OneFlow vm_template_contents values are not taken into account

Hi all

We are using OneFlow to start a virtual cluster, in our workers VMs we need a contextualization variable to set the master ip called MASTER_IP . To avoid to include this variable hardcoded in our VM template we use the oneflow template instead, so it looks like this:

      "name": "Worker",
      "cardinality": 2,
      "vm_template": 1,
      "shutdown_action": "terminate",
      "parents": [
      "vm_template_contents": "MASTER_IP=\"\"",
      "elasticity_policies": [],
      "scheduled_policies": []

The problem is that vm_template_contents only is included when you instantiate the oneflow service the first time, if you scale the cluster or if you add a new Worker the MASTER_IP does not appear within the user VM section after instantiation, only the ROLE_NAME .

Someone else has found the same issue? we are using OpenNebula 5.2.1 at this moment



I’m not sure if this is related with this problem but if you create several onflow roles, each one with different vm_template_contents variables, you only see from Suntone (via “Advanced Role Parameters” option) the values from the last role. It looks that from suntone only are visible the vm_template_contents from the last role. The “advanced role parameters” looks empty for the rest (but the values are displayed correctly running oneflow-template show command)


I think that I found the issue, in fact there are 2 different issues.

The first problem (I don’t know if this is fixed in the latest ONE version) is if you use the oneflow wizard from suntone and you reload a oneflow service template the wizard only reloads the latest Role VM template content params, so it wipes the first roles vm_template_contents if you do not notice this and you save the changes.

The second problem is if you modify the vm_template_contents values that is not taken into account if the oneflow service is already running and you add more VMs or you resize a role, the new VMs will still use the old service template. (maybe this is done on purpose).

I will update to the latest OpenNebula version (I’m still using 5.2.1) just to see if I get the same results…