| 12 Jul 2024 |
 nrv | * The config is... complicated in addition to all the trouble, i'm trying to do 1M BIOS offset + 600G luks zfs and then -1G reserve at the end and -512M off that ESP -4096 off that ext4 boot. So, ESP and another ext4 boot partition counting from the end, leaving a lot of space to possibly expand the ZFS later (1T drive).
For the time being i commented out the end part and try to create only the BIOS + main partition part.
No part of the disk is in use. I blkdiscarded and sgdisk --zapped it and rebooted and partprobed and whatever multiple times.
It is also healthy disk that i can manually partition and use | 09:55:37 |
| nrv | * The config is... complicated in addition to all the trouble, i'm trying to do 1M BIOS offset + 600G luks zfs and then -1G reserve at the end and -512M off that ESP -4096 off that ext4 boot. So, ESP and another ext4 boot partition counting from the end, leaving a lot of space to possibly expand the ZFS later (1T drive).
For the time being i commented out the end part and try to create only the BIOS + main partition part.
No part of the disk is in use. I blkdiscarded and sgdisk --zapped it and rebooted and partprobed and whatever multiple times.
It is also healthy disk that i can manually partition with luks/zfs/ext4 and use | 09:56:48 |
| nrv | * The config is... complicated in addition to all the trouble, i'm trying to do 1M BIOS offset + 600G luks zfs and then -1G reserve at the end and -512M off that ESP -4096 off that ext4 boot. So, ESP and another ext4 boot partition counting from the end, leaving a lot of space to possibly expand the ZFS later (1T drive) OR use it for something else.
For the time being i commented out the end part and try to create only the BIOS + main partition part.
No part of the disk is in use. I blkdiscarded and sgdisk --zapped it and rebooted and partprobed and whatever multiple times.
It is also healthy disk that i can manually partition with luks/zfs/ext4 and use | 09:58:58 |
| nrv | So, sgdisk commands succeed, but then there is no cryptsetup in the log. It is in the script though, but refers to hardcoded random partlabels(?), dunno how that works, are they static and calculated at build time? | 10:01:27 |
| nrv | * So, sgdisk commands succeed, but then there is no cryptsetup in the log. It is in the script though, but refers to hardcoded random partlabels(?), dunno how that works, are they static and calculated at build time? They seem to exist | 10:02:30 |
| nrv | /tmp/secret.key is also created | 10:03:08 |
| nrv | i did actually forget that before and the error was different and intelligible of the issue | 10:03:33 |
| nrv | ..so it's failed to open keyfile | 10:16:55 |
| nrv | but cat /tmp/secret.key works | 10:17:08 |
| nrv | (and of course it's not in the logs, disabled explicitly for the cat but not for the cryptsetup so hmm) | 10:19:32 |
| nrv | * (and of course it's not in the logs, disabled with +x explicitly for the cat but not for the cryptsetup so hmm) | 10:19:44 |
| nrv | also, disko completely misses this error and proceeds with the zfs stuff | 10:20:44 |
| nrv | * ..so it's failed to open keyfile (the pipe substitution or whatever it's called in bash) | 10:22:28 |
| nrv | * ..so it's failed to open keyfile (the pipe substitution of cat or whatever it's called in bash) | 10:22:40 |
| nrv | * ..so it's failed to open keyfile (the process substitution of cat) | 10:23:34 |
| nrv | * ..so it's failed to open keyfile (the process substitution of cat /tmp/secret.key) | 10:23:44 |
| nrv | * (and of course it's not in the logs, disabled with +x explicitly for the cat but not for the cryptsetup so hmm, maybe the big ( ) subshell captures it?) | 10:24:56 |
 lassulus | hmm, weird, I don't see any code that ignores the cryptsetup error. like it should fail if cryptsetup fails. The partlabels are calculated inside the gpt type and are used onwards from there | 10:26:11 |
| nrv | * At first i think i miscalculated the sizes. sgdisk would say cannot make partition and refuse going further. That is ok.
But then this, and also partprobe broke and said to reboot, and also sgdisk were spitting random numbers instead of ones that i give to it | 10:26:13 |
| lassulus | but you can also overwrite the labels there, if they are too long, they are changed to a hash | 10:33:30 |
| nrv | could it be that sgdisk --clear doesn't work properly and there's still partitions on the disk when i run disko multiple times? | 10:38:05 |
| nrv | even though it runs --clear and partprobe and everything | 10:38:17 |
| nrv | because now i see sgdisk failing again (though i tweaked the sizes) | 10:38:56 |
| nrv | In reply to @nevrv:matrix.org
could it be that sgdisk --clear doesn't work properly and there's still partitions on the disk when i run disko multiple times? sure it could. blkdiscard fixes that. | 10:42:08 |
| nrv | bug? | 10:42:31 |
| nrv | * bug on the disko part? | 10:42:43 |
| nrv | In reply to @nevrv:matrix.org
could it be that sgdisk --clear doesn't work properly and there's still partitions on the disk when i run disko multiple times? * sure it could. blkdiscard fixes that. success after blkdiscard, no changes to disko config | 10:43:26 |
| lassulus | what command are you running? | 10:43:51 |
| nrv | In reply to @nevrv:matrix.org
sure it could. blkdiscard fixes that. success after blkdiscard, no changes to disko config wait, abort | 10:44:47 |
| nrv | i got excited on one success message which was not the right one | 10:45:12 |
