Hyper-V: What kind of checkpoint with ubuntu server vm

[TheFu]

Snapshots are not backups. They are frozen points in time, but on the same storage, so a disk failure takes the snapshot with it.
Snapshots quiesce specific blocks and provide a name for those blocks so they can be mounted (read-only) and a perfect backup can be made while the system keeps running full speed.

Since we are all pros here, we need to ensure that lurkers in 2-5 yrs don't get confused. Even a "backup" on other storage really isn't a backup until it is verified as consistent and tested to ensure it can actually be restored. Nobody does these things, but we all should validate (or have a backup validation tool) check that the non-live storage isn't corrupted AND we should test a full restore at least once a year to ensure we didn't forget the egg with the chicken.

The first 5 attempts to restore a backup will likely fail, so whenever setting up a new backup process, be prepared to test multiple times to ensure critical information is available and understood for the guy paged at 2am to do the restore while half asleep. I've inherited some terrible restore processes and I've inherited some fantastic processes. These days, I can restore any system here, usually in under 30 minutes, but no longer than 45 minutes. The restored system can run on completely different CPUs, different storage architectures, with nearly completely different hardware. If the building is a crater and the offsite backups are available, we can restore to a VPS the most business critical stuff, if needed, while we search for a replacement building to lease and order hardware. Our backups feed into a DR plan. Fortunately, it wouldn't be the end of the world if the systems were down a day or two. I've worked places were 1 hour of unavailable systems would kill people. Those had DR plans with live replication to alternative DCs and we'd test the failover 2x a month. The client wasn't interested in all the testing until there was a corrupted DB at 10am one day. The client still didn't make it a priority until their bosses-bosses MADE it a priority for them.

[MAFoElffen]

LMAO... PTSD moments for me also.

LOL! Bravo! Well said.

The problem is too late when it occurred 8 months previous and no-one noticed... (Experience from being brought into someone else's raging bonfires)

[kathrin-huber]

okay.... it was mentioned several times that a snapshot is not a backup now - I have a backup strategy but that was not the question - the question was, what kind of snapshot I should use so I do not need a backup when I use it

[CharlesA]

okay.... it was mentioned several times that a snapshot is not a backup now - I have a backup strategy but that was not the question - the question was, what kind of snapshot I should use so I do not need a backup when I use it

Standard checkpoint would be ideal in this case.

[TheFu]

okay.... it was mentioned several times that a snapshot is not a backup now - I have a backup strategy but that was not the question - the question was, what kind of snapshot I should use so I do not need a backup when I use it

Nevermind. The hostOS is Windows.

[MAFoElffen]

Terminology: Snapshot (Other than Windows = Checkpoint (In Windows)... Microsoft just can't use the name "Snapshot" because they lost a suit from Apple over that. Just a trivial pursuit question.

If you are taking a Checkpoint from Hyper-V of a VM Guest containing anyhting other than a Microsoft Host, use "Standard Checkpoint".

As for Checkpoint as related to backups... Please refer to Microsofts own flagship System Center Configration Manager (SCCM) Virtual Machine Manager (VMM) Documentation. From the SCCM-VMM 2019 Doc's:

Back up and restore VMM
...

Before you start

 - Don't use checkpoints for disaster recovery. Checkpoints do not create full duplicates of the hard disk contents, nor do they copy data to a separate volume.
 - You can use a checkpoint to serve as temporary backup, before updating an operating system on a virtual machine. This allows you to roll back the update if it has adverse effects.
 - You should use a backup application to back up and recover your data in case of catastrophic data loss. One option is System Center Data Protection Manager (DPM).
 - Data such as Remote Access Authorization (RAA) passwords and the product key can be entered when you reinstall VMM. However, some encrypted data such as Virtual Machine Roles cannot be reentered.
 - You can't back up and restore such data if you use the Data Protection application programming interface (DPAPI) for backing up VMM.
 - The data will be lost if the VMM management server fails.


Create and implement a backup plan

Basic elements of a backup plan include a list of what needs to be backed up, and an outline of what is changed frequently (and therefore need to be backed up frequently) in your environment.

No one besides MS has VSS or DPM... Those are specific to Microsoft and their own branding's. So for that Checkpoint to work to get back to a specific point in time (on anything other than an MS OS), you need the "State", which for your question, means a "Standard Checkpoint"... And just because you take a Checkpoint, doesn't mean it's safe.

There where tips on how that is best done via IT "best practices", but that also falls under, and is affected by what your company policies and business priorities apply. Cloud, Server Farms, Server Clusters and replication servers, although safer, do get backed up... and if "always up" is the priority, have multiple failover/fallback locations, so that can be done. That is not to change the answer you are looking for, but just additional information to it. Those apply to your own company's Disaster Recovery Plan's Risk Impact acceptance.

I have no idea what your infrastructure is. I'm just someone who has had to support Windows, Linux, UNIX, VMware, etc... Living in Datacenters, in the same infrastructure. This is at least my understanding at this specific point in time.

So yes, the simple answer to your question does seem to be: Check the box for Standard Checkpoint...