I know there are some posts already in the forum related to “CloudInit- No datasource found”. I have read though those, but I could not figure out the problem in my case.
Here is my setup.
1 VM 4vCPU / 16GB Memory hosting MaaS Regiond + Rackd (tried with snap earlier, same bug. So switched to PPA to see if it fixes, but the problem remains. Now MaaS is installed from PPA).
Now I boot a machine, shows up in MaaS UI, I manually commission the machine, all good the machines comes to ready state.
When I deploy the machine to
Memory: All is fine. The OS comes up and I can ssh in to the machine. All good
Disk: Here is my problem. During the deployment’s first stage when the os is loaded to memory, it can connect to cloud init fine. Now when it reboots from the disk data, it cannot connect to cloud init, says “Cloud Init: No data source found”. All the older forum post suggested issues with Maas controller connectivity from the target machine. But it did connect to cloudinit during the first phase of deployment.
What I did now to troubleshoot?
I logged in to single user mode, checked netplan all good. IP’s were good. Its from the same segment as maas. (MaaS is .130, the machine had .132). All good here. I can ping and curl maas IP.
I check the cloud init data source file, that looked fine as well. I restarted the cloud-init in the singleuser mode and checked cloud-init status --long. It shows no errors and shows MaaS :5248/MAAS/metadata as the datasource.
Another observation: this problem happens in our other setup, that used to work fine earlier. Same exact problem.
Deploying Jammy 22.04 BTW. Any help would be appreciated. If you need any detailed logs, I can provide you that too. Please let me know. We are in a time crunch to deploy 50 servers and this bugs happens!
We’ve discovered several issues that lead to such behaviour, however without knowing more details it is hard to diagnose if thats your case.
Can you tell us a bit more about your network topology and if you are using stable or candidate images or custom built?
Another observation: this problem happens in our other setup, that used to work fine earlier. Same exact problem.
Based on this statement, it feels like it could a cloud-init related issue.
Are you using OVN? Because we’ve discovered an issue related to OVN recently. And if thats the case, these two changes might fix it for you (since you are using deb package, should be relatively straightforward to apply these changes). https://git.launchpad.net/maas/commit/?id=cb8d38ef978e5048125cc36a294055b51dd528f6
Our network topology is pretty simple on the switching Layer. One subnet / vlan dedicated to PXE and only used for MaaS provisioning. MaaS (Regiond+Rackd) is on the same subnet as the target machines (L2)
On the OS side, its just 2 interface bonded with active-backup setup. No OVN or OVS bridge and our Ubuntu Jammy is coming from the MaaS Stable repo that comes configured as default.
Booting to single user mode and downgrading cloud-init to cloud-init=23.1.2-0ubuntu0~22.04.1 and then rebooting the node fixes the issue. My MaaS machine is synced with the latest image (Wed, 02 Apr. 2025 07:23:04 UTC)
The bug report say its fixed with " 24.4.1-0ubuntu0~24.04.2" But this does not seem to the case. I still get the same error with this version.
I have used packer-maas repo and build the cloud-img with this mod. That worked fine for me. Is there any ETA to have the updated patch in the maas.io image?
The latest cloudinit version did not fix the error, btw.
