerios213]

Thanks for your help but it still has this error.
Do i have to add a no_root_squash to /etc/export
I have already tried this but was not successful.

 

[mir]

What are the permissions on Freenas for the share?

 

[kerios213]

Yes,
I have 2 datasets were set to 777.

 

[kerios213]

And some pics from SAN and Proxmox Configuration files :

 

[mir]

Try changing permissions on the dataset from root:wheel to root:root

 

[kerios213]

Thanks Mir,
But there is no option to change the group owner to root.


The group is not valid
And I tried to edit by command line in the SAN BOX:
[root@freenas]chown -R root:root /mnt/vol1/images-nfs

and i received "Chown root invalid argument"

 

[mir]

See last post in this thread: http://forum.proxmox.com/threads/16595-Cannot-mknod-Input-output-error-on-CT

 

[kerios213]

Yes,
It works perfectly.
I want to say thanks.
If anybody has this problem just do as following instruction :


1. Creat a dataset with correct permission(maybe 775 or 777).
2.Share via NFS on freenas but pay a attention to the permission on this :
Maproot User: N/A
Maproot Group: N/A
Mapall User: root
Mapall Group: wheel
3. Add a noacl to the /etc/pve/storage.cfg.
Done

 

[lucaprete]

Hello, I'm running into the same issue with my actual deployment and I don't understand why - even after doing steps described above things still don't work.
I've the latest version of ProxMox (3.4) and the latest version of freenas (9.3).

On freenas I've zfs volumes + dataset + NFS share.

Few questions:
- What's the real problem behind this? I know the cause and the possible practical solution, but I didn't understand why this is happening
- Should the permissions on freenas side be really 777 or 775? What's the reason for this?
- You configured the user as root and the group as wheel. As far as I imagine, any combination of user and group should be fine as long as it's consistent between what has been setup in the dataset and in the NFS share mapall parameters. Is this assumption correct?
- I added noacl to storage.cfg, replacing options vers=3 with options vers=3,noacl. Is this correct?
- I'm running a proxmox cluster with different nodes. When I modified storage.cfg I saw the modification happen automatically (without restarting anything) also on the other nodes - I imagine because of the distributed store used by proxmox. What should I do to have the mod apply on the single node and/or in the entire cluster? Is it ok not rebooting? Should I reboot node by node?

Anyway, even following what was reported above didn't work. Any other thought?

Thank you!

Luca

 

[mir]

Did you also follow this advise?
http://forum.proxmox.com/threads/16595-Cannot-mknod-Input-output-error-on-CT?p=97100#post97100

 

[lucaprete]

Thanks!

I didn't and now, after trying it works.
What I did at the end was:
- Set my user and group on freenas, consistently between the dataset and the nfs share mapall parameter
- My permissions on the dataset are 700 (no need of 777 or 775)
- I modified on one of my nodes in the cluster the file /etc/pve/storage.cfg
---- from "options vers=3" to "options vers=3,noacl,aclmode passthrough,aclinherit passthrough-x"

The following questions still remain open:
- What's the error happening behind this behavior
- What the parameters noacl,aclmode passthrough,aclinherit passthrough-x mean (and not sure if all of them are needed)

FYI, anyway backing up and restoring a CT is really slower compared to a VM. I don't see differences instead for live migration.

Thanks a lot!

Luca
Luca