hashicorp/terraform · critical

missing provider schema

Error message

missing provider schema

What it means

Thrown by the Terraform plugin v6 gRPC client (GRPCProvider.GetProviderSchema) when GetProviderSchema succeeded without diagnostics but the response `Provider` field is nil. Identical semantics to the v5 'missing provider schema' error: the provider must declare a top-level provider schema, and a nil one is a provider defect. The v6 path then continues to fetch resource identity schemas.

Solutions

1. Upgrade or pin a compatible v6 provider version; run `terraform init -upgrade`.
2. Clear the plugin cache (.terraform/providers) and re-init.
3. If you maintain the provider, ensure the framework provider schema is declared and returned.
4. Check provider logs for a crash during schema serving.

Example fix

# before: v6 provider returns nil schema
# provider-side: declare schema in terraform-plugin-framework provider

# user-side:
$ terraform init -upgrade

# shell: verify the v6 provider serves a schema
$ terraform providers schema -json | jq '.provider_schemas'

# bash: clear cache and re-init on schema failure
if ! terraform init 2>err.log; then
  if grep -q 'missing provider schema' err.log; then
    rm -rf .terraform/providers && terraform init -upgrade
  else cat err.log; exit 1; fi
fi

When it happens

Trigger: A v6-protocol provider's GetProviderSchema handler returned Provider == nil with no error. Protocol-version mismatch, a mis-built v6 provider, or a provider crash yielding an empty response.

Common situations: Provider built against terraform-plugin-go/terraform-plugin-framework v6 that omits the provider schema. Corrupted plugin cache. Version skew between core and provider.

Related errors

• action schema not found for action
• identity schema not found for type
• list resource schema not found for type
• list resource schema not found for type
• missing provider schema

AI-assisted analysis of hashicorp/terraform@d32a084675 (2026-08-11). Data as JSON: /api/errors/b99b44cef9f84437. Report an issue: GitHub.

Appendix: source

Thrown at internal/plugin6/grpc_provider.go:146

	// Note: this option is marked as EXPERIMENTAL in the grpc API. We keep
	// this for compatibility, but recent providers all set the max message
	// size much higher on the server side, which is the supported method for
	// determining payload size.
	const maxRecvSize = 64 << 20
	protoResp, err := p.client.GetProviderSchema(p.ctx, new(proto6.GetProviderSchema_Request), grpc.MaxRecvMsgSizeCallOption{MaxRecvMsgSize: maxRecvSize})
	if err != nil {
		resp.Diagnostics = resp.Diagnostics.Append(grpcErr(err))
		return resp
	}

	resp.Diagnostics = resp.Diagnostics.Append(convert.ProtoToDiagnostics(protoResp.Diagnostics))

	if resp.Diagnostics.HasErrors() {
		return resp
	}

	if protoResp.Provider == nil {
		resp.Diagnostics = resp.Diagnostics.Append(errors.New("missing provider schema"))
		return resp
	}

	identResp, err := p.client.GetResourceIdentitySchemas(p.ctx, new(proto6.GetResourceIdentitySchemas_Request))
	if err != nil {
		if status.Code(err) == codes.Unimplemented {
			// We don't treat this as an error if older providers don't implement this method,
			// so we create an empty map for identity schemas
			identResp = &proto6.GetResourceIdentitySchemas_Response{
				IdentitySchemas: map[string]*proto6.ResourceIdentitySchema{},
			}
		} else {
			resp.Diagnostics = resp.Diagnostics.Append(grpcErr(err))
			return resp
		}
	}

	resp.Diagnostics = resp.Diagnostics.Append(convert.ProtoToDiagnostics(identResp.Diagnostics))

View on GitHub (pinned to d32a084675)