[ML] Integrate calls to EIS for preconfigured inference endpoints for ModelRegistry functionality

x-pack/plugin/build.gradle

@@ -155,6 +155,7 @@ tasks.named("yamlRestCompatTestTransform").configure({ task ->
   task.skipTest("esql/46_downsample/Query stats on downsampled index", "Extra function required to enable the field type")
   task.skipTest("esql/46_downsample/Render stats from downsampled index", "Extra function required to enable the field type")
   task.skipTest("esql/46_downsample/Sort from multiple indices one with aggregate metric double", "Extra function required to enable the field type")
+  task.skipTest("inference/inference_crud/Test get missing model", "Error message changed")
 })
 
 tasks.named('yamlRestCompatTest').configure {

x-pack/plugin/esql/qa/server/src/main/java/org/elasticsearch/xpack/esql/qa/rest/SemanticMatchTestCase.java

@@ -67,7 +67,7 @@ public void testWithInferenceNotConfigured() {
             """;
         ResponseException re = expectThrows(ResponseException.class, () -> runEsqlQuery(query));
 
-        assertThat(re.getMessage(), containsString("Inference endpoint not found"));
+        assertThat(re.getMessage(), containsString("Inference endpoint [inexistent] not found"));
         assertEquals(404, re.getResponse().getStatusLine().getStatusCode());
     }
 

x-pack/plugin/inference/build.gradle

@@ -9,6 +9,7 @@ apply plugin: 'elasticsearch.internal-es-plugin'
 apply plugin: 'elasticsearch.internal-cluster-test'
 apply plugin: 'elasticsearch.internal-yaml-rest-test'
 apply plugin: 'elasticsearch.internal-test-artifact'
+apply plugin: 'elasticsearch.yaml-rest-compat-test'
 
 restResources {
   restApi {
@@ -407,6 +408,13 @@ tasks.named('yamlRestTest') {
   usesDefaultDistribution("Uses the inference API")
 }
 
+tasks.named("yamlRestCompatTestTransform").configure({ task ->
+  task.skipTest("inference/40_semantic_text_query/Query a field with an invalid inference ID", "Error message changed")
+  task.skipTest("inference/40_semantic_text_query/Query a field with an invalid search inference ID", "Error message changed")
+  task.skipTest("inference/70_text_similarity_rank_retriever/Text similarity reranking fails if the inference ID does not exist", "Error message changed")
+  task.skipTest("inference/70_text_similarity_rank_retriever/Text similarity reranking fails if the inference ID does not exist and result set is empty", "Error message changed")
+})
+
 artifacts {
   restXpackTests(new File(projectDir, "src/yamlRestTest/resources/rest-api-spec/test"))
 }

x-pack/plugin/inference/qa/inference-service-tests/src/javaRestTest/java/org/elasticsearch/xpack/inference/BaseMockEISAuthServerTest.java

@@ -27,11 +27,6 @@ public class BaseMockEISAuthServerTest extends ESRestTestCase {
     protected static final MockElasticInferenceServiceAuthorizationServer mockEISServer =
         new MockElasticInferenceServiceAuthorizationServer();
 
-    static {
-        // Ensure that the mock EIS server has an authorized response prior to the cluster starting
-        mockEISServer.enqueueAuthorizeAllModelsResponse();
-    }
-
     private static ElasticsearchCluster cluster = ElasticsearchCluster.local()
         .distribution(DistributionType.DEFAULT)
         .setting("xpack.license.self_generated.type", "trial")

x-pack/plugin/inference/qa/inference-service-tests/src/javaRestTest/java/org/elasticsearch/xpack/inference/InferenceGetModelsWithElasticInferenceServiceIT.java

@@ -10,7 +10,6 @@
 package org.elasticsearch.xpack.inference;
 
 import org.elasticsearch.inference.TaskType;
-import org.junit.BeforeClass;
 
 import java.io.IOException;
 import java.util.List;
@@ -22,24 +21,11 @@
 import static org.hamcrest.Matchers.is;
 
 public class InferenceGetModelsWithElasticInferenceServiceIT extends BaseMockEISAuthServerTest {
-
-    /**
-     * This is done before the class because I've run into issues where another class that extends {@link BaseMockEISAuthServerTest}
-     * results in an authorization response not being queued up for the new Elasticsearch Node in time. When the node starts up, it
-     * retrieves authorization. If the request isn't queued up when that happens the tests will fail. From my testing locally it seems
-     * like the base class's static functionality to queue a response is only done once and not for each subclass.
-     *
-     * My understanding is that the @Before will be run after the node starts up and wouldn't be sufficient to handle
-     * this scenario. That is why this needs to be @BeforeClass.
-     */
-    @BeforeClass
-    public static void init() {
-        // Ensure the mock EIS server has an authorized response ