dnyberg Posted May 28, 2022 Share Posted May 28, 2022 Ok, so this influences procedure: Suppose we have an array with a bunch of drives, say 6TB each. Parity drive(s) exist, one data drive fails. Suppose also that no replacement drive is currently available, but the surviving data drives have sufficient empty space, say 10 TB free. Do we absolutely have to have a replacement drive to swap in for the failed one before we regenerate, or can unraid regenerate data and put the recovered data into that existing free 10 TB? The answer to this tells us if we have to have a spare drive on the shelf at all times, or if all available drives can be installed from the beginning. Thanks! Quote Link to comment
Squid Posted May 28, 2022 Share Posted May 28, 2022 You would copy all the data from the emulated drive to one (or more) of the other drives (Use Dynamix File Manager). Then you would do a New Config, reassign all the drives excluding the one you're removing and rebuild the parity information. Quote Link to comment
dnyberg Posted May 28, 2022 Author Share Posted May 28, 2022 Hm, ok, so "emulated drive" is scratch storage used by the rebuild process, and then either handled as described above, or in the more usual situation (we do have a fresh drive to pop in) copied to the replacement drive? What's the physical basis of the emulated drive; is that space borrowed from the surviving healthy drives, or a physically separate drive set up just for that purpose, or...? Quote Link to comment
Squid Posted May 28, 2022 Share Posted May 28, 2022 The emulated drive is the result of the mathematical equation resulting from reading the contents of all the other drives and calculating what that drive should be. https://wiki.unraid.net/Parity Quote Link to comment
JonathanM Posted May 28, 2022 Share Posted May 28, 2022 1 hour ago, dnyberg said: The answer to this tells us if we have to have a spare drive on the shelf at all times, or if all available drives can be installed from the beginning. Every bit of every drive in the parity array, filled with data, empty, doesn't matter, is part of the parity equation. The upshot of this, is that if an empty drive fails, rebuilding a full drive that fails before the empty drive is rebuilt is impossible, because the bits on that empty drive are still part of the rebuilding process. So, best practice to reduce points of failure is to only add drives that you need to actually hold data, plus a margin of empty space on each drive. Personally only I keep enough array space empty as my largest data drive, for example, if I was using 6TB drives I wouldn't add array drives until I had less than 6TB free across all the drives. That way if you needed to, you can empty your largest data drive by moving the data to the others, at least temporarily for organizational reasons, plus file systems need some free space for checks and stuff. So, yes, all available drives CAN be installed right away, but I would strongly recommend NOT doing that. All drives will fail. Predicting when is more art and mentalism than science. Always be sure ALL your array drives are healthy, replace promptly when needed, keep good backups disconnected from the array to allow recovery from accidental deletion or corruption. 1 Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.