Hi all,
as seen in other posts, I'm getting to know HPE VME and it's options. I notice that there is a built-in capability to backup the appliance itself, and separately to backup every provisioned instance, as needed.
Now, I'm confused about several things in this process, so I'll separate them by numbers.
1. Why don't backup jobs create folders with names that make sense? I.e. - to pull the Instance name in the backup job and create a folder with the name of the instance?
I currently have three folders containing backups and I can only guess as to their purpose, judging from their size. Is the idea that you never have to use these folders from the filesystem?
2. Directly linked to 1. I am now trying to learn how to manage a complete node failure, while the node is holding the VME Manager. Let's say you're that unlucky and you also lost your shared storage, so now you don't even have your VMEM disk, but you magically DO have a backup of the VMEM and the Morpheus database, located in one of the folders seen above. How would I go about restoring VMEM and it's database? How would I even know which files to use, without guessing from filesizes?
Am I missing the ball completely here? Am I even supposed to backup VMEM like this?
I imagine that since the backup job creates a .QCOW2 folder with all files browsable, that this folder can somehow be converted to a QCOW2 image and then attached to a freshly-provisioned empty VM, serving as a quick placeholder. Is this a legitimate method? Or is this not even meant to be used this way? The linked guide found below looks like it may be the way. But, am I even supposed to be thinking in this direction?
https://serverfault.com/questions/246835/convert-directory-to-qemu-kvm-virtual-disk-image
I can obviously see an easier path to just backup VMEM using Veeam and restore it to another node, thus eliminating all these shenanigans. I'm just trying to grasp the intended idea for backup and restore of the VMEM.
3. And then we get to the point of the Morpheus database, which is also confusing me. Namely - if we are intended to backup the VMEM disk using Veeam, why would I ever restore the Morpheus database manually? Am I just overthinking these things? I can imagine installing a fresh, new VMEM on a working node and then restoring the database, but why would you ever do this, if you did your due dilligence and just backed up the disk?
I'd be very happy to talk to someone from HPE about these things via e-mail or a call, if this is the kind of thing you people do...
I'm interested in all of these answers as well. All of this is very new. Trying to understand how you would do recovery from these backup files as well as the "Morpehus Appliance" backups which I'm assuming is the Morpheus database?
Also second the notion of naming the folders based on jobs. Would also like to see these stored with it's own folder instead of all on the root. Morpheus-virtual-images gets it's own root folder, why not the backups? This will just get cluttered over time.
Also I've seen instances of where I delete a backup job from the UI and it not removing the related backup folders on the SMB share as well. Does it not do cleanup of old backup jobs\plans that get deleted?
Lastly, for KVM VM Snapshot backups to work, I'm supposed to just allow SMB access direct from the VM to the SMB share based on the message I got from this first backup attempt saying it has to go through the appliance? This job took 4 hours to finish, which clearly copied data to the SMB share but still ended in a failed state, nothing in the History to indicate why.
I'm also concerned since I can't require encryption on SMBv3 then the backup jobs streaming to the SMB path are not transfering by a secure method, right?