session affinity no longer set on backend services

[thecodeassassin]

Terraform Version

v0.10.2

Affected Resource(s)

google_compute_backend_service

If this issue appears to affect multiple resources, it may be an issue with Terraform's core, so please mention this.

Terraform Configuration Files

esource "google_compute_backend_service" "legacy" {

    name = "legacy-backend-${var.environment}"
    description = "Legacy backend service"
    port_name = "http"
    protocol = "HTTP"
    timeout_sec = 120
    session_affinity = "GENERATED_COOKIE"

    depends_on = ["google_compute_instance_group_manager.legacy"]

    backend {
        group = "${google_compute_instance_group_manager.legacy.instance_group}"
    }

    backend {
        group = "${google_compute_instance_group_manager.legacy-fallback.instance_group}"
    }

    health_checks = ["${google_compute_http_health_check.legacy.self_link}"]
}

Debug Output

https://gist.github.com/thecodeassassin/ebf4df4769fe38d6febf1529e014b2ee

Expected Behavior

Should create a backend service that has the session affinity set.

Actual Behavior

What actually happened?

Steps to Reproduce

1. terraform apply

Important Factoids

I think it broke after upgrading from 0.9.x to 0.10.x.

[thecodeassassin]

It seems to be a UI problem on Google Cloud's side. issue can be closed i guess.

[thecodeassassin]

I was wrong, it seems that it's not setting the session affinity correctly when updating.

  ~ google_compute_backend_service.legacy
      backend.#:                                 "1" => "2"
      backend.1795308435.balancing_mode:         "UTILIZATION" => "UTILIZATION"
      backend.1795308435.capacity_scaler:        "1" => "1"
      backend.1795308435.description:            "" => ""
      backend.1795308435.group:                  "https://www.googleapis.com/compute/v1/projects/REDACTED/zones/europe-west1-d/instanceGroups/legacy-group-manager-pele" => "https://www.googleapis.com/compute/v1/projects/REDACTED/zones/europe-west1-d/instanceGroups/legacy-group-manager-pele"
      backend.1795308435.max_rate:               "0" => "0"
      backend.1795308435.max_rate_per_instance:  "0" => "0"
      backend.1795308435.max_utilization:        "0.8" => "0.8"
      backend.~3824898581.balancing_mode:        "" => "UTILIZATION"
      backend.~3824898581.capacity_scaler:       "" => "1"
      backend.~3824898581.description:           "" => ""
      backend.~3824898581.group:                 "" => "${google_compute_instance_group_manager.legacy-fallback.instance_group}"
      backend.~3824898581.max_rate:              "" => ""
      backend.~3824898581.max_rate_per_instance: "" => ""
      backend.~3824898581.max_utilization:       "" => "0.8"

[selmanj]

Was able to reproduce. Whenever a backend_service is updated, session_affinity is reset to NONE (and a subsequent terraform apply notices this and corrects it).

A quick glance at the code seems to suggest the problem is that the update code is written to only set options that have changed, which then results in session_affinity having a default value.

[selmanj]

While attempting to fix this, I uncovered an inconsistency with how connection draining default values are set in the GCP API. Filed a bug upstream.

[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!