[3.0.1-rc6] order of cicustom elements forces replacement of entire vm #1224
Labels
issue/confirmed
Issue has been reviewed and confirmed to be present or accepted to be implemented
issue/workaround
it is or has a workaround
resource/qemu
Issue or PR related to Qemu resource
type/bug
Im using the
cicustom
attribute to provide my VMs with a cloud-init config that consists ofmeta-data
,network-config
anduser-data
. In code this looks something like this:This works and my VM's get successfully deployed. But when make a change to the code i.e. assign more ram or something else that would not justify a replacement, I get the following output from
terrafrom apply/plan
(example for vm name lff-r-vslnxc04):The order of the cloud-init files that is received from the current state is different to the order I have in code. I would assume this should be no problem, but this triggers a complete destroy and reapply of my infrastructure. When I change my code to reflect the
cicustom
order which the current state provides it works and all the desired vms get an update instead of a replacement.So the issue is rather minor and could probably be addressed by a comment in the documentation regarding the ordering of the
cicustom
cloud-init config elements. I don't know if this behavior is intentional hence I decided to report it just in case.The text was updated successfully, but these errors were encountered: