jxi7eoiigzns7oaf7

So far I've added a single device L2ARC and a two device mirror as a SLOG using this method and it's gone well!

Awesome thank you very much for testing that out! I think I will try by just adding some L2ARC and hopefully nothing blows up if I make a mistake. That seems the easiest to undo if something goes awry. Do you know, if you've already started the array once and put in the disk encryption password in the process, and then stop the array, can you just do cryptsetup luksOpen /dev/sdX1 sdX1 without specifying a keyfile or passphrase? Would it just use the one it has cached from you entering it to start the array? Presumably it's still cached since you can start the array again without re-entering the password. Is there some sort of 'crypto context' or something that would make this possible (and hopefully ensure the exact same settings are applied as when you use the GUI)? I really just don't want to mess anything up by encrypting things incorrectly. I am not sure if Unraid uses a salt, for example, that wouldn't be automatically included even if I used the same password when doing the cryptsetup manually. Thanks again for your help

Yeah I thought about either trying to encrypt them manually or with UD and then trying to pass the resulting LUKS volumes (partitions?) to 'zpool add' but I am not sure how Unraid handles opening encrypted volumes that you encrypt manually. I am going to see if there's a safe way I can test it with spare disks later when I get home. From what I can I see looking at the existing disks in my zpools and from watching the logs when I clear and format a drive with UD (both XFS - Encrypted and ZFS - Encrypted), all that's done is: 1) Wipe the partition table, 2) Create a new MBR partition table (in the case of SSDs), 3) cryptsetup luksFormat /dev/sdX 4) cryptsetup luksOpen /dev/sdX which then gives you the expected /dev/mapper/sdX entry. However, I haven't yet found where in the config Unraid stores the list of devices it tries to mount/encrypted volumes it tries to open when you start the array. I am hoping if I can ensure these devices are added there that the pool import would be automatic after all the LUKS volumes are opened.

I figured out the problem eventually. It was a bad stick of RAM. I replaced it and have not had this issue since.

I have a zpool of mirror vdevs which I created using the GUI and selecting 'zfs-encrypted' for filesystem type. I would like to add some nvme ssds as l2arc, slog, special, or dedup vdevs. I know this is not currently supported via the GUI, so I will have to use the CLI for this. I found JorgeB's excellent post here: explaining how to add such vdevs to an existing zpool. However, since my devices are also LUKS encrypted I am not sure how best to go about this. I looked at zpool history to see how the original pool was created, and found this: zpool create -f -o ashift=12 -o autotrim=on -O compression=on -O dnodesize=auto -O acltype=posixacl -O xattr=sa -O normalization=formD -m /mnt/tank tank mirror /dev/mapper/sdn1 /dev/mapper/sdo1 mirror /dev/mapper/sdp1 /dev/mapper/sdq1 mirror /dev/mapper/sdr1 /dev/mapper/sds1 So from that I gather 1) the devices are passed to zfs after the LUKS encryption is applied (makes sense), and 2) only one specific partition on each drive is passed to zfs, rather than the entire block device (also makes sense). In my case each drive only has one partition, but as a note I have also seen other posts mentioning that unraid only supports creating zpools this way using a single partition per drive at the moment, and that it HAS to be the first partition. My question is: how do I accomplish this (i.e., a) encrypt the drives with LUKS, b) add them as vdevs to my zpool, and c) get unraid to import the pool automatically) without destroying my zpool or having any weird graphical issues like other people seem to be encountering while trying to modify their zpools via the CLI (e.g., drives not showing up after starting the array, missing drives from the GUI, being unable to format just the drives you're trying to add upon start of the array rather than being prompted to format every drive in your pool, etc.)?

I am having this same issue. I tried updating to the RC version Jorge suggested but unfortunately I still get a bunch of spam of this whenever I try to stop the array: Jul 15 10:28:28 vault emhttpd: Sync filesystems... Jul 15 10:28:28 vault emhttpd: shcmd (1362): sync Jul 15 10:28:28 vault emhttpd: shcmd (1363): /usr/sbin/zfs unmount -a Jul 15 10:28:28 vault emhttpd: shcmd (1364): umount /mnt/user0 Jul 15 10:28:28 vault emhttpd: shcmd (1365): rmdir /mnt/user0 Jul 15 10:28:28 vault emhttpd: shcmd (1366): umount /mnt/user Jul 15 10:28:28 vault root: umount: /mnt/user: target is busy. Jul 15 10:28:28 vault emhttpd: shcmd (1366): exit status: 32 Jul 15 10:28:28 vault emhttpd: shcmd (1367): rmdir /mnt/user Jul 15 10:28:28 vault root: rmdir: failed to remove '/mnt/user': Device or resource busy Jul 15 10:28:28 vault emhttpd: shcmd (1367): exit status: 1 Jul 15 10:28:28 vault emhttpd: shcmd (1369): /usr/local/sbin/update_cron Jul 15 10:28:28 vault emhttpd: Retry unmounting user share(s)... Jul 15 10:28:33 vault emhttpd: shcmd (1370): /usr/sbin/zfs unmount -a Jul 15 10:28:33 vault emhttpd: shcmd (1371): umount /mnt/user Jul 15 10:28:33 vault root: umount: /mnt/user: target is busy. Jul 15 10:28:33 vault emhttpd: shcmd (1371): exit status: 32 Jul 15 10:28:33 vault emhttpd: shcmd (1372): rmdir /mnt/user Jul 15 10:28:33 vault root: rmdir: failed to remove '/mnt/user': Device or resource busy Jul 15 10:28:33 vault emhttpd: shcmd (1372): exit status: 1 Jul 15 10:28:33 vault emhttpd: shcmd (1374): /usr/local/sbin/update_cron Jul 15 10:28:33 vault emhttpd: Retry unmounting user share(s)... Jul 15 10:28:38 vault emhttpd: shcmd (1375): /usr/sbin/zfs unmount -a Jul 15 10:28:38 vault emhttpd: shcmd (1376): umount /mnt/user Jul 15 10:28:38 vault root: umount: /mnt/user: target is busy. Jul 15 10:28:38 vault emhttpd: shcmd (1376): exit status: 32 Jul 15 10:28:38 vault emhttpd: shcmd (1377): rmdir /mnt/user and it just hangs forever, forcing me to have to reboot.