Optimal approach for host backups with Media content

[Mirmanium]

Hi community

I have been using PVE and PBS for some time and 0 issues so far with the backup system so I am very happy.
Currently I backup only LXC and VMs and any restoration I had to do just worked.
Now, I would like to include backup of a shared folder where all my pictures are. The thing is, althought I played with the cli "proxmox-backup-client backup host ..." option and works same way than LXC or VMs, since backup is done at block level, I don't have any way to check how many files have been added, deleted etc... on each incremental copy. I can see the amount of data copied in MegaBytes in the task but this does not give me an indication of what changed from previous backup.

I am looking for a way to have confidence on the backups and identify if from one backup to another, for instance, thousand of files were deleted it may be an indicator that something happened by mistake.

I was playing with some very rudimentary process to mount using Fuse the previous and new backup in 2 folders and do a "diff -r folder1 folder2 | diffstat" to get the differences... but is far from ideal. Have not seen any other option in documentation as an alternative.

I totally understand the benefit of deduplication and for VMs I don't need this file comparison but for photos or media I really would like.

Someone in the same situation? I want to avoid to have another backup tool like duplicaty for files and PBS for rest if I can use PBS for both (gives me much more confidence among other benefits).

Thank you,

 

[shanreich]

It is possible to get a list of all files contained in a backup archive via:

proxmox-backup-client catalog dump <snapshot_name>

You could then try comparing (e.g. via diff) the output of the two different backup archives.

 

[Mirmanium]

Mmm... interesting. I don't know why I missed the dump option... sorry.
I tested and notice that list of files are shown as this example:

./xxx.pxar.didx/var/www/html/file.txt" 3354 2022-09-15T03:00:41Z

If timestamp is diffent between 2 snapshots dump, I guess is because at block level file changed and PBS included the new one? is this the timestamp from PBS or from the source file modification attribute? Because I have no way to check the "hash" or something similar to ensure file has not been modified even though it can have same size.

Thank you,

 

[shanreich]

Yes, it should take the modification time from the file system.

 

[Mirmanium]

Great!
A combination of dump information + deduplication amount of data may provide what I need.
Thanks!