RAID0 drives failing... but which one?

[dhenzler]

I have built several FreeNAS systems. All work well. I downloaded TrueNAS and wanted to PLAY with it. So used the only server that remained. Unfortunately it has a RAID controller. So... I set up the drives as RAID0 and proceeded.
Loved the way TrueNAS worked... However after 30 hours of operation or so it started acting squirly. I didn't NEED it so, dumped the jail, and plugin, and attempted to dump the pool. Failing this just rebooted the machine and cleared the 3 drives I was using for TrueNAS.
All seemed good until I had completed re-scanning the content. Took hours to do... 925 folders. But when I took a look at the Dashboard noticed an alarm saying drives were failing.
A look at the pool validated that, however instead of identifying which drive was bad like it would do with an HBA controller, it simply said ONE or more of my drives was failing.
No indication of which. I used smart to check them quickly and two of them had 44K hours and one had <400 hours. Obviously one of the two high hour drives is suspect. But which one?

Anyone out there have knowledge how to proceed. I guess I could boot Linux and use SMART from there, but would the RAID0 prevent an accurate picture there as well?

Help !

 

[Samuel Tai]

Sorry, we can't help you. Using a hardware RAID controller is not supported. RAID0 (striping) has no redundancy, and is also not recommended.

 

[Yorick]

Assuming you mean each drive set up as an individual raid0 of one drive. What you found out in your playing is that ZFS needs raw access to disks, and if it doesn’t have that, figuring out what is failing when it’s failing is impossible. Recovery, even more so.

Good lesson to learn first-hand, I suppose. That server of yours, does it have a free PCIe slot? If so, an LSI HBA in IT mode is around usd 35 on eBay. Get one of those, connect your drives there, and start over from scratch. The pool as is obviously won’t come over from that raid0 Setup. If there’s data on there you care about, get it off first.

If that raid card is LSI, some can be flashed to IT mode to become HBAs. Depends very much on your exact hardware.

 

[dhenzler]

Sorry, we can't help you. Using a hardware RAID controller is not supported. RAID0 (striping) has no redundancy, and is also not recommended.

I know... that's why I posted this "OFF TOPIC" figured someone bright enough to use FreeNAS might have an answer

 

[dhenzler]

Assuming you mean each drive set up as an individual raid0 of one drive. What you found out in your playing is that ZFS needs raw access to disks, and if it doesn’t have that, figuring out what is failing when it’s failing is impossible. Recovery, even more so.

Good lesson to learn first-hand, I suppose. That server of yours, does it have a free PCIe slot? If so, an LSI HBA in IT mode is around usd 35 on eBay. Get one of those, connect your drives there, and start over from scratch. The pool as is obviously won’t come over from that raid0 Setup. If there’s data on there you care about, get it off first.

If that raid card is LSI, some can be flashed to IT mode to become HBAs. Depends very much on your exact hardware.

 

[dhenzler]

Thanks... I have a few laying around... thought about doing exactly that. Just as a diagnostic. Luckily I don't need it. Was simply an experiment with a server that only had a RAID controller.

Good part is... I like TrueNAS.... think it's good... perhaps Great !

 

[dhenzler]

Installed an LSI SAS9211-8i and of course this erased everything on the drives. I decided to add one more drive, and use RAIDZ2 and so am not NOT seeing any sign of drive failure. Perhaps some anomaly caused by RAID0 drives. The LSI card works fine. Won't be loading it fully as moving 2.2TB of storage takes a while. If anyone knows a good way to move these fast... let me know... I'm using Samba

Dave