Can't run terraform destroy on workers been rebooted

[simox-83]

If we reboot the worker instances for any reasons and then trying to destroy the cluster, when running terraform destroy on the worker.tf we get an error message:

* data.local_file.user_data: 1 error(s) occurred:

* data.local_file.user_data: data.local_file.user_data: open /tmp/kube-workers/worker-bootstrap.sh: no such file or directory

Can't we just upload that script somewhere else?
Thanks!

[edtoon]

The script isn't "uploaded" to that path - it's read from that path on the machine that Terraform is running on, I assume as a quick workaround to Terraform's weird user_data behavior in the past. That will have to be handled differently in order to avoid this error.

(See hashicorp/terraform#145 for more about the "weirdness", it should be fixed now...)

[lander2k2]

@edtoon is correct. The file /tmp/kube-workers/worker-bootstrap.sh is created on the local machine during cluster installation. kube-cluster.sh generates that file when masters are created. Then it is referenced when the worker ASG is created.

@simox-83 Have you tried creating an empty file at /tmp/kube-workers/worker-bootstrap.sh? Since you're destroying, it shouldn't matter that it doesn't have the correct content.

[simox-83]

@lander2k2 Thanks for your suggestion. I tried to do what you suggested and it works. Thanks for the clarifications too @edtoon