Re-applying over a failed run is not supported until the installer migrates to Terraform 0.9. If you see the following output:
Error applying plan:
3 error(s) occurred:
* ignition_config.master: invalid file "8f8b62c46063e64e3709a2be1317bd83562abf43c3866d7adae023477b57fb47", unknown file id
* ignition_config.worker: invalid file "8f8b62c46063e64e3709a2be1317bd83562abf43c3866d7adae023477b57fb47", unknown file id
* ignition_config.etcd: invalid systemd unit "297994451bbc66a0b227eea50017c4d05fde37b99130d1efc2f6b14b3bd2a283", unknown systemd unit id
Destroy your cluster and start over.
If you see a failed tectonic.service unit, you will need to manually copy a file over to complete the install:
$ scp ./kubeconfig core@<your api public ip>:/home/core
Then SSH to the box:
$ ssh core@<your api public ip>
And execute this script to finish the install:
$ sudo bash /opt/tectonic/tectonic.sh kubeconfig /opt/tectonic/tectonic
If you see errors about missing modules:
Error configuring: 2 error(s) occurred:
* module.bootkube: provider localfile couldn't be found
* module.tectonic: provider localfile couldn't be found
These errors indicate that you are not using the customized Terraform binary bundled with the project. You can check this:
$ which terraform
/Users/coreos/tectonic-installer/bin/terraform/terraform
$ terraform plan -var-file=terraform.tfvars platforms/aws
2 error(s) occurred:
* module.masters.aws_autoscaling_group.masters: : invalid or unknown key: tags
* module.workers.aws_autoscaling_group.workers: : invalid or unknown key: tags
This error indicates that the .terraformrc is not being used. Be sure you have exported your config environment variable:
export TERRAFORM_CONFIG=$(pwd)/.terraformrc
Afterwards, you should be able to execute the desired action.