NVME issues - wrong partition names if nvme is used (missing p)

[jTd7bPLFb]

I tried to install using your script (thank you btw.) on an nvme device.

the exports for partitions are not usable for nvme

ls -la /dev/nvme*
nvme0n1 nvme0n1p1 nvme0n1p2 nvme0n1p3 nvme0n1p4

invocation "sh install /dev/nvme0n1"

the exports in your script
export INSTALL_DISK=$DISK
export INSTALL_EFI_DEV=$DISK""2
export INSTALL_SWAP_DEV=$DISK""3
export INSTALL_ZPOOL_DEV=$DISK""4
export INSTALL_ZPOOL=${INSTALL_ZPOOL:-rpool}

so basically, they become in this case
/dev/nvme0p1
/dev/nvme0p12
/dev/nvme0p13
/dev/nvme0p14

which is wrong, as it misses the p

I noticed this on mkswap command failing.
mkswap: can't open '/dev/nvme0n13': No such file or directory

maybe put a if/then in the install.sh script, and also a small echo for each of the variables?

[psy0rz]

Yes this is indeed a problem.
I think there is a better way to determine device names from partition numbers, (sfdisk?)

Is there a reason why you dont use the image-method of installing?

[jTd7bPLFb]

I tried once again with the image-install method.
Did not work for me.

1. boot alpine iso
2. setup-network, setup-dns, setup-ssh

using the simple install image method

localhost:~# sh image /dev/nvme0n1
ALPINEBOX: Getting latest stable version...
ALPINEBOX: Writing https://github.com/psy0rz/alpinebox/releases/latest/download/alpine.img.gz to disk /dev/nvme0n1 ...
Connecting to github.com (140.82.121.3:443)
Connecting to github.com (140.82.121.3:443)
Connecting to objects.githubusercontent.com (185.199.110.133:443)
writing to stdout

•                100% |********************************|  467M  0:00:00 ETA
  

written to stdout
ALPINEBOX: All done, will reboot in 5 seconds..

after reboot, the machine hangs on
'Tearing down USB controller'

4. so, I to using the other method, ,with the separate script, when using official alpine installer
5. just added the p to the disk/device names, as mentioned, and installer working without issues.

So, to sum up, I used the Alpine iso, that is why I use the non image install method.
:)

[psy0rz]

Thanks,

Are you sure it didn't just hang during the shutdown sequence? Did you try hard-resetting it to see if it would boot?

[jTd7bPLFb]

Yes.

[psy0rz]

Its hard to reproduce here. Could you make a screenshot/photo of it? I assume not since its already installed now :)

[jTd7bPLFb]

which one? the issue is as stated at the beginning

if disk is sda
partitions sda1, sda2, sda3... and so on

export INSTALL_EFI_DEV=$DISK""2
export INSTALL_SWAP_DEV=$DISK""3
export INSTALL_ZPOOL_DEV=$DISK""4

but when nvme is used
partions are prepended with p
nvme0p1, nvme0p2 ... and so on

so basically I would just put something like
grep nvme $DISK
if true then just add p

export INSTALL_EFI_DEV=$DISK"p"2
export INSTALL_SWAP_DEV=$DISK"p"3
export INSTALL_ZPOOL_DEV=$DISK"p"4

[psy0rz]

No i meant when you use the imaging method. Could you screenshot that at the point it fails?