Open myxxmikeyxx opened 3 years ago
I am using Ubuntu 20.04 LTS in a virtual machine on virtual box. I don't think that should matter but I wanted to point this out incase it helps to find the problem.
@pandel may be able to help with that one.
In the meanwhile, you could try building with USE_QCOW2=0
@myxxmikeyxx I have no idea what's happening here. You are too far into the build to just say, that your NBDs simply don't work. If I find time, I'll setup a virtual machine like you did and try to find a reason for this.
Okay, not a problem. I am currently setting up a new VM on Ubuntu LTS 16.04.6 amd64. I saw on earlier post that some people were using that and didn't have any build problems.
On Ubuntu 16 I am getting this error:
root@ubuntu16VM:/home/install17/pi-gen/stage0# ./prerun.sh ./prerun.sh: line 4: bootstrap: command not found root@ubuntu16VM:/home/install17/pi-gen/stage0#
What I don't understand is the device name /dev/mapper/nbd10p2
. So inside the build process the system is trying to access nbd10, but it is nbd1 actually. I am not sure why this is happening and have never noticed such behaviour in my environment...
You cannot issue the prerun.sh command outside the build process, as it is executed inside the docker environment... that simply does not work.
Sorry, last comment was wrong, I deleted it...
Try using the Docker build variant via build-docker.sh perhaps.
I have switched over to a raspberry pi 3. I am testing doing the build on it. The build seems to be working, slower, so I do not know if it will get hung up on the same spot. I also downloaded the x86 Raspbian img and I am trying that on the Virtual Box, I just got all the dependences and cloned it. I will update you on how it goes. If it fails then I will try it back on Ubuntu (both) and try the docker build to see if it works.
Also thank you @pandel for trying to help. I am not super tech savvy. I really want to get this to work so I can create my own IMG that has different packages pre-installed. My goal is to basically try to make a LazyCast img file. That still works with Raspberry Pi Imager.
I want it to have the necessary packages but still be able to get the Wi-Fi info and then I parse that info from the wpa_supplicant and run a script to take the Wi-Fi name and password and connect to it through network-manager in command line.
sudo nmcli dev wifi connect 'Wi-Fi Name' password 'my Wi-Fi Password
Raspberry pi x86 error:
Raspberry pi: Still running
@myxxmikeyxx Strange, as the build succeeded itself, only image conversion at the end failed. I know that this "failed to setup loop device" error was something, that could happen in an older Docker release on my Mac, too, and I also noticed it in a Debian ARM64 VM on my Proxmox system. But it was related to timing issues in combination with read/write cycles of the Proxmox hypervisor and solved itself via a Proxmox kernel update. I am really sorry, but do not have any idea, how I could help you without a deep dive into your system by hand, which isn't possible, obviously...
I get the same error message. Ubuntu 20.04, on vmware esxi. I'm building with ./build-docker
:
[01:25:22] End /pi-gen/stage1/03-install-packages/00-packages
[01:25:22] End /pi-gen/stage1/03-install-packages
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs: 10.9 GiB (11649982464 bytes) trimmed
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs/sys
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs/proc
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs/dev/pts
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs/dev
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs/boot
/pi-gen/work/2021-06-01-Raspbian/stage1/rootfs
/dev/nbd0 disconnected
[01:25:22] End /pi-gen/stage1
[01:25:22] Begin /pi-gen/stage2
Creating backing image: image-stage2.qcow2 <- /pi-gen/work/2021-06-01-Raspbian/image-stage1.qcow2
Formatting '/pi-gen/work/2021-06-01-Raspbian/image-stage2.qcow2', fmt=qcow2 size=12884901888 backing_file=/pi-gen/work/2021-06-01-Raspbian/image-stage1.qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16
read error, sector 0
llseek error
llseek error
llseek error
mount: /pi-gen/work/2021-06-01-Raspbian/stage2/rootfs: special device /dev/mapper/nbd0p2 does not exist.
[01:25:22] Unloading image
Building with USE_QCOW2=0
works, so that's what I'm doing for now.
@rianadon Thanks for your information. It must have something to do with 20.04. Unfortunately, I am not able to test this atm, as I am still moving and busy. I try to get back to this as soon as possible...
Thanks for looking into this @pandel! For me at least, this isn't a blocker (qcow2 disabled fixes the read error sector 0 problem and successfully builds).
I wonder if it's running in a virtual machine that could be the problem as well. I was running inside docker, which as far as I understand, would eliminate ubuntu quirks. If it would help I also have an Arch VM I could try building inside to test the VM theory.
I am running into this error and I am not sure what is wrong.
[11:01:27] Begin /home/install17/pi-gen/stage4 Creating backing image: image-stage4.qcow2 <- /home/install17/pi-gen/work/2021-05-26-Raspbian/image-stage3.qcow2 Formatting '/home/install17/pi-gen/work/2021-05-26-Raspbian/image-stage4.qcow2', fmt=qcow2 size=12884901888 backing_file=/home/install17/pi-gen/work/2021-05-26-Raspbian/image-stage3.qcow2 cluster_size=65536 lazy_refcounts=off refcount_bits=16 read error, sector 0 llseek error llseek error llseek error mount: /home/install17/pi-gen/work/2021-05-26-Raspbian/stage4/rootfs: special device /dev/mapper/nbd0p2 does not exist. [11:01:28] Unloading image
The build.log file is bellow: