NCP Backup Strategy: Are BTRFS Snapshots Sufficient?

[Puilxico]

Hello community
For over a year, I have been running a Nextcloud instance with NCP on a Raspberry Pi 4 (4G). The data is stored on an external drive at /media/myCloudDrive. I am very happy with it. However, so far I have neglected having a proper backup solution and only had dataless backups to the same hard-drive the data is already on. I would like to a have a backup that allows me to recover everything if the primary drive fails.
To prepare the setup, I have read the blogpost by nachoparker, everything I could find in this subforum relating to the topic and on the NCP docs. Based on that, I decided to add a second hard-drive (at /media/myBackupDrive/, formatted with nc-format-USB) and added the directories btrfs_snps and dataless_bkp to it.
First, I changed nc-backup-auto to make dataless backups to /media/myBackupDrive/dataless_bkp. This works fine.
Next, I activated nc-snapshot-auto. As expected, this creates snapshots in /media/myCloudDrive/ncp-snapshots/. I want to sync these snapshots with nc-snapshot-sync using the following config:
Active: yes, Snapshot directory: /media/myCloudDrive/ncp-snapshots, Destination: /media/myBackupDrive/btrfs_snps, Compression: no, Sync periodicity: 1. After 24 hours, the snapshots do not yet show up on the backup drive. But this might just be a matter of time and is not my actual reason for asking here.

If I have this setup fully running. Every change made afterwards will be stored in a snapshot and back-upped to my backup drive. However, as far as I can tell, all the data I already had previously is nowhere in the backup. Am I misunderstanding the issue? If not, what backup strategy would you recommend me to keep my data safe from hardware failure?
As you can see from the config below, there is about 1 TB of data already on my instance. Of course, I can do a “datafull” backup to my backup drive and I am absolutely willing to do that once. But as this would take upwards of 10 hours of downtime each, it does not seem a good strategy for new changes moving forward.

Any suggestions, recommendations or links to explanations I might have missed are much appreciated. I also hope that the answers to this post might be helpful for all others out there who made the mistake to only consider a proper backup after some time.

Kind regards,
Puilxico

NextCloudPi version	v1.34.1
NextCloudPi image	NextCloudPi_07-21-19
distribution	Raspbian GNU/Linux 10 \n \l
automount	yes
USB devices	sda sdb
datadir	/media/myCloudDrive/ncdata
data in SD	no
data filesystem	btrfs
data disk usage	1.1T/4.6T
rootfs usage	4.9G/29G
swapfile	/var/lib/swap
dbdir	/var/lib/mysql
Nextcloud check	ok
Nextcloud version	20.0.4.0

[OliverV]

@Puilxico

Sync periodicity: 1. After 24 hours, the snapshots do not yet show up on the backup drive.

Could you share a screenshot of your settings, And check/share content of the cronjob created for it, i think it is in /etc/cron.d/

Btrfs snapshots are enough for me. It depends how valuable your data is to you. I rsync backups (db+NC) and btrfs snapshots to 2 external server/drives. In ideal situation, one of these is in another location. One should be aware that Nextcloud is a cloud file sharing system, not a backup of your files.

[Puilxico]

Sure. Thanks for answering!

Or did you mean other settings, such as the “Nextcloud configuration” tab?

In /etc/cron.d/ncp-snapshots, there is one line with:
30 4 */1 * * root /usr/local/bin/btrfs-sync -qd “/media/myCloudDrive/ncp-snapshots” “/media/myBackupDrive/btrfs_snps”

So, that looks fine and should sync the snapshots at 04:30 a.m. It didn’t work last night, but I’ll give it another go tonight.

Concerning the backup strategy, I am aware that Nextcloud is not a backup for my personal files and that’s not what my family and I use it for. The PCs/personal files are backuped differently. My intention with this backup project is to easily restore the instance in case of HDD failure, which is likely to happen in the long run. I’d like to be able to do such a restore to a reasonably recent point in time (a few days), but without the hours-long maintenance mode a complete “datafull” backup would cause every few days. And as written above, I believe btrfs snapshots are not sufficient for this as all data uploaded to the instance before setting up the new backup regime would not be covered.

[bubo]

@Puilxico : Did you find answers to your questions? I tried exactly the same. I am wondering if snapshot-sync is sufficient as a data-backup, too. And snapshot-sync isn’t working on my system as well.

[Puilxico]

No, unfortunately I could not find an answer concerning the BTRFS snapshots. I also didn’t manage to get snapshot-sync working, but of course such a sync could be done manually.
I have resigned to do a full backup (with data) on a regular basis. The downtime is about 6 hours in my case, so if I let it run at nighttime, it’s acceptable.

[OliverV]

Maybe a typo or permission issue, worked out of the box for me.
Try again and
Check /var/log/ncp.log