| 13 Apr 2022 |
 @elvishjerricco:matrix.org | what does it do if the passphrases differ between disks? | 17:27:55 |
| @elvishjerricco:matrix.org | Or like, if two of them share one password and two others share a different one? | 17:28:09 |
 @janne.hess:helsinki-systems.de | then it fails and asks | 17:28:27 |
| @elvishjerricco:matrix.org | Will it try to share for the ones that can share? | 17:28:54 |
| @janne.hess:helsinki-systems.de | huh? | 17:29:49 |
| @elvishjerricco:matrix.org | Like in the 2x2 I just described | 17:30:12 |
| @janne.hess:helsinki-systems.de | ahhh I see | 17:30:29 |
| @janne.hess:helsinki-systems.de | I don't know how the kernel keyring is implemented, but if it's a key-value store then no, 2x2 doesn't work | 17:30:55 |
| @elvishjerricco:matrix.org | Actually | 17:31:10 |
| @elvishjerricco:matrix.org | I bet it would ask for the first one, and before you can even enter the second one, it would have tried the first on all the disks, so the ones it would work for would be unlocked | 17:31:32 |
| @elvishjerricco:matrix.org | I'm gonna try it... | 17:31:52 |
| @janne.hess:helsinki-systems.de | I can push my WIP stuff soon | 17:32:03 |
| @elvishjerricco:matrix.org | Ok I see. The first one you enter is the only one it remembers. So if drives 1 and 2 use password A, but drives 3 and 4 use password B, then I'll enter A once, it'll get remembered for all four drives, and 3 and 4 will both need me to manually enter B | 17:45:16 |
| @elvishjerricco:matrix.org | Still that's really cool | 17:45:47 |
| @elvishjerricco:matrix.org | And they're unlocked in parallel, which should give me a small speed boost for my 6 disk raidz2 zfs pool :) | 17:46:51 |
| @janne.hess:helsinki-systems.de | whoop whoop | 17:47:02 |
| @elvishjerricco:matrix.org | Actually, I tried it with all four drives having the same password and it was weirdly serial... | 17:49:19 |
| @janne.hess:helsinki-systems.de | you can probably disable the prompt | 17:49:37 |
| @janne.hess:helsinki-systems.de | it can only prompt for one password at a time | 17:49:44 |
| @janne.hess:helsinki-systems.de | --no-tty? | 17:50:29 |
| @elvishjerricco:matrix.org | Well yea but I answered the first prompt and then, as expected, did not get a second prompt. But it would only say "Found device ..." after it finished mounting the previous one | 17:50:45 |
| @janne.hess:helsinki-systems.de | that's sad | 17:51:04 |
| @elvishjerricco:matrix.org | So find, open LUKS, mount, repeat | 17:51:05 |
| @elvishjerricco:matrix.org | serially | 17:51:06 |
| @elvishjerricco:matrix.org | no idea why | 17:51:12 |
| @elvishjerricco:matrix.org | That's very said. I wonder if the kernel can only unlock one at a time? That doesn't seem likely | 17:51:37 |
| @elvishjerricco:matrix.org | * That's very sad. I wonder if the kernel can only unlock one at a time? That doesn't seem likely | 17:52:22 |
| @janne.hess:helsinki-systems.de | yeah, that seems unlikely | 17:58:17 |
| @elvishjerricco:matrix.org | Hm. After the first drive finishes unlocking, systemd debug messages give me cryptsetup.target: starting held back, waiting for: systemd-cryptsetup@luks2.service | 18:09:04 |
| @elvishjerricco:matrix.org | and then that drive's stuff gets to work | 18:09:18 |
