Backup PVE Host and Restore to new disk?
- Thread starter 1liminal1
- Start date
Backing up a mounted filesystem is always problematic. That why you boot into clonezilla instead of installing it as a package. If you want consistent backup of a running system you would need to snapshot that entire system and then create a backup based on that snapshot.
Ok, so clonezilla didn't work. I have an old Mac Pro for the host and the drive is actually 500GB not the made up 500GB that most disk makers say their drives are.
So I guess now, I have to find another way to do it.
I'm very surprised that Proxmox does not have a recommended way to do this. hosts must fail from time to to time right LOL?
So I guess now, I have to find another way to do it.
I'm very surprised that Proxmox does not have a recommended way to do this. hosts must fail from time to to time right LOL?
I also had not much luck with clonezilla.
The way I did it, was to setup a SystemRescue usb (https://www.system-rescue.org/), ssh is enabled, with password set in 100-defaults.yaml, then I simply insert usb drive in host machine and reboot, ssh in to host, and all drives will then be unmounted. Next I use dd (handle with care!) to make an image of host to another drive which I mount.
The whole process is pretty seamless and headless.
Good luck. Go to the above site at SystemRescue, and its pretty well documented.
The way I did it, was to setup a SystemRescue usb (https://www.system-rescue.org/), ssh is enabled, with password set in 100-defaults.yaml, then I simply insert usb drive in host machine and reboot, ssh in to host, and all drives will then be unmounted. Next I use dd (handle with care!) to make an image of host to another drive which I mount.
The whole process is pretty seamless and headless.
Good luck. Go to the above site at SystemRescue, and its pretty well documented.
Thats why you ...:
A.) ... should run raid with mirrors or parity data. SSDs/HDDs are consumables and will fill sooner or later, so you should run at least a raid1 for redundancy. Then it doesn't really matter if one of your disks is failing, as everything will continue running normally, except that you will have to pay for a new one.
B.) ... could run a cluster, so it's no problem, even when a whole server fails. Everything will continue working while you replace the dead server with new one.
C.) ... should have recent host, guest and data backups. Then you could always restore a backup from a day or hour ago in case A and B are for some reason failing.
Thanks for the info
A. Totally agree with you, but this is just a home server. I'm not made of money or electricity HAHA
B. I am running a cluster actually, I've moved my workloads to the other host. So no issue there
C. All backed up.
Its just a matter of all the configs, I have certs set up. Bonded NICs If I just create a new host in the cluster, how can I backup and restore those configs? I guess thats the solution?
A. Totally agree with you, but this is just a home server. I'm not made of money or electricity HAHA
B. I am running a cluster actually, I've moved my workloads to the other host. So no issue there
C. All backed up.
Its just a matter of all the configs, I have certs set up. Bonded NICs If I just create a new host in the cluster, how can I backup and restore those configs? I guess thats the solution?
You get a new enterprise grade SSD (Samsung PM883 240GB), perfectly fine as a ZFS mirror for your PVE installation, for 37€ only. That one idles at 1.3W, so one of them running 24/7, at expensive 0,40€/kWh electricity price, would just be 0,37€ per month. So costs shouldn't be a problem. You probably already spend way more for that host
You can't just overwrite all the config files. But a backup of "/etc" (especially "/etc/pve") still helps to set up a new node. Another option would be to backup the system disk on block level using something like clonezilla. Another possibility would be to backup the config.db (which mounts the PMXCFS in /etc/pve).
