error: /bin/sh: 2: gcloud not found about terraform-google-gcloud HOT 10 CLOSED

The way it is now just setting destroy_cmd_body to something like "" would still break because it'll still attempt to just run gcloud at destroy time before the executable is there.

If you also set destroy_cmd_entrypoint = "" that should at least let terraform destroy finish. Obviously the iam-service-account would still be disabled for you since the enable on destroy wouldn't run.

from terraform-google-gcloud.

What platform are you using? And can you run terraform version?

from terraform-google-gcloud.

@chiokejjones could you try explicitly setting the version for this module to v0.2.0 and let us know if it was successful? There was a big change in v0.3.0 that shifted things around quite a bit.

from terraform-google-gcloud.

Hey fellas.

@morgante I'm using Terraform Enterprise running Terraform v0.12.18.

@milesmatthias I receive the same error when explicitly setting v0.2.0 or v0.1.0.

from terraform-google-gcloud.

@chiokejjones How about with v0.3.0? @taylorludwig Could your changes have caused this? Maybe we can test on Terraform Cloud.

from terraform-google-gcloud.

@morgante the command looks good. I see the PATH updated as expected with the new v0.3.0 changes.

I did notice in the output above the command is gcloud iam service-accounts enable projects/... but enable is the destroy_cmd_body

So it would make sense that it fails on destroy - the decompress and upgrade resources wouldn't have run on a fresh new vm/container (as terraform cloud would do) just to do a destroy so gcloud command wouldn't be there.

I believe it would have failed prior to v0.3.0 for the same reason.

@chiokejjones could you confirm if its failing on destroy and not apply?

I can setup a terraform cloud project to replicate

from terraform-google-gcloud.

I was able to replicate the same error on terraform cloud during a destroy.

apply worked as intended. I'll see if I can find a way to force the provision steps to happen on destroy for the setup gcloud resources.

from terraform-google-gcloud.

@morgante I received the same error on all three versions.

@taylorludwig It seems to force the destroy of the null_resources regardless if I'm making a change to the service account or not. I can try removing the destroy_cmd_body from the module logic.

from terraform-google-gcloud.

I was able to get create and destroy working with my PR here #11

@morgante seems a bit hacky and a lot of duplicate null_resource local-exec resources. But it does make sure that gcloud is decompressed and setup in the proper order during a destroy.

Output of my terraform cloud runs

apply.txt

destroy.txt

from terraform-google-gcloud.

The way it is now just setting destroy_cmd_body to something like "" would still break because it'll still attempt to just run gcloud at destroy time before the executable is there.

If you also set destroy_cmd_entrypoint = "" that should at least let terraform destroy finish. Obviously the iam-service-account would still be disabled for you since the enable on destroy wouldn't run.

@taylorludwig Thanks for the suggestion. Nulling those two parameters allowed the destroy to finish.

from terraform-google-gcloud.