Hi,
On 9/4/20 5:21 PM, proxmox wrote:
Hello
I didn't know this patch mail got approved, so sorry for the (very) late
response.
My intention for not going with snippets was the fact that they could not be
created via the API and one would have to manually create a file on the target
machine for cloud-init userdata.
There is currently a bug report open for this:
https://bugzilla.proxmox.com/show_bug.cgi?id=2208
This together with backup and migration of snippets could be useful.
One possible use case was to spin up a kubernetes cluster on proxmox only via
API.
I wanted to have something similar to the hetzner cloud API where the full
userdata can be submitted for VM provisioning:
https://docs.hetzner.cloud/#servers-create-a-server
So going further here you want me to submit the MTU patches separately?
The MTU patches separately would be great.
Should I integrate userdata into the cicustom field? I thought this would make
things more complex in favor of parsing out the base64 stuff. So I would still
go with an extra field.
After some discussions we think that putting the userdata in the config
file is not the right approach. As the cluster filesystem is limited to
32M and a single VM config file is limited to 512K you can easily run
into the limit with a small number of VMs. We would most likely have to
limit the userdata per config to a very small amount (1K?). But with
those limits it is difficult to get everything in.
Thoughts?
_______________________________________________
pve-devel mailing list
[email protected]
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
_______________________________________________
pve-devel mailing list
[email protected]
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-devel
