NFS not mounted after reboot

[Morten]

Our PVE cluster is backing up to a NFS share that mounts perfectly fine when I open the web interface on the cluster node and click "Storage". After rebooting the server the share is not mounted which I know is as intended.
But the backup process doesn't always mount the NFS itself which causes the backup to store the backup files locally in the path that corresponds to the share mount point.
Here's the line from the correctly mounted share:

192.168.9.5:/mnt/vg10k/xfs10k1/nfs10k1 on /mnt/pve/backup type nfs (rw,addr=192.168.9.5)

When the NFS is not mounted the backup files are located in /mnt/pve/backup.
The NFS share is located on a Openfiler server and is rarely (never) rebooted.
What should I do?

pve-manager: 1.5-9 (pve-manager/1.5/4728)
running kernel: 2.6.24-11-pve
proxmox-ve-2.6.24: 1.5-23
pve-kernel-2.6.24-11-pve: 2.6.24-23
qemu-server: 1.1-14
pve-firmware: 1.0-5
libpve-storage-perl: 1.0-13
vncterm: 0.9-2
vzctl: 3.0.23-1pve11
vzdump: 1.2-5
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.12.4-1

 

[dietmar]

Why does the mount fail? Any hint in the backup log?

 

[hk@]

hi
we actually have a similar problem here - we do use rsnapshot in order to keep previous backups for some time.
Now nfs is not always mounted (stuipd question: why is this intended?) - so rsnapshot trying to rotate the directories on the nfs-mount doesn't always work well.
is there a way to get the nfs mounts active before doing rsnapshot?

also vzdump offers some commandline-options like bwlimit and maxfiles - I guess the only way to utilize those is by using the /etc/vzdump.conf on a global basis?

and finally it would be most conventient if we could not only define maxfiles, but rsnapshot like rotation (daily, weekly, monthly) somehow.

regards
hk

 

[dietmar]

hi
we actually have a similar problem here - we do use rsnapshot in order to keep previous backups for some time.
Now nfs is not always mounted (stuipd question: why is this intended?) - so rsnapshot trying to rotate the directories on the nfs-mount doesn't always work well.
is there a way to get the nfs mounts active before doing rsnapshot?

You can use 'pvesm' (PVE storage manager) to activate the storage:

# pvesm enable <STORE_ID>

also vzdump offers some commandline-options like bwlimit and maxfiles - I guess the only way to utilize those is by using the /etc/vzdump.conf on a global basis?

Yes, or pass it directly to the vzdump command.

and finally it would be most conventient if we could not only define maxfiles, but rsnapshot like rotation (daily, weekly, monthly) somehow.

It is somehow unclear howto handle multiple backups with 'daily, weekly, monthly' - suggestions?

 

[JustaGuy]

It is somehow unclear howto handle multiple backups with 'daily, weekly, monthly' - suggestions?

I've yet to implement a complete solution as of yet, however I've collected a few ideas so far along with various components which wait for me to find time to install & configure properly.

One of which is a mechanism with which to automate offline backups similarly to how tape drives have traditionally been in use, except with a stack of large external USB drives instead.

This in order to mitigate the risk inherent in having automated data propagation systems share an online presense in one way or another.
All networked systems are thus vulnerable to the attack vector which involves a core breach.

I found a relatively affordable and featureful product called SyncBackPro, which, among other things employs job trigger logic based on the presence of HDD serial numbers (AND these serials' mountpoints if that's what you need).

The day's drive can be plugged into a Windows workstation at the beginning of the day, import the contents of a share, check the integrity of written data, run a script & eject. No HID required.

A manager can then rotate these drives to a fireproof safe, or wherever.




As for how to lessen the complexity of implementing elements of a robust backup scheme, I'm at a loss as I imagine many are. It's a big job.
I like to hear people discuss it here in the forum, to exchange ideas on the subject for the benefit of us all.