Found a typo in the DHCP server config that was giving the BOOTP clients the wrong path for the O/S image. After fixing this, I’m finally seeing proof that the two are talking as the image is downloaded and executed by the SOPINE compute module.
This is major progress! However the node still fails to boot Linux. Based on the console output it looks like I need more files on the tftp server. I have a vague memory of reading about this somewhere so it’s not a complete surprise.
I think the next step is to make an Armbian SD card, boot the SOPINE module with it and complete the setup process via the serial connection. Then I can take the contents of the /boot directory and copy them over to the tftp server root. I think this will give me what’s needed to get past this error.
I don’t think that will be enough to get all the way to a login proper though. I think I still need to share the rest of the O/S files over NFS from the head node, and tell the compute node to mount this shared directory (maybe by tweaking the files in the tftp root?). My plan for now is to take this next guessed step and learn what I can from the error messages.
The system is slowly coming together. Hardware-wise is have a bunch of ideas to refine the setup but I’m trying really hard to avoid getting distracted by that until I get the whole cluster booting.
#journal