Comment by thedougd | Hacker News Reader

Comment by thedougd | Hacker News Reader

Parent Full thread

Exactly the route I took. I had an aging tower machine full of spinning disks running on an old LSI adapter that was doing hardware raid. They were out of space and I began to get nervous the LSI adapter could die and I would trouble replacing it. Decided JBOD for the future.

External drives were on sale, I bought several and setup with a RPI. Lots of headaches. It took effort to iron out all the USB and external disk issues. Had to work out alternative boot. Had power adapters fail for the RPI. Had to enhance cooling. etc. Kept running into popular Docker containers still not having aarch64 variants.

I finally replaced the RPI with a used Dell SFF. Kept the USB drives and it's been solid with similar power draw and just easier to deal with all around.

Though I am considering going back to a tower, shucking the drives (they're out of warranty) and going back to SATA.

I think most LSI adapters you can get a battery backup for. I've got one on mine, plus a spare battery sitting on a shelf somewhere. I admit when I put the system together for the first time I was a little hesitant to go with hardware RAID but it's worked out fine so far.

I reckon the issue is more in replacement than transient data loss: what are you going to do when you can't find a replacement controller card, or it only available at ludicrous prices?

With a proprietary on-disk format you can't exactly hook them up to any random controller and expect it to work: either you find a new one from the same controller family, or your data is gone.

Replacing your RAID controller is already major maintenance, so there's going to be downtime. I wouldn't be opposed to just wiping the drives and restoring from the latest backup. I routinely do this anyway, just to have assurance that my backups are working.

And a risk! I've had this on a premium machine put together specifically for that purpose and when the raid controller died something got upset to the point that even with a new raid controller we could not recover the array. No big deal, it was one of several backups, but still, I did not expect that to happen.

Reply on news.ycombinator.com