fix: grpc_service_config timeout settings (#3200)

* fix: set max & initial timeouts to total timeout

When using grpc_service_config the max & initial rpc timeout
should be the same as the total timeout. This essentially
renders the hedging logic unused because the initial timeout
is the limit. This means one RPC is made rather than several
smaller ones that are backed off until the total is reached.

* switch showcase to gapic v2 & grpc_service_config
* retry mapping: fix NPE when a Service is not present in protos
* add blockTimeout showcase test
* add valid block showcase test
* update gapic-showcase image for java to v0.10.1
* lengthen delay on flaky showcase java streaming tests

.circleci/config.yml

@@ -316,7 +316,7 @@ jobs:
       - image: circleci/openjdk:8-jdk
     working_directory: /tmp/workspace
     environment:
-      SHOWCASE_VERSION: 0.7.0
+      SHOWCASE_VERSION: 0.10.1
     steps:
       - checkout:
           path: gapic-generator
@@ -326,13 +326,18 @@ jobs:
             git clone --single-branch https://github.com/googleapis/googleapis.git googleapis
       - run:
           name: Setup showcase protos.
+          # Remove the other showcase protos, testing only uses echo.proto
           command: |
             mkdir showcase-protos
             curl -sL https://github.com/googleapis/gapic-showcase/releases/download/v${SHOWCASE_VERSION}/gapic-showcase-${SHOWCASE_VERSION}-protos.tar.gz | tar -C showcase-protos -xz
+            curl -sL https://github.com/googleapis/gapic-showcase/releases/download/v${SHOWCASE_VERSION}/showcase_grpc_service_config.json > showcase-protos/google/showcase/v1beta1/showcase_grpc_service_config.json
             TEST_SRC="gapic-generator/src/test/java/com/google/api/codegen/testsrc/showcase"
             cp ${TEST_SRC}/showcase.yaml showcase-protos/google/showcase/
             cp ${TEST_SRC}/showcase_gapic.yaml showcase-protos/google/showcase/v1beta1/
             cp ${TEST_SRC}/artman_showcase.yaml showcase-protos/google/showcase/
+            rm showcase-protos/google/showcase/v1beta1/identity.proto
+            rm showcase-protos/google/showcase/v1beta1/messaging.proto
+            rm showcase-protos/google/showcase/v1beta1/testing.proto
 
       - run:
           name: Build and install local toolkit
@@ -528,7 +533,7 @@ jobs:
       <<: *anchor_artman_vars
     docker:
       - image: circleci/openjdk:8-jdk
-      - image: gcr.io/gapic-images/gapic-showcase:0.2.0
+      - image: gcr.io/gapic-images/gapic-showcase:0.10.1
     steps:
       - attach_workspace:
           at: workspace

showcase/java/src/test/java/com/google/api/showcase/ShowcaseTest.java

@@ -25,8 +25,11 @@
 import com.google.api.gax.rpc.ClientStream;
 import com.google.api.gax.rpc.ServerStream;
 import com.google.api.gax.rpc.StreamController;
+import com.google.protobuf.Duration;
 import com.google.rpc.Code;
 import com.google.rpc.Status;
+import com.google.showcase.v1beta1.BlockRequest;
+import com.google.showcase.v1beta1.BlockResponse;
 import com.google.showcase.v1beta1.EchoClient;
 import com.google.showcase.v1beta1.EchoRequest;
 import com.google.showcase.v1beta1.EchoResponse;
@@ -172,7 +175,7 @@ public void onCompleted() {
     }
     requestStream.onCompleted();
 
-    latch.await(2, TimeUnit.SECONDS);
+    latch.await(7, TimeUnit.SECONDS);
 
     assertThat(collections).containsExactly("a b c done");
   }
@@ -224,8 +227,35 @@ public void onComplete() {
               }
             });
 
