Filesystem for reliability (HDD)

I have ditched cloud storage completely as it was costing way too much lately, everything has seemingly gotten more expensive nowadays…

I now have in my PC an nvme SSD and two 2TB enterprise hard drives, all formatted XFS

I am planning to get two more 6-8 Tb enterprise hard drives for archival purposes and a general mirror of everything. Not in a RAID, but me manually mirroring the more important files, to maximize storage.

Do I stick to XFS, or is is wiser to use Btrfs? If so, how would you suggest I format that future volume?

LE: Also, in theory, how can I protect a file system against file deletions, or to not delete more than one file at a time etc? Is there a better way than just mounting it read only?

Hopefully, you can get some folks to chime in about filesystems; I’m an ext4 guy…

Maybe this SE article can give you some inspiration.

Don’t mount it at all until you need it; then unmount it immediately afterwards.

1 Like

If you want reliability and dont need a bunch of fancy things … then I would be using ext4
(btrfs for example is known to not always be reliable)


In our dataplexes, or massive servers, we use Ext4 as it is hassle free and proven to be reliable. Indeed, those servers are used for reliable physical simulations data storage!

Hope that helps, have a wonderful day!

1 Like

Am I the only one who doesn’t understand what you mean? :stuck_out_tongue_winking_eye:
Please, describe a bit more what you want to achieve!

i understood guys, so btrfs falls out of the question

Now between ext4 and xfs… Is xfs still in favour? I have heard that it had some journaling features removed but it’s sequential performance is still the highest.
In journal, indexing etc reliability, are ext4 and xfs different of they do the same thing?

I don’t want to use raid anymore since it’s too complicated for me, even in a simple raid 1 mirroring, who knows what I need to change then array issues and other stuff. Had a raid 5 once and never again, hehe

Basically I’ll have most important files + about 90% of others copied from one hdd to the other, as in case i have data loss i will lose 10% unimportant stuff

And yeah one of the future 6+TB drives i will have always disconnected, stored in a drawer @merlock , thanks for the suggestions btw. With the other drives i plan to routinely unmount them with a cron script and have hdparm tunr them off, so that they don’t accumulate power on hours for nothing.

I will have at most only one of the hdd’s active 24/7 for filesharing. The PC itself is on 24/7.

1 Like

You could use systemd mount units in combination with automount units where you can set the delay :wink:
Not sure how you could add the hdparm off switching in this combo yet though as i never used that.

1 Like

btrfs is actually great for sending read-only snapshots from one disk to another. You just need to automate that. I haven’t researched which tool can automate send-receive feature of btrfs properly. This one looks promising:

But there are plenty of other solutions which are not based on btrfs:

And/or read what Fabby recommends:

But I still don’t know what you want to achieve.

1 Like

My sole professional advise is Ext4 !

1 Like

Some people recommend ZFS. But not for the system partition, just for data.

1 Like

I have been using BTRFS for my system SSD for two years and I have never experienced any problems. When you read about BTRFS mind the date of articles because this file system was actively developed and some of them may be out of date. BTRFS is great file-system for operating system purpose because of snapshots that take a second or two, literally. And for system backups with BTRFS you need to have a separate backup storage with BTRFS, one partition with BTRFS, other partition might be different. I have it this way, HDD with about 30GB for system incremental BTRFS backups, but only half of it is taken so it actually could be smaller, and the rest ext4 for normal storage.

So, for backup or normal storage outside SSD, I would use ext4 or if you need to use Windows NTFS, so that Windows can read from them without problem. And you can use some synchronization tools, even with GUI for example luckyBackup. It depends how advanced Linux user you are, and if the fancy solutions you want to apply are appropriate for the purpose.

1 Like

When you want to use btrfs send-receive feature you need two disks both with btrfs.
Having btrfs and not to use send-receive is a waste of opportunities.

And I was using btrfs and had issues every now and then. Usually extent tree corruption and btrfs repair --init-extent-tree can take days and still end without success. With backups not a big issue, but sometimes the backup disk is needed for something else and just then the corruption will happen. But if you have a proper backup filesystem corruption is not an issue.

1 Like

Thank you all! This will help me take a decision for my completely amateurish setup :slight_smile:

This topic was automatically closed 15 days after the last reply. New replies are no longer allowed.