[SOLVED] Boot issue NVME HP SSD EX 900 with pve-5.4.55-1-pve

[irattarto]

Hi Everyone,

I just upgraded the system (and kernel got updated too) to pve-kernel-5.4.55-1-pve
But it does not find root zfs anymore. If boot to previous kernel it works. I even tried reinstall, and afther upgrade the same problem.



If i try to manually import, no pool found.

 

[Stoiko Ivanov]

seems the pool could not be imported - maybe related to a missing/corrupt cache file...

what's the output of:
`zpool import`
(run in the initramfs shell)

 

[irattarto]

what's the output of:
`zpool import`

It is showing the other pools, but rpool is not showing.

 

[Stoiko Ivanov]

do you see the disk where rpool is supposed to be on in the output of `blkid` or `lsblk` (not sure if all of them are available) or in `ls -laR /dev`?

 

[irattarto]

do you see the disk where rpool is supposed to be on in the output of `blkid` or `lsblk` (not sure if all of them are available) or in `ls -laR /dev`?

No, it is completely missing.

 

[irattarto]

I read somewhere that HP EX 900 SSD had issues with the controller of the SSD drive, but in older kernels, and in theoretically it is fixed...
With other (previous) kernel it just boots fine for me (but maybe finaly it will be only inside the 5.5?). Do not find the link again.

 

[Stoiko Ivanov]

Hmm - if you can boot into older kernels - you could try to narrow down until which version it works (just install the pve-kernel packages and see from when it does not boot)

Additionally - are there maybe any firmware upgrades available for your system or the SSD? - This can help in such situations

 

[irattarto]

Hmm - if you can boot into older kernels - you could try to narrow down until which version it works (just install the pve-kernel packages and see from when it does not boot)

Only the current version 5.4.55-1-pve does this. All other previous I tested works.

 

[Stoiko Ivanov]

anything in the `dmesg` output when booting into 5.4.55-1, which might indicate what the problem is?
(if possible compare to the dmesg output of 5.4.44-2)

 

[irattarto]

Well the difference related to the nvme drive and dmesg I found:

Working kernel:
[ 3.396539] nvme nvme0: missing or invalid SUBNQN field. [ 3.405780] nvme nvme0: allocated 64 MiB host memory buffer. [ 3.436120] nvme nvme0: 15/0/0 default/read/poll queues [ 3.443329] nvme nvme0: Identify Descriptors failed (2) [ 3.445452] nvme nvme0: Identify Descriptors failed (2) [ 3.446815] nvme0n1: p1 p2 p3

And 5.4.55-1-pve
[ 3.402276] nvme nvme0: missing or invalid SUBNQN field. [ 3.411731] nvme nvme0: allocated 64 MiB host memory buffer. [ 3.441636] nvme nvme0: 15/0/0 default/read/poll queues [ 3.448693] nvme nvme0: Identify Descriptors failed (2)

I did not find other difference, but seems the new kernel does not handle the device well.
BIOS update did nothing, I look for nvme firmware update.

 

[irattarto]

anything in the `dmesg` output when booting into 5.4.55-1, which might indicate what the problem is?
(if possible compare to the dmesg output of 5.4.44-2)

Hm looking afther it. My problem seems like this:
https://bugzilla.kernel.org/show_bug.cgi?id=205679
But thats a theoretically fixed one, and I did not have any issue until now.

 

[Stoiko Ivanov]

Hmm - seems that the partitions are not found - maybe try setting the kernel cli parameter from:

https://unix.stackexchange.com/questions/470778/nvme-missing-or-invalid-subnqn-field

Hm looking afther it. My problem seems like this:
https://bugzilla.kernel.org/show_bug.cgi?id=205679
But thats a theoretically fixed one, and I did not have any issue until now.

yes and that fix is the current kernel - so that's not the cause.

 

[irattarto]

Hmm - seems that the partitions are not found - maybe try setting the kernel cli parameter from:
https://unix.stackexchange.com/questions/470778/nvme-missing-or-invalid-subnqn-field

Does not work

 

[irattarto]

I stayed with previous kernel, and now the 5.4.60-1-pve solved it.
Thanks for trying to help me!