Import Container Nodepool doesn't work with 4 parts as documented

[jamielennox]

The import documentation says we can import nodepools like:

$ terraform import google_container_node_pool.mainpool my-gcp-project/us-east1-a/my-cluster/main-pool

$ terraform import google_container_node_pool.mainpool us-east1-a/my-cluster/main-pool

I'm using the 4 part version and the import step completes successfully. Terraform then runs the exists check and in my case the d.Id() is:

<project>/<zone>/<cluster>/<nodepool-name>

which when run through:

name := getNodePoolName(d.Id())

func getNodePoolName(id string) string {
	// name can be specified with name, name_prefix, or neither, so read it from the id.
	return strings.Split(id, "/")[2]
}

means that it's returning my cluster name instead of the nodepool name.

I can then see it try and check the existance of the nodepool:

GET /v1beta1/projects/<project>/locations/<zone>/clusters/<cluster>/nodePools/<cluster>?alt=json

which obviously fails and prevents my import from working. In my case I don't have a provider project configured so I can't use the 3 param form.

google_container_node_pool.nodepool: project: required field is not set

I'll investigate a little further, but somewhere along the line the ID you end up with from import is wrong - or the getNodePoolName function needs to be updated to support it. I'm not sure which direction you're going in here.

I'm running master 65669fc.

[jamielennox]

This quick patch is getting me past my problem. I don't know if it's actually a good idea or not.

From 0566c6abb98399a4547c0466a1fc50c427730859 Mon Sep 17 00:00:00 2001
From: Jamie Lennox <[email protected]>
Date: Tue, 11 Sep 2018 12:33:24 +1000
Subject: [PATCH] Support nodepool container name having 3 or 4 segments

Using nodepool import lets us use a 3 or 4 segment version. When you use
the 4 segment version the ID that gets created contains more information
than a regular ID and so we can't use the same portion of the path to
identify name.

Fix it temporarily to unblock myself until upstream confirms a
direction.
---
 google/resource_container_node_pool.go | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/google/resource_container_node_pool.go b/google/resource_container_node_pool.go
index 17c4b482..d0307c9d 100644
--- a/google/resource_container_node_pool.go
+++ b/google/resource_container_node_pool.go
@@ -724,5 +724,15 @@ func nodePoolUpdate(d *schema.ResourceData, meta interface{}, nodePoolInfo *Node

 func getNodePoolName(id string) string {
 	// name can be specified with name, name_prefix, or neither, so read it from the id.
-	return strings.Split(id, "/")[2]
+	parts := strings.Split(id, "/")
+
+	switch len(parts) {
+	case 3:
+		return parts[2]
+	case 4:
+		return parts[3]
+	default:
+		log.Printf("[ERR] Unexpected nodepool name segments: %s", id)
+		return ""
+	}
 }
--
2.17.1

[jamielennox]

Originally created in #1653 notifying @dtan4 and @danawillow based on that. (Hope that's ok)

[jamielennox]

Actually looking at this a bit later, almost certainly what we want is a:

d.SetId(strings.Join(parts[1:], "/"))

in case 4: of the importer to drop the project from the start and use the standard id.

[danawillow]

I'm going to go ahead and close this as a duplicate of #1846, but thanks for the report! Just a note that if you do have a solution in mind, we'd be happy to look at a PR for it :)

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 [email protected]. Thanks!