node_count is required

[minac]

Without node_count, autoscaling will never kick in because e.g. 0 is not between 2 and 5.

[danawillow]

Hi @minac, I'm not convinced this is the correct approach. This variable essentially says "node count should be X, hard stop". If autoscaling is also defined, then it won't work: any time terraform is re-run, it'll try to set the node count back to whatever was defined in the configuration. Instead, I'd recommend using initial_node_count to create the node pool with the specified number of nodes, while not enforcing that it remains the same size it was when it was originally created.

Because either initial_node_count or node_count can be specified, I don't feel comfortable changing the docs to say either are required. However, I'm happy to accept a PR that clarifies when to use one vs the other, especially in relation to autoscaling.

[minac]

Hi @danawillow! Using initial_node_count seems like a better alternative. I just wish it was required to avoid the issue described above. If one does not use the field and sets the min of an autoscaling group to 2 and max to 4, the cluster will never create correctly, because it will never reach the minimum.
This was my experience. As you say this may not be a documentation issue, but a module issue, as an autoscaling group should guarantee that the number of nodes is between a min and a max.

[danawillow]

Neither field should be required though, because they both have their use cases. initial_node_count works better when you want to use autoscaling, and node_count works better when you don't. We also used to check that node pools had a count > 0, but people wanted the ability to scale their cluster down (#752).

I think the correct approach is not to say that the field is required, but instead to add a note to the autoscaling docs to say that it only works if the initial node count starts in that range.

[minac]

Sounds good

[rileykarson]

Based on https://github.com/kubernetes/autoscaler/blob/master/cluster-autoscaler/FAQ.md#my-cluster-is-below-minimum--above-maximum-number-of-nodes-but-ca-did-not-fix-that-why, I don't think that your initial_node_count needs to be within autoscaler ranges, so I'm going to close this out. If I'm wrong, please correct me though!

[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 👉 hashibot-feedback@hashicorp.com. Thanks!