Iivo

Finally I had the time to finish the backup configuration during the Xmas vacations. I installed the QEMU guest agent to all VMs I could to enable the snapshot feature in the VM Backup plugin - the snapshot feature is a must. I moved any problematic VMs back to the Array (XFS) and I'll use the old User script backup script for these. For example pfSense has the QEMU guest agent installed and the network interface listing is visible in Unraid, but when the backup starts, an error occurs: "failure: snapshot command failed on pfsense-vdisk2.snap for pfSenseXXX." "failure: snapshot_fallback is 0. skipping backup for pfSenseXX to prevent data loss. no cleanup will be performed for this vm.". I checked that the virtual disk paths are not using /mnt/user, but didn't investigate much further. Edit: under the "Number of backups to keep" tooltip it mentions that if multiple vdisks are present, they need to be sequentially numbered - there is only one vdisk in this VM, but it ends with 2, could maybe be an issue... This solution is good enough for now, let's see if something changes with Unraid 7. Have a happy new year!

Thanks for the suggestions! I'll start experimenting and will get back with the results. The VM Backup plugin is not as "production grade" as I would like it to be (it's in beta and the only developer is quite busy). Due to the plugin status I actually made a feature request for Lime back in 2022 to help make it stable or include similar features in Unraid, and the response was "That's a very requested feature and one we plan on adding in the future, but no ETA for now"

Thanks for the suggestion! The backup pool is not currently ZFS, but I could convert it to be. Do you think these commands would work on Unraid as well or should I use something else: https://serverfault.com/questions/812351/synchronize-zfs-pools-fast ?

Hello! I'm using Unraid 6.12.10 and recently started using ZFS on one of the pools. I'm now facing a small issue with backing up virtual machines stored in the ZFS pool: I haven't found a simple way of making a "consistent" backup of the virtual machine disks. To elaborate, when using XFS I had a scheduled User script like this: echo "Backing up machine XXX" cp --reflink=always "/mnt/disk1/domains/XXX/vdisk1.img" "/mnt/disk1/domains/XXX/vdisk1-SNAPSHOT.img" cp "/mnt/disk1/domains/XXX/vdisk1-SNAPSHOT.img" /mnt/backuppool/virtualmachines rm "/mnt/disk1/domains/XXX/vdisk1-SNAPSHOT.img" It makes a consistent snapshot of the disk image and copies it into another pool, so no matter how hard I screw up the original pool, the backup can be found from the other pool. With ZFS the "--reflink=always" does not work, and I'm afraid that without using it, the copies might become corrupted. I've watched the ZFS videos by Spaceinvader One, where the backup script plays the Imperial march tune and all kinds of cool features, but that seems overly complicated for my use case. I want to KISS. Any tips on what kind of script should I use with ZFS?

I'm running a testing Unraid in Windows with Virtualbox using a bit different steps, maybe simpler. The steps are: create a USB drive normally with the Unraid USB Creator tool (keep the "Allow UEFI boot" checkbox checked) in Virtualbox VM instance settings on the System tab, check the "Enable EFI (special OSes only) checkbox in the same settings window on the USB tab add a USB filter to attach the USB drive to the VM start the VM. It fails to boot but goes into some UEFI BIOS thingy. Select the "Boot Manager" menu option and then choose your USB drive start Unraid in the non-GUI mode if the GUI mode hangs

I think this is a good idea as well. In addition to the whole engine, it would be nice if the RAM usage could be controlled with a GUI selection in individual container settings. It is possible to use the Extra Parameters field to adjust the limits (if you know the correct parameter), but there is a CPU Pinning selection for CPU and would be nice to have something similar for RAM as well.

As this is the top result in Google for "unraid ntp not working", I'll share my problem (and solution) that happened with two systems: due to a specific VLAN configuration, I had two network adapters (eth0 and eth1) so that eth0 was physically disconnected and eth1 contained only VLANs. The NTP service does not seem to be able to find its way through to the Internet (community plugin downloads etc. do work). I connected a cable to the eth0 and disabled VLANs (for eth0), and now NTP works.