-    latch.await(2, TimeUnit.SECONDS);
+    latch.await(7, TimeUnit.SECONDS);
 
     assertThat(responses).containsExactlyElementsIn(inputs).inOrder();
   }
+
+  @Test(expected = StatusRuntimeException.class)
+  public void blockTimeout() {
+    try {
+      client.block(
+          BlockRequest.newBuilder()
+              // Set a longer timeout than the 5 seconds specified in the grpc_service_config.
+              .setResponseDelay(Duration.newBuilder().setSeconds(10L).build())
+              .build());
+    } catch (Exception e) {
+      assertThat(e.getCause()).isInstanceOf(StatusRuntimeException.class);
+      StatusRuntimeException error = (StatusRuntimeException) e.getCause();
+      assertThat(error.getStatus().getCode().value()).isEqualTo(Code.DEADLINE_EXCEEDED_VALUE);
+      throw error;
+    }
+  }
+
+  @Test
+  public void block() {
+    BlockResponse result =
+        client.block(
+            BlockRequest.newBuilder()
+                .setResponseDelay(Duration.newBuilder().setSeconds(2L).build())
+                .setSuccess(BlockResponse.newBuilder().setContent("Hello, World!").build())
+                .build());
+    assertThat(result.getContent()).isEqualTo("Hello, World!");
+  }
 }

src/main/java/com/google/api/codegen/config/GrpcGapicRetryMapping.java

@@ -139,9 +139,13 @@ private static void applyRetryPolicyToName(
       return;
     }
 
-    // apply the RetryPolicy to all methods in the service interface that don't already have an
-    // entry
+    // Apply the RetryPolicy to all methods in the service interface that don't already have an
+    // entry. If the Service is not present in the protos, skip it.
     Interface interProto = protoInterfaces.get(service);
+    if (interProto == null) {
+      return;
+    }
+
     for (Method methodProto : interProto.getMethods()) {
       String fullName = methodProto.getFullName();
       methodCodesMap.putIfAbsent(fullName, codesName);
@@ -155,7 +159,10 @@ private static RetryParamsDefinitionProto.Builder retryPolicyToParamsBuilder(
         .setMaxRetryDelayMillis(Durations.toMillis(retryPolicy.getMaxBackoff()))
         .setInitialRetryDelayMillis(Durations.toMillis(retryPolicy.getInitialBackoff()))
         .setRetryDelayMultiplier(convertFloatToDouble(retryPolicy.getBackoffMultiplier()))
+        .setRpcTimeoutMultiplier(1.0)
         .setTotalTimeoutMillis(timeout)
+        .setMaxRpcTimeoutMillis(timeout)
+        .setInitialRpcTimeoutMillis(timeout)
         .setName(policyName);
   }
 

src/test/java/com/google/api/codegen/protoannotations/testdata/java_library_with_grpc_service_config.baseline

@@ -14487,13 +14487,20 @@ public class LibraryServiceStubSettings extends StubSettings<LibraryServiceStubS
           .setInitialRetryDelay(Duration.ofMillis(200L))
           .setRetryDelayMultiplier(1.3)
           .setMaxRetryDelay(Duration.ofMillis(35000L))
+          .setInitialRpcTimeout(Duration.ofMillis(60000L))
+          .setRpcTimeoutMultiplier(1.0)
+          .setMaxRpcTimeout(Duration.ofMillis(60000L))
           .setTotalTimeout(Duration.ofMillis(60000L))
           .build();
       definitions.put("retry_policy_1_params", settings);
       settings = RetrySettings.newBuilder()
+          .setRpcTimeoutMultiplier(1.0)
           .build();
       definitions.put("no_retry_params", settings);
       settings = RetrySettings.newBuilder()
+          .setInitialRpcTimeout(Duration.ofMillis(60000L))
+          .setRpcTimeoutMultiplier(1.0)
+          .setMaxRpcTimeout(Duration.ofMillis(60000L))
           .setTotalTimeout(Duration.ofMillis(60000L))
           .build();
       definitions.put("no_retry_1_params", settings);
@@ -15491,9 +15498,13 @@ public class MyProtoStubSettings extends StubSettings<MyProtoStubSettings> {
       ImmutableMap.Builder<String, RetrySettings> definitions = ImmutableMap.builder();
       RetrySettings settings = null;
       settings = RetrySettings.newBuilder()
+          .setRpcTimeoutMultiplier(1.0)
           .build();
       definitions.put("no_retry_params", settings);
       settings = RetrySettings.newBuilder()
+          .setInitialRpcTimeout(Duration.ofMillis(60000L))
+          .setRpcTimeoutMultiplier(1.0)
+          .setMaxRpcTimeout(Duration.ofMillis(60000L))
           .setTotalTimeout(Duration.ofMillis(60000L))
           .build();
       definitions.put("no_retry_2_params", settings);

src/test/java/com/google/api/codegen/protoannotations/testdata/php_library_with_grpc_service_config.baseline

