Can't find bootable partition

[h***@tuta.io]

Hi,

I have 14.0-RELEASE installed with two drives in a ZFS mirror on a x86_64 system. I've been testing a very minor patch, rebooting, testing, etc. This is installed with EFI.

Out of the blue, it won't come back.

It detects both drives. I see these errors.

zio_read error: 5
zio_read error: 5
zio_read error: 5
ZFS: i/o error - all block copies unavailable
ZFS: failed to read pool zroot directory object

Can't find /boot/zfsloader
Can't find /boot/loader
Can't find /boot/kernel/kernel

I'm a little mystified. I know there were some talks of ZFS corruption. I was using the latest patches for the kernel, although world was from 14.0-RELEASE as it came out.

I rebooted this probably a dozen times at least with no issue prior to this.

Where should I start? Does this sound like I made some blunder or that the hardware randomly failed in some strange way?

Thank you!

-Henrich


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[h***@tuta.io]

I booted the 14.0-RELEASE installer and mounted zroot to /tmp/mnt.

There's nothing in the zroot, except other mounts (like /usr, /home, and /var, the defaults). Those appear to have good data. But there's no /tmp/mnt/COPYRIGHT, /tmp/mnt/boot, etc.

I did a very lengthy memtest on this machine without issue. I don't have any particular reason to suspect hardware.

It's as if every file/folder directly under zroot/ROOT/default has vanished.

There's no zpool errors.

I made no changes to the filesystem in the boot prior to the reboot.

-Henrich

Post by h***@tuta.io
Hi,
I have 14.0-RELEASE installed with two drives in a ZFS mirror on a x86_64 system. I've been testing a very minor patch, rebooting, testing, etc. This is installed with EFI.
Out of the blue, it won't come back.
It detects both drives. I see these errors.
zio_read error: 5
zio_read error: 5
zio_read error: 5
ZFS: i/o error - all block copies unavailable
ZFS: failed to read pool zroot directory object
Can't find /boot/zfsloader
Can't find /boot/loader
Can't find /boot/kernel/kernel
I'm a little mystified. I know there were some talks of ZFS corruption. I was using the latest patches for the kernel, although world was from 14.0-RELEASE as it came out.
I rebooted this probably a dozen times at least with no issue prior to this.
Where should I start? Does this sound like I made some blunder or that the hardware randomly failed in some strange way?
Thank you!
-Henrich

--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[h***@tuta.io]

Hi everyone,

I had previous thread titled: Can't find bootable partition

I also have a forum thread for this, but it's slow going as my account is not yet verified and each post requires approval: https://forums.freebsd.org/threads/14-0-stopped-booting-mysteriously-zfs-mirror.93101/

Here's what I had to do to produce a non-booting system:

Details:

14.0-RELEASE
x86_64
2x 16TB harddrives
ZFS mirror installed with GPT UEFI.
"Automatic" ZFS install.
8GB RAM
Hardware has ECC and passed memtest.
Fully updated with freebsd-update.
Intel S1260 Atom (non-speculative execution processor)

After writing over 1TB of random data to the filesystem and pressing the reset button on the server, (a regular reboot might've been sufficient as well, but not sure) it booted but the boot loader had "zio_read error: 5" errors plastering over top of the menu. It still booted, nonetheless.

After that successful boot, I rebooted and the boot loader failed quickly, dumping me into a UEFI shell.

I've had at least two similar failures like this already. I don't suspect a full 1TB written is required to cause it. BIOS boot, from a previous install, also did not appear to work. I also don't suspcet that patches, or lack of patches, influences the issue.

It's *possible* that somehow hardware is to blame, but while running the server has had no instability or odd behaviors. I've done make kernel and make world without incident. I've rsynced several TB without issue as well.

Whenever this happens, I can zpool import zroot from the installer and zpool shows no errors. And smartctl shows no errors (this is the third pair of drives I've had this happen with). I've yet to have a zpool scrub come back with errors.

Has anyone else had any issues like this? It's very perplexing and concerning.

Thank you!

-Henrich


--
Posted automagically by a mail2news gateway at muc.de e.V.
Please direct questions, flames, donations, etc. to news-***@muc.de

[Stefan Bethke]

Post by h***@tuta.io
I also have a forum thread for this, but it's slow going as my account is not yet verified and each post requires approval: https://forums.freebsd.org/threads/14-0-stopped-booting-mysteriously-zfs-mirror.93101/
2x 16TB harddrives

Some BIOSes (and maybe UEFI too) might have issues accessing blocks beyond a certain number. Try creating a pool just for the OS that is small-ish (1 TB max) and see if you can still reproduce the issue.


Stefan