This is just an anecdote, I gotta need to tell. 😑 No blame or suggestion whatsoever. 

 

I started using Unraid in 09/2023. I frankensteined together an old, but adequately performant and power friendly little NAS with an i7-4770T on a Q87 board with 32G DDR3. I didn't have HDDs at hand, so I bought two handful of ST4000 drives. 🤡 I now know I shouldn't have done this, but back then I didn't fully understand how conveniently Unraid is supposed to work. I'm currently running a 2-parity, 4-drives btrfs array with 3 spares.

 

Anyway, I also had a 4x2.5" SATA IcyDock drive bay, which nicely fitted into that Revoltec Sixty-1 case that had several 5.25" bays. 💽

I had it set up with a "persistent" cache pool with 2x512G Samsung SSD 840 Pro (MZ-7PD512) in RAID1 and a "temporary" cache pool with 2x500G Sandisk Ultra II (SDSSDHII-500G) in RAID0.

 

Both Samsungs were pretty much worn out already and one quickly threw some SMART errors; thus the've been replaced with a couple 1TB Samsung 870 QVO (MZ-77Q1T) just 2 months later. The Sandisks were performing unexpectedly well, though, until now. 🤞

 

This week I upgraded to Unraid 7️⃣ after a nice, but not particularly remarkable uptime of over 4 months. Everything went well, but the "temporary" cache pool was gone after rebooting.

 

The system only found one Sandisk. A reboot didn't help. The first one was unrecognizable by the system (it didn't respond on the bus, but past me didn't know that yet). So, at first I was not too concerned, since there's a known issue with RAID0 after upgrading with a guide forward; failing those I got a bit suspicious. 🤨

 

Seeing those COMRESET errors in dmesg, I went down into the basement and wanted to fetch the failing drive from the bay to look at it on another machine. Unfortunately I had labelled them wrongly in the disk location plugin 🤦, so I threw out the one that was still recognized by the system.

Boom 💥, that "killed" the second one, now both are unrecognizable.

 

I thought, maybe the chipset is going on a permanent hiatus🎊, but even one of the less worn out Samsung 840 showed up immediately after plugging in.

 

So, long story short, or TL;DR: This got me started thinking 🤔 about the "persistent" cache, though RAID1, but only 2 drives that could as easily fail at once, too... 

 

PS: If you made it that far down here, you've earned a hug or more: 🤗 🍻

Coincidentally I just lost two of four in a RAID10 pool - both wd red 2tb, the other two are crucial mx500 2tb. Unfortunately it seems they were the same part of the mirror (or I did something wrong) as the pool was toast. Completely dead drives, no system recognizes them. Power loss and UPS triggered graceful shutdown, and they never returned - gone on boot.

Backups restored fine though.