kops_cluster_updater should have more verbose logs

[shapirus]

module.cluster.kops_cluster_updater.updater: Still creating... [19m40s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [19m50s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [20m0s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [20m10s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [20m20s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [20m30s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [20m40s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [20m50s elapsed]
module.cluster.kops_cluster_updater.updater: Still creating... [21m0s elapsed]

At this point, all pods are up and running, all containers are healthy, all nodes are healthy, but kops_cluster_updater is still not happy. Why? What is it waiting for? This is where the ability to make it show the actual kops logs would be very helpful.

[sl1pm4t]

I don't think Terraform allows provider plugins to print any useful info to the console, unless TF is run in debug mode:

TF_LOG=DEBUG terraform apply

Perhaps a compromise is for the kops provider to intercept the kops output and write it to a log file?!

[shapirus]

There's probably a way, but it may work only on errors, I'm not sure. For example, with the aws provider, when there's an error creating something in the cloud and AWS returns error 400 (e.g., resource already exists), there is an actual error message printed on the console. Not sure if it's possible to do that outside of the error context.