On the Download Fedora server page, which image is recommended? Standard, netinstall or raw.
I’m going to use it for a home file server, which is not too heavy on network traffic.
Use the mount command and if you see the device mounted then it needs to be unmounted. lsblk also gives you the device path which may be a /dev/sdX or /dev/mmcblkX device.
For example, with my sd card reader I see the sd card as /dev/sdf and it usually mounts /dev/sdf1 and /dev/sdf2. Thus I use the command sudo umount /dev/sdf1 and repeat that for sdf2 so both partitions are unmounted.
The image you select for the transfer using the arm-image-installer should be the one you download in from question 3. Note that the image installer is only for use with the raw image.
Thus --image will be the path/to/the/image/downloaded (full or relative). It is easiest if you download to some location then cd to that location before running the installer. I download to Downloads then ‘cd Downloads’ and use ‘ls’ to see the exact file name to use as the image name. --target is not needed for either the Pi 3 or Pi 4. That target is only if using uboot which those boards do not need. Fedora by default installs grub to boot. --media is the device identified in 1 above. For me it is usually /dev/sdf but should be whatever you identified it as in step 1.
I also add --resizefs so the installer can automatically expand the / file system to fill the sd card. The image is only ~9 GB without that and you would manually have to expand the file system if not using the --resizefs option.
You can get a lot of info about the arm-image-installer by simply using that command with no options and it will give a help screen.
use the raw image. That is the one the image installer is designed to use.
It is your choice on which type image to download, but I use the server image since I manage the Pi from my desktop and do not need the gui desktop running on the Pi. However, if you choose to use VNC or similar to manage the Pi then you would probably select the workstation image.
That is assuming that the command “lsblk” (with the card plugged in) showed the device /dev/mmcblk1. If it did not then use whichever device “lsblk” shows but the whole path as I showed above.
I checked point 1 with the following steps.
$ sudo fdisk -l
Device Boot Start End Sectors Size Id Type
/dev/mmcblk1p1 8192 62521343 62513152 29.8G c W95 FAT32 (LBA)
/dev/mmcblk1p2 2048 8191 6144 3M 83 Linux
Partition table entries are not in disk order.
I guess this answers your question 3.
I missed the step to unmount - my bad. So I unmount it and re-run the image writing commands.
$ umount /dev/mmcblk1
umount: /dev/mmcblk1: not mounted.
However, it comes with the following message.
Partition table entries are not in disk order.
The partition table has been altered.
Calling ioctl() to re-read partition table.
/dev/mmcblk1: fsync device failed: Input/output error
Error: mount /dev/mmcblk12 /tmp/boot failed
mmcblk1 is the device not the partition. The device is not normally mounted, but the partitions with file systems usually are.
If when you insert the card reader it mounts the partitions you can see that with the mount command. The individual partitions will be what you just saw with the fdisk -l command.
Thus the commands to ensure the device is not mounted before writing to it would be
The partitions were not created correctly.
What I did is;
$ sudo fdisk /dev/mmcblk1
$ sudo mkfs.vfat -f /dev/mmcblk1
The partition type was set to W95 FAT32 (LBA).
$ lsblk
mmcblk1 179:24 0 29.8G 0 disk
├─mmcblk1p1 179:25 0 5.6G 0 part
└─mmcblk1p2 179:26 0 24.2G 0 part
Done.
I recall the partitions need to be formatted (activated) and labeled.
What else did I miss and what did I do wrong?
Please bear with me. I used the Fedora Media Writer for the Workstation set-up and never did the file system/partitions, let alone Raspberry Pi Partition.
Put the newly written card into the R Pi 3 device and boot.
You should do nothing else but what I have listed to the sd card.
Step 4 is to clean out any corruption that may remain from prior failed attempts. As long as the device is still identified as /dev/mmcblk1 in step 2 then the commands as written should be all that is needed.
It seems you are trying too hard, and doing a lot of things that may interfere.
Do you not understand that the image writer does a complete partitioning, formatting, and writing process to create a bootable device from the downloaded image?
Do you not understand the difference between a device (/dev/mmcblk1) and a partition on that device (/dev/mmcblk1p1)?
Do you not understand that a partition with a file system (/dev/mmcblk1p1) is what the system mounts (and needs to be unmounted before writing the image to the device)?
It is not required that you do anything else but use the image writer to create the bootable sd card from the image.
No wonder it always shows as mmcblk1. I suspect the mmcblk0 is reserved for the full size sd card slot. I had a usb card reader that did similar with 2 slots. My newest one does not do that.
If you use a usb card reader it likely will show the sd card as a standard flash drive designation (/dev/sdX) instead of /dev/mmcblk1. Other than that everything else should be the same.
I believe, when first booting up, you get an initial setup dialogue.
There you can set your computer name, interface config, user and things I don’t remember.
The network time comes to mind.
This is at least with the pi 4 the case.
But one shouldn’t quit this dialogue before setting up an account.
The arm-imagerdoesn’t seem to work on SilverBlue right now (I’m on Aurora-dx), so the process of creating a bootable Fedora arm image is a maze of twisty passages, all different.
At this point, I’m tempted to stay with Debian on the pi because that takes 5 mins and this it just going to be n Ansible control server.
Please do not reopen necro threads that already have a solution.
I am closing this one and if you need to discuss an issue with your current problem open a new thread. 3 years ago fedora 35 would have been the current platform.
