[SOLVED] Cloud-Init Drive Not Auto-Regenerating

[martincarpenter]

Hello,

I am experiencing an issue with the CloudInit Drive 'auto-regenerating' after a clone, a virtual hardware hotswap, or even a reboot. The Regenerate Image button must be manually clicked in the GUI.

• I have used Packer to create an image of Alma 9 Linux with cloud-init installed. It also pushed the below line to /etc/cloud/cloud.cfg as per the docs.
  datasource_list: [ NoCloud, ConfigDrive ]
• I then provision a VM via Terraform with only the ipconfig0 for cloud-init defined.
• After booting if I run systemctl start cloud-init I get a line in the logs saying no datasources found. If I do any combination of hotswapping hardware, rebooting, running cloud-init manually, no datasources are found and the ipconfig0 IP.
• After clicking the Regenerate Image button in the GUI and rebooting the VM, the CloudInit Drive and IP will be picked up by the VM as expected.

I have had this kind of thing running with an older setup with an older Proxmox/CentOS in the past, without using Packer and Terraform.

Have I come across a bug? Is there a workaround for this perhaps? This is essentially completely blocking the automation of any VM provisioning with this method.

Any help would be hugely appreciated !

 

[martincarpenter]

Just incase anyone comes across this same issue, I have got it working by simply apending ide0 to the the bootlist. I am not sure WHY this would be needed but it is. I found it by grabbing the output of the Proxmox QEMU VM config file VMID.conf output before clicking Regenerate Image, and then again after and diffing them both - the difference being the below.

4c4 < boot: order=scsi0 --- > boot: order=scsi0;ide0 9c9 < ide0: ssd:vm-10139-cloudinit,media=cdrom,size=4M --- > ide0: ssd:vm-10139-cloudinit,media=cdrom

Adding the below to the Terraform Proxmox resouce .tf allowed the VM to pick up the cloud-init config immidiately after cloning.
boot = "order=scsi0;ide0"

Previously I had this.
boot = "order=scsi0"