Can't create VM without specifying an affinity_host - issue with an empty string

[killroy1971]

I'm trying to create a diskless VM. The VM creation part works, however I need to include an affinity_host line in my VM's configuration, so I used the data.xenorchestra_pool.pool.id item, which in my mind should allow the hypervisor to choose the best host within the Homelab pool. The VMs create, but they can't be started because the hypervisor never assigns a host to the VMs.

Ultimately, that's what I would like to do. Omit the affinity_host configuration item entirely, or specify the xenorchestra.host as "none" on the affinity_host line, and have the hypervisor choose the host.

main.tf has:
data "xenorchetra.pool" "pool" {
name_label = "Homelab"
}

vm.tf has:
...
memory_max = 107374182400
cpus = 2
hvm_boot_firmware = "uefi"
name_label = each.key
name_description = each.value
affinity_host = data.xenorchestra_pool.pool.id
template = data.xenorchestra_template.talos_control.id

network {
network_id = data.xenorchestra_network.net.id
}
...

[collegue1337]

Same issue here,

 Error: jsonrpc2: code 10 message: invalid parameters: {"errors":[{"instancePath":"/affinityHost","schemaPath":"#/properties/affinityHost/minLength","keyword":"minLength","params":{"limit":1},"message":"must NOT have fewer than 1 characters"},

[illethrias]

+1 with that issue

[olivierlambert]

Pinging @julien-f so we can fix this (likely in XO)

[ddelnano]

This is essentially the same root cause as #238 and my work in progress PR is one way to address this (#239). I unfortunately haven't had time to focus on that change and it still needs more work before it's ready to merge.

If there is a way to handle this on the XO side, that would be great.

[killroy1971]

Is the error message coming from the XO side, because of a requirement or because of how the VM is being created by terraform?

[ddelnano]

The error message is coming from XO. The api was recently updated to enforced that optional fields should be null rather than empty strings. See #231 for the link to the XO commits.

[marlluslustosa]

#231 (comment)

[killroy1971]

Sorry for the delay. Travel, work, and stuff. Plus, it's been nice outside in my area.

I ran across this in the documentation:
# Prefer to run the VM on the primary pool instance
affinity_host = data.xenorchestra_pool.pool.master
network {
network_id = data.xenorchestra_network.net.id
}

This solves my problem, I can't say if this was added since I started playing with terraform or not. Either way, my problem is solved.

[OTonGitHub]

What on earth pool.master mean? What's the point of adding in pool via label as a data parameter then?

[nathanael-h]

It's the master node (=physical server) of a pool. Read more in XCP-ng and Xen-orchestra docs 😉 Le 14 février 2025 21:44:46 GMT+01:00, OTonGitHub ***@***.***> a écrit :

OTonGitHub left a comment (vatesfr/terraform-provider-xenorchestra#243) What on earth pool.master mean? What's the point of adding in pool via label as a data parameter then? -- Reply to this email directly or view it on GitHub: #243 (comment) You are receiving this because you are subscribed to this thread. Message ID: ***@***.***>

-- Nathanaël Envoyé depuis /e/OS Mail.

[OTonGitHub]

@nathanael-h I apologize for my stupid comment, I later on read a lot, and understood it well and made a little template the next day here as well https://github.com/OTonGitHub/Terraform-XO to get started.

[olivierlambert]

There's no stupid questions, glad you managed to move forward!