ext4, RAID, and data recovery

[NeghVar]

System:
MoBo with PCI-X slot
16 1TB HDD in RAID 10 with ext4 partition
1 PCI-X 16-port RAID controller
Scenario:
MoBo fails.
Question:
Would the data be recoverable if I could not get another MoBo with a PCI-X? Instead, I use the SATA ports on a new MoBo, a PCI and a PCIe SATA controller card, and the USB ports on the MoBo to external HDD enclosures. Remember, this is an ext4 partition
I know this would work with a BTRFS partition because I did exactly that. In fact, this is the Frankenstein system I made to do it. (spare parts from around the warehouse)

The reason for this question is that I am looking to purchase a new NAS to replace my old Synology NAS. It currently uses BTRFS. But I have noticed that QNAP does not support BTRFS. In fact, they are actually quite hostile towards it. So, that is why I want to know if I could recover data from an ext4 partition as I did with the BTRFS partition.

 

[Eulogy]

Personally, I'd just build the new NAS with ZFS, then restore from backup. Quicker, easier, and more manageable long term.

 

[mwroobel]

Honestly the answer is maybe. It depends on which RAID card you had (and even which firmware it was running), how the array was created (RealRAID, FakeRAID/MB RAID or SW RAID) and/or expanded/extended over the years, how the array failed, what RAID level you were using etc etc etc. There are simply too many variables to give you a definitive Yes/No.

As far as replacing your older Synology I would suggest a newer Synology (That is if you want an "appliance" you can just unbox and 10 minutes later you are golden) and stick with SHR2/RAID6 on top of BTRFS. Otherwise you can roll your own and go ZFS. QNAP has had a number of issues over the past few years with global, successful RansomWare attacks on their end user devices.

One question now that I re-read it. Do you have a single logical ext4 volume spread across MB SATA, PCI-X SATA and USB drives? That is generally a bad idea, because USB drives (in general) in RAID is a headache waiting to happen. Spanning across SATA and USB in the same array is a disaster (usually) waiting to happen! If it has been working for you long term, you are lucky!

 

[NeghVar]

That system was only temporary. On the Frankenstein system, I installed barebone CentOS with ssh, webmin, and BTRFS. This was completely software RAID. All temporary just to recover the data. Once the volume was accessible, I copied all the data off and built a new system with 4x 4TB HDD in RAID 10 on a PCIe SATA RAID controller. The old system I believe used an Areca ARC-1160
I have a personal love of BTRFS because of what happened above. I found out that part of why I was able to rebuild the array was because BTRFS keeps info on the whole volume on every drive. This is why, despite being on numerous different interfaces and controllers, the volume was successfully rebuilt.
The QNAP NAS I have my eye on has an expansion port for a WIFI 6 card and it is within my budget. Synology has one too, but their cost is way outside my budget and 8 bays is impractical for my needs.
I know some of you will think "Just run a network cable." My whole house runs on WIFI. And the provided router/gateway (I hate Eero) only has a single 1 Gb port. And there is no place near the router to adequately place the NAS

There is no ext4 volume. There never was. I am asking this because I want to know if the same thing that happened above on the 16x 1 TB HDD system could have been done if I had an ext4 volume instead of a BTRFS.

But to be honest, this is mostly just a curiosity. No significant need for the info.

 

[Eulogy]

ZFS still wipes the floor with BTRFS. BTRFS is trying to get close to feature parity, but, it's not there yet and has quite a way to go... the community around it is shrinking a bit, too, unfortunately. ZFS's is larger and older, and more active.
ZFS has data dedupe and compression, and has for a long while. Been a few months since I last looked, but, BTRFS only had dedupe as an experimental feature, and no compression.
ZFS is end to end checksum for data, and any errors are automatically corrected.
ZFS is also much more performant in medium to large data sets. The two systems are about even on small datasets.

The only real advantage BTRFS has at this point is that it's native to the linux kernel... but I'm not even sure I'd really count that as an advantage at this point. If I were building a new NAS today for home especially, it would be a hard sell to get me to consider BTRFS still. Maybe in a couple more years when it's at least feature parity with ZFS I would, though.