@@ -6089,27 +6089,27 @@ class MyProtoClient extends MyProtoGapicClient
           "initial_retry_delay_millis": 200,
           "retry_delay_multiplier": 1.3,
           "max_retry_delay_millis": 35000,
-          "initial_rpc_timeout_millis": 0,
-          "rpc_timeout_multiplier": 0.0,
-          "max_rpc_timeout_millis": 0,
+          "initial_rpc_timeout_millis": 60000,
+          "rpc_timeout_multiplier": 1.0,
+          "max_rpc_timeout_millis": 60000,
           "total_timeout_millis": 60000
         },
         "no_retry_params": {
           "initial_retry_delay_millis": 0,
           "retry_delay_multiplier": 0.0,
           "max_retry_delay_millis": 0,
           "initial_rpc_timeout_millis": 0,
-          "rpc_timeout_multiplier": 0.0,
+          "rpc_timeout_multiplier": 1.0,
           "max_rpc_timeout_millis": 0,
           "total_timeout_millis": 0
         },
         "no_retry_1_params": {
           "initial_retry_delay_millis": 0,
           "retry_delay_multiplier": 0.0,
           "max_retry_delay_millis": 0,
-          "initial_rpc_timeout_millis": 0,
-          "rpc_timeout_multiplier": 0.0,
-          "max_rpc_timeout_millis": 0,
+          "initial_rpc_timeout_millis": 60000,
+          "rpc_timeout_multiplier": 1.0,
+          "max_rpc_timeout_millis": 60000,
           "total_timeout_millis": 60000
         }
       },
@@ -6740,17 +6740,17 @@ return [
           "retry_delay_multiplier": 0.0,
           "max_retry_delay_millis": 0,
           "initial_rpc_timeout_millis": 0,
-          "rpc_timeout_multiplier": 0.0,
+          "rpc_timeout_multiplier": 1.0,
           "max_rpc_timeout_millis": 0,
           "total_timeout_millis": 0
         },
         "no_retry_2_params": {
           "initial_retry_delay_millis": 0,
           "retry_delay_multiplier": 0.0,
           "max_retry_delay_millis": 0,
-          "initial_rpc_timeout_millis": 0,
-          "rpc_timeout_multiplier": 0.0,
-          "max_rpc_timeout_millis": 0,
+          "initial_rpc_timeout_millis": 60000,
+          "rpc_timeout_multiplier": 1.0,
+          "max_rpc_timeout_millis": 60000,
           "total_timeout_millis": 60000
         }
       },

src/test/java/com/google/api/codegen/testsrc/showcase/artman_showcase.yaml

@@ -4,6 +4,8 @@ common:
   organization_name: google
   service_yaml: showcase.yaml
   gapic_yaml: v1beta1/showcase_gapic.yaml
+  grpc_service_config: v1beta1/showcase_grpc_service_config.json
+  proto_package: google.showcase.v1beta1
   src_proto_paths:
   - v1beta1
   proto_deps:

