commands that exceed timeouts are considered a success

[emilioziniades]

Hi there, thanks for the cool provider. Am playing around with it, and ran into some weirdness. Consider this example main.tf:

terraform {
  required_providers {
    ssh = {
      source  = "loafoe/ssh"
      version = "2.6.0"
    }
  }
}

resource "ssh_resource" "long_running_ssh_command" {
  host        = "15.26.187.21"
  user        = "root"
  password    = "mysecurepassword"
  timeout     = "10s"
  retry_delay = "1s"

  commands = [
    "sleep 300 && echo finished > /tmp/test.txt",
  ]
}

As you can see, the timeout limit is 10 seconds, but the command will actually take 300 seconds (5 minutes) to complete.

When I do terraform apply, this is the output:

ssh_resource.long_running_ssh_command: Creating...
ssh_resource.long_running_ssh_command: Still creating... [10s elapsed]
ssh_resource.long_running_ssh_command: Creation complete after 11s [id=3395189971859253104]

Expected behaviour: resource creation fails, since it didn't complete within the timout.
Actual behaviour: resource creation succeeds, despite the command not completing before the timeout.

My short term hack is to use a gigantic timeout of "120m".

[loafoe]

Thanks for reporting this, it is indeed not what the resource should do. I'll think about a possible solution for this.