Some questions about making effective backups

Hi, folks:
I have a question about how I would do this should I ever have the need to.
My server, is a box in a datacenter that uses RAID1 by default.
The partitions it uses are actually mounted using LVM.
So filesystems are made in LVM etc.
I’m wondering something.

In the event that the hardware ever failed, hopefully prior to this I’d make a backup like a raw filesystem dump (and,) I would, of course, turn off MySQL during such a situation, so as to avoid the situation Joe warned of when discussing the backup and restore feature, that is, untrusted databases that you can’t use.

But I’m wondering what ways to restore a backup to this remote box would be best?

I can, have the machine booted if I ask the provider into a rescue environment.

Wouldn’t it be best to go into the webmin section, choose the backup thing an dmake a backup that way, and then restore?

I was wondering one thing though.

In my LVM filesystem, I expanded the logical volume for /home to take up all free space.

Then, in the case of restoring from a dump, would I need to do this again if restoring virtualmin etc?

Well, first of all, please note that having a RAID is no reason for not making regular backups. :slight_smile: RAID helps with single-disk failure, but won’t help you at all in case of any other problems, e.g. malware infection, accidental file deletion, dual-disk failure, physical damage to the server etc.

For my Virtualmin hosting servers, I use two “levels” of backups. Every day, an incremental backup of all virtual servers is made through Virtualmin’s backup function. Once per month, a full backup of the virtual servers is made.

Additionally, I use the “Filesystem Backup” module of Webmin to create a TGZ archive of the server’s root file system, also once per month. The active MySQL binary data files are not included in this, because I have separate file systems for / and /home; MySQL data directory resides under /home. Everything under /home is virtual server stuff and gets backed up by Virtualmin.

So in case of catastrophic server failure, I just need to partition the new disk, extract the TGZ archive, and install the boot loader. Then use Virtualmin’s restore feature to pull in the incremental backups.

Thanks for the tips.
This is the first server I’ve ever operated that was configured with RAID.
And yes, I know that RAID in and of itself won’t save me from total disk destruction, etc.
AndI do have to accept that sooner or later my server will probably have something like this happen throughout it’s lifetime where the provider will have to replace at worst the entire drive and ask me to re-image the server entirely.
I’m actually responsible for re-imaging, with the provider I’m working with.
They offer a web-based image tool that does automatic network installations of say CentOS, or whatever.

All my provider is responsible for is replacing hardware, and that’s all.

So that being said, what suggestions do you folks have for insuring that all data under LVM is taken into account when doing a full filesystem backup?

Or is webmin’s backup module smart enough to help me with this task.


So that being said, what suggestions do you folks have for insuring that all data under LVM is taken into account when doing a full filesystem backup?

The procedure I outlined (Virtualmin backups and Webmin’s “Filesystem Backup” module) works no matter if your partitions are on LVM, mdadm or directly on a disk.