Backups and Snapshots

[TN68]

TL:DR
When recovering from a hypothetical disaster, can you recover your datapool shares/data AND it's respective snapshots? Or will the last backup only contain the latest visible files... and data that was protected by snapshots may be lost?

- I have a snapshot task running on my entire single pool, and you see folder/file history via windows "previous versions"... so that's great.
- I have an offsite backup of the SMB shared-shares using "IDrive" in a jail, mounted to the said shares/directories.

Hypothetical scenario I'm debating... what if files are deleted/missing (unknowingly) but would NORMALLY still protected by existing in some past snapshot on the system, but alas, some disaster happens (fire, stolen).... You restore your data from an offsite service and THEN you notice some folder/file! Are the snapshots somehow going to be preserved through the backup/restore process for you to go explore and search for the missing data that was THERE but hidden and in the snapshot history? I understand the data doesn't "go anywhere" when you delete it (if it's snapshot'ed) but clearly your OS thinks it's deleted and doesn't show it... so I'm not sure how IDrive deals with this.

I guess I can test-restore some dummy folder and find out but that's hard to test b/c i'm not going to spool up a fresh trueNAS to test the full disaster scenario.

IDrive supposedly has 10-30 snapshots of the stuff you send to it as well... so that's at least some piece of mind if you can't restore snapshot history to TrueNAS.

 

[jenksdrummer]

If you have periodic snapshots set up, and, you set up replication to occur every so often, then the snapshots do come along for the ride and will be on the replicated-to system; there are even options relating to retention of those replicated snapshots.

To the same effect, if you nuke the source system, you can then set up a replication job going the other direction and the snapshots go back with it.

Test, test, test, test and validate :D

Re-reading your post though, I am curious if what you have is a snapshot of a VM; to which, if it's VMWare integrated, I have tested that scnario between 2 units and was able to recover the VM. Hyper-V is a different beast, and success was more like a crazy rollover crash where you happen to land on the wheels. Basically, use the snapshot data to attach to a new VM and then configure as needed because Hyper-V does not like it when you have recover a VM from a snapshot that it didn't manage. VMWare couldn't give 2 ducks a quack.