NVMe drive in software raid fails to format

nullroute

Hello, I am having a strange problem when formatting 2 NVMe Netac disks, they were in Software Raid 1 with md0 array however when formatting these disks they are recognized as multiparthed and the operating system is not able to build a Raid again.

Another problem is that these disks in Debian 11 also seem to have problems, the system has Kernel Panic and still seems to have some remnants of previous boot files

I also used this formatting technique and nothing resolved:

sudo dd if=/dev/zero of=/dev/[PATH OF DEVICE] bs=1M count=1" to clear the partition information on the disk. Be sure to replace "[DEVICE PATH]

Here's how it recognizes when starting with ubuntu:

Debian 11 crash:

Nvme disks automatically pull the multipath and single-raid it in multipath.
When trying to force it not to mount the formatting, it generates an error when installing and does not mount the image.

Does anyone have any idea what could be causing the problem or how to solve it?

djn

try wipefs --all /dev/DEVICE

nullroute

@djn said:
try wipefs --all /dev/DEVICE

Technically debian does this when installing the OS, but even so the problem persisted, as these disks are already outside the dedicated server, do I need to install the nvme unitary and type the command?

Calin

Hello @nullroute you usage NVME Adapter? For this NVME? I'm remember , in the past I'm usage NVME Adapters for very old servers with no NVME Support , and I m make lot of problems

Regards,
Calin

sparek

Try clearing the superblocks on each device

mdadm --misc --zero-superblock /dev/%devicename%

Also, correct me if I'm wrong, but:

dd if=/dev/zero of=/dev/[PATH OF DEVICE] bs=1M count=1

is only writing 1MB of /dev/zero to /dev/[PATH OF DEVICE], because you have the count set to 1.

nullroute

@Calin said:
Hello @nullroute you usage NVME Adapter? For this NVME? I'm remember , in the past I'm usage NVME Adapters for very old servers with no NVME Support , and I m make lot of problems

Regards,
Calin

Hello, no. They are mostly on a B550M with 2 m.2 slots directly on the motherboard.

vsys_host

You can also formatting dd if=/dev/zero of=$YOUR_DEV bs=512 seek=$(( $(blockdev --getsz $YOUR_DEV) - 1024 )) count=1024

rm_

@nullroute said: Here's how it recognizes when starting with ubuntu:

Debian 11 crash:

Are there supposed to be screenshots? Not seeing any. And "I get an error" is not a useful problem report.

nullroute

@rm_ said:

@nullroute said: Here's how it recognizes when starting with ubuntu:

Debian 11 crash:

Are there supposed to be screenshots? Not seeing any. And "I get an error" is not a useful problem report.

Hi, there is a screenshot in the thread, maybe try disabling your adblock/vpn

nullroute

@vsys_host said:
You can also formatting dd if=/dev/zero of=$YOUR_DEV bs=512 seek=$(( $(blockdev --getsz $YOUR_DEV) - 1024 )) count=1024

Also tried, it looks like these drives have dead sectors that are not successfully cleaned, currently replaced with perfectly functioning Kingston drives.

rm_

@nullroute said:

Hi, there is a screenshot in the thread, maybe try disabling your adblock/vpn

Got it now, imgur returns 0 byte files when saying "too many requests" (from OVH).

As for the first screenshot, it is not clear why use multipath, the software RAID should be set up with mdadm. For the second one, there doesn't seem to be a crash, just the installation in progress. Or did it hang always at this moment with no further progress and no error messages?