Hey all!
I hope you’re well and good 
So I have a test install on bare metal, I’m sure it consisted of 6 drives (RAIDZ1), but after multiple old drive failures, it eventually resilvered itself onto 5 drives.
Is that a thing that can happen or have I wondered over into a strange world?
Thanks!
Chris
I have never heard of that. If you added a hot spare, would it give you back your 6 drive vdev?
Reducing the stripe width of RAIDZ is technically impossible. RAIDZ Expansion feature (soon lol) allows for adding drives, but not remove them.
If it was a 6-wide RAIDZ and now is a 5-wide RAIDZ, you can’t read any data from that pool and it should report as faulted or wouldn’t import at all.
If it is a RAIDZ1, you may have mistaken the numbers with 5 drives worth of data aka 5+1 notation and subconsciously created a 5-wide RAIDZ1 with 4 disks worth of data + 1 worth of parity.
Would be my humble guess.
Thanks for this. I wonder if I did have a hot spare, I should document things better 
I’m relieved to hear this, that it’s technically impossible. I must be mis-remembering.
Here’s the situation (yes, it’s awful…diagnostic reasons you understand). One of them is new and nothing to do with the pool in question, so that gets us down to 7. One of them is new and recently added, down to 6. I thought that I replaced at least 2 of them, but somehow it dropped the need for a drive. Very strange.
OK, I’ve had a memory refresh, the TL:DR is that I always had 5 drives
Soon I need to replace a mirrored pair with a 3 drive RAIDZ1, due to space requirements. Before I do this on the live machine, I thought I should get well practiced with it on a dummy machine.
I planned to replace a pair (2 drives), with a 3 drive RAIDZ1…so 2 + 3 is, yes, you guessed it. 5.
Apologies for wasting your time @Exard3k @MikeGrok , I promise I won’t do it again!
No harm done. You wouldn’t be the first one making a mistake. My pool is more easy in terms of RAID config (3 mirrors is as simple as it gets), but I’ve created many pools in the last years and created wrong (not desired) vdevs at least twice, but I was always able to correct it without losing anything.
My last VM with 48 virtual disks running with dRaid certainly had a couple of mistakes on my side, both in disk name, amount and dRaid notation.
Always good to learn and remember to pay attention to things next time.
Thank you, I’m glad it’s not just me to forget or do vdev layouts wrongly, glad you didn’t lose anything though.
Wow, 48 virtual disks, that’s busy!
Yes indeed, always good to learn/recall these things, I have to say this is where the disadvantage of not being in IT full time, doesn’t help. If I did it for other people every few days, the procedure would be ingrained in my brain.
I work in IT every day, (independent contractor) I make stupid mistakes especially when I am in a hurry to finsh a project.
