Boy, am I glad to have found this thread. I did an upgrade from 24.04.2.3 to 24.10.0.1 via GUI, and everything worked OK, incl. application migration. Then, I wanted to make a clean install from *.iso. And that’s about that. No boot… This disk wiping of the installer is absolutely not working (also mirrored drives here, in case this matters). Nothing would install anymore, even several 24.04.x *.iso installers would not generate a bootable system.
Only after using @apple4ever 's solution (thanks ) could I install successfully and also boot afterwards…
Hoping, this gets fixed or made more resilient in future versions…
I just wiped anotyher boot drive and rebooted from a usb stick with 24.10.02 then reloaded the saved configureation file. Sadly I got the same pool_name KeyError so perhaps it wasnt my boot drive! This is very frustrating as it stops any further disk configurations.
My only next safe thing to try is copying off everything on the mirrored pair of ssds and wipe those and copy back.
I just ran into this issue with 24.10.1 and 24.10.2.
HW: HP Proliant Micro N54L, 16 GiB RAM, LSI 9211 HBA for both original drive slots and four 2.5" slots added in place of optical drive.
When I got to the error message I manually tried importing the boot-pool.
/sbin/zpool import showed no pools, but /sbin/zpool import -d /dev showed my boot-pool. I imported it /sbin/zpool import -d /dev boot-pool and then exit and the first boot after install continued fine.
BUT … I still have the problem whenever I reboot. I started with 24.10.1 and upgraded to 24.10.2 and still have the issue. So I have to be at the console when I boot to import the boot-pool. I did open a BUG ticket using the smiley face.
My boot -pool is a 2-way mirror of Kingston V300 120GB SSD.
I also noted that during subsequent reboots, a /sbin/zpool import does find my non-boot-pool pools. I have a 2-way mirror of Samsung 850 500GB SSDs (Apps) and a 4 drive RAIDz1 for backing up my primary server (ZFS replication).
The Kingston SSDs did have 24.04 installed perviously and I have NOT tried the blkdiscard solution given above, but if that is still the problem, then the fix for it in 24.10.1 did not work.
Late to the party and definitely not a Linux expert, but did anyone experiencing these problems try and wipe the drive(s) destined for the boot pool with dd(1) already?
If you spot any errors or typos please let me know.
The underlying problem was corrupt ZFS disk labels (positions 2 and 3) on the base device (/dev/sda) while the correct labels were on partition 3 (/dev/sda3). I do NOT know WHY this caused the zpool import to not find the zpool but for a zpool sport -d /dev did find the zpool.
I’m sure that would have worked if allowed to get all the way to the end of the device, since I would the corrupt labels were in positions 2 and 3 (at the end of the device) the labels at positions 0 and 1 were correctly cleared.
The installer is NOT clearing ALL the old ZFS labels, only the two at the start of the device, it is leaving the 2 at the end there (and corrupt in some way).