src/test/java/com/google/api/codegen/testsrc/showcase/showcase_gapic.yaml

@@ -3,7 +3,7 @@
 # retry_codes_name, required_fields, flattening, and timeout properties cannot
 # be precisely decided by the tooling and may require some configuration.
 type: com.google.api.codegen.ConfigProto
-config_schema_version: 1.0.0
+config_schema_version: 2.0.0
 # The settings of generated code in a specific language.
 language_settings:
   java:
@@ -20,139 +20,3 @@ language_settings:
     package_name: Google\Showcase\V1beta1
   nodejs:
     package_name: showcase.v1beta1
-# The configuration for the license header to put on generated files.
-license_header:
-  # The file containing the copyright line(s).
-  copyright_file: copyright-google.txt
-  # The file containing the raw license header without any copyright line(s).
-  license_file: license-header-apache-2.0.txt
-# A list of API interface configurations.
-interfaces:
-  # The fully qualified name of the API interface.
-- name: google.showcase.v1beta1.Echo
-  # A list of resource collection configurations.
-  # Consists of a name_pattern and an entity_name.
-  # The name_pattern is a pattern to describe the names of the resources of this
-  # collection, using the platform's conventions for URI patterns. A generator
-  # may use this to generate methods to compose and decompose such names. The
-  # pattern should use named placeholders as in `shelves/{shelf}/books/{book}`;
-  # those will be taken as hints for the parameter names of the generated
-  # methods. If empty, no name methods are generated.
-  # The entity_name is the name to be used as a basis for generated methods and
-  # classes.
-  # Definition for retryable codes.
-  retry_codes_def:
-  - name: idempotent
-    retry_codes:
-    - UNAVAILABLE
-    - DEADLINE_EXCEEDED
-  - name: non_idempotent
-    retry_codes: []
-  # Definition for retry/backoff parameters.
-  retry_params_def:
-  - name: default
-    initial_retry_delay_millis: 100
-    retry_delay_multiplier: 1.3
-    max_retry_delay_millis: 60000
-    initial_rpc_timeout_millis: 20000
-    rpc_timeout_multiplier: 1
-    max_rpc_timeout_millis: 20000
-    total_timeout_millis: 600000
-  # A list of method configurations.
-  # Common properties:
-  #
-  #   name - The simple name of the method.
-  #
-  #   flattening - Specifies the configuration for parameter flattening.
-  #   Describes the parameter groups for which a generator should produce method
-  #   overloads which allow a client to directly pass request message fields as
-  #   method parameters. This information may or may not be used, depending on
-  #   the target language.
-  #   Consists of groups, which each represent a list of parameters to be
-  #   flattened. Each parameter listed must be a field of the request message.
-  #
-  #   required_fields - Fields that are always required for a request to be
-  #   valid.
-  #
-  #   resource_name_treatment - An enum that specifies how to treat the resource
-  #   name formats defined in the field_name_patterns and
-  #   response_field_name_patterns fields.
-  #   UNSET: default value
-  #   NONE: the collection configs will not be used by the generated code.
-  #   VALIDATE: string fields will be validated by the client against the
-  #   specified resource name formats.
-  #   STATIC_TYPES: the client will use generated types for resource names.
-  #
-  #   page_streaming - Specifies the configuration for paging.
-  #   Describes information for generating a method which transforms a paging
-  #   list RPC into a stream of resources.
-  #   Consists of a request and a response.
-  #   The request specifies request information of the list method. It defines
-  #   which fields match the paging pattern in the request. The request consists
-  #   of a page_size_field and a token_field. The page_size_field is the name of
-  #   the optional field specifying the maximum number of elements to be
-  #   returned in the response. The token_field is the name of the field in the
-  #   request containing the page token.
-  #   The response specifies response information of the list method. It defines
-  #   which fields match the paging pattern in the response. The response
-  #   consists of a token_field and a resources_field. The token_field is the
-  #   name of the field in the response containing the next page token. The
-  #   resources_field is the name of the field in the response containing the
-  #   list of resources belonging to the page.
-  #
-  #   retry_codes_name - Specifies the configuration for retryable codes. The
-  #   name must be defined in interfaces.retry_codes_def.
-  #
-  #   retry_params_name - Specifies the configuration for retry/backoff
-  #   parameters. The name must be defined in interfaces.retry_params_def.
-  #
-  #   field_name_patterns - Maps the field name of the request type to
-  #   entity_name of interfaces.collections.
-  #   Specifies the string pattern that the field must follow.
-  #
-  #   timeout_millis - Specifies the default timeout for a non-retrying call. If
-  #   the call is retrying, refer to retry_params_name instead.
-  methods:
-  - name: Echo
-    retry_codes_name: idempotent
-    retry_params_name: default
-    timeout_millis: 60000
-  - name: Expand
-    flattening:
-      groups:
-      - parameters:
-        - content
-        - error
-    retry_codes_name: idempotent
-    retry_params_name: default
-    timeout_millis: 60000
-  - name: PagedExpand
-    page_streaming:
-      request:
-        page_size_field: page_size
-        token_field: page_token
-      response:
-        token_field: next_page_token
-        resources_field: responses
-    retry_codes_name: idempotent
-    retry_params_name: default
-    timeout_millis: 60000
-  - name: Collect
-    retry_codes_name: non_idempotent
-    retry_params_name: default
-    timeout_millis: 60000
-  - name: Chat
-    retry_codes_name: idempotent
-    retry_params_name: default
-    timeout_millis: 60000
-  - name: Wait
-    retry_codes_name: idempotent
-    retry_params_name: default
-    timeout_millis: 60000
-    long_running:
-      return_type: google.showcase.v1beta1.WaitResponse
-      metadata_type: google.showcase.v1beta1.WaitMetadata
-      initial_poll_delay_millis: 20000
-      poll_delay_multiplier: 1.5
-      max_poll_delay_millis: 45000
-      total_poll_timeout_millis: 86400000