clickhouse_service_scheduled_scaling: "inconsistent result after apply" for entries with idle_scaling = false

[joecanueldd]

Description

clickhouse_service_scheduled_scaling: "inconsistent result after apply" for entries with idle_scaling = false

Description

clickhouse_service_scheduled_scaling cannot be applied for any entry with idle_scaling = false. terraform apply fails with "Provider produced inconsistent result after apply … does not correlate with any element in actual". Since disabled idle scaling is the common case for fixed-size scheduled windows, the resource is effectively unusable for those entries.

Cloud provider

AWS

Service tier

Not applicable

Affected resource(s)

clickhouse_service_scheduled_scaling

Provider version

3.18.0 (resource introduced in 3.17.0; affects all versions that ship it)

Terraform version

1.12.2

Terraform configuration

resource "clickhouse_service_scheduled_scaling" "reader" {
  service_id = clickhouse_service.reader.id

  entries = [
    {
      name                  = "Business Hours"
      weekdays              = [0, 1, 2, 3, 4, 5, 6]
      start_hour_utc        = 5
      end_hour_utc          = 19
      min_replicas          = 10
      max_replicas          = 10
      min_replica_memory_gb = 236
      max_replica_memory_gb = 236
      idle_scaling          = false
    },
  ]
}

Expected behavior

The schedule is created and the apply completes cleanly, with the entry stored as written (idle_scaling = false).

Actual behavior

The API call succeeds, but the apply then aborts with a provider-consistency error:

Error: Provider produced inconsistent result after apply

When applying changes to clickhouse_service_scheduled_scaling.reader, provider
"provider[\"registry.terraform.io/clickhouse/clickhouse\"]" produced an unexpected
new value: .entries: planned set element cty.ObjectVal(map[string]cty.Value{
"end_hour_utc":cty.NumberIntVal(19), "idle_scaling":cty.False,
"idle_timeout_minutes":cty.NumberIntVal(5), "max_replica_memory_gb":cty.NumberIntVal(236),
"max_replicas":cty.NumberIntVal(10), "min_replica_memory_gb":cty.NumberIntVal(236),
"min_replicas":cty.NumberIntVal(10), "name":cty.StringVal("Business Hours"),
"start_hour_utc":cty.NumberIntVal(5), "weekdays":cty.SetVal([]cty.Value{...})})
does not correlate with any element in actual.

This is a bug in the provider, which should be reported in the provider's own
issue tracker.

The same happens whether idle_timeout_minutes is set or omitted. Retrying loops: a subsequent plan shows -/+ replacement because the failed apply leaves the resource tainted in state.

Additional context

Root cause (from reading the source at v3.18.0):

entries is a SetNestedAttribute whose nested fields are Optional + Computed. Two problems compound:

1. Terraform correlates set elements by a hash of their contents. Any nested field left unset is (known after apply) at plan time; a set element containing an unknown can't be hash-matched to a returned element → "does not correlate."
2. For a non-idle entry the server does not echo the idle fields, and apiEntryToModel maps them back with nil → null (boolPtrToValue / int64PtrToValue in pkg/resource/service_scheduled_scaling.go). So the applied element differs from the planned one (idle_scaling: false → null) even when every field is specified.

The existing round-trip test (TestRoundTrip_NoServerNormalization) doesn't catch this because it only exercises idle_scaling = true, where the server preserves the idle fields.

Likely fix: model entries as a ListNestedAttribute (lists correlate by index and tolerate Computed fields resolving after apply) and map a nil idle_scaling to an explicit false.

Cloud provider

AWS

Service tier

Production

Affected resource(s)

No response

Provider version

3.18.0

Terraform version

1.12.2

Terraform configuration

resource "clickhouse_service_scheduled_scaling" "reader" {
  service_id = clickhouse_service.reader.id

  entries = [
    {
      name                  = "Business Hours"
      weekdays              = [0, 1, 2, 3, 4, 5, 6]
      start_hour_utc        = 5
      end_hour_utc          = 19
      min_replicas          = 10
      max_replicas          = 10
      min_replica_memory_gb = 236
      max_replica_memory_gb = 236
      idle_scaling          = false
    },
  ]
}

Expected behavior

Created correctly.

Actual behavior

Error: Provider produced inconsistent result after apply
When applying changes to clickhouse_service_scheduled_scaling.reader, provider
"provider["registry.terraform.io/clickhouse/clickhouse"]" produced an unexpected
new value: .entries: planned set element cty.ObjectVal(map[string]cty.Value{
"end_hour_utc":cty.NumberIntVal(19), "idle_scaling":cty.False,
"idle_timeout_minutes":cty.NumberIntVal(5), "max_replica_memory_gb":cty.NumberIntVal(236),
"max_replicas":cty.NumberIntVal(10), "min_replica_memory_gb":cty.NumberIntVal(236),
"min_replicas":cty.NumberIntVal(10), "name":cty.StringVal("Business Hours"),
"start_hour_utc":cty.NumberIntVal(5), "weekdays":cty.SetVal([]cty.Value{...})})
does not correlate with any element in actual.
This is a bug in the provider, which should be reported in the provider's own
issue tracker.

Additional context

No response

[serdec]

@joecanueldd this issue was solved server side, you shouldn't experience it anymore. Let me know if that is not the case