Hello, I get "ZFS: i/o error - all block copies unavailable" whenever I try to boot from the zones zpool, however the machine boots fine using the usb stick. I've tried piadm -d and piadm -e to no avail. The pool is fine according to zpool scrub. Also, I've a few crashes under load and vmdumps available if needed.

huy bios or UEFI boot? any other error messages?

how big are the disks used with zones pool?

with BIOS boot, it is possible you have hit by INT13 limit to access large disks, UEFI boot should be fine. if you have crashes (vmdumps), then it would be nice to at least see the panic messages (vmdumps are not related to boot loader error above).

tsoome_: bios. There are 4 4Tb disks configured in zraid-1

if available, try UEFI boot.

unfortunately, the microserver gen8 has no uefi apparently

I see, yea, gen9 is the first one.

essentially, the problem is that you need to load kernel + boot archive from smaller disk or alternate storage (like usb or network boot)

Is there a way to make sure /boot is within INT13 limit? I'm pretty sure it used to work

Here's the panic message: 2025-10-13T09:57:49.107237+00:00 proliant savecore: [ID 570001 auth.error] reboot after panic: pageout_deadman: stuck pushing the same page for 90 seconds (freemem is 17425)

I think the crashes happen when I compile stuff under LX zones

thats part of the problem -- with zfs there is no way to enforce particular location on disk. So when you do update, the PI will be stored in new blocks and when pool is getting full, eventually new blocks are allocated from high LBA and if there is actually LBA limit, you will hit it...

ah, the pageout_deadman issue is known, it happens when you are in low memory situation and the system starts to shrink zfs cache (arc) but it will take too long, so pageout_deadman will fire and kill your system.

thanks

we are looking for relief for deadman issue but this will take some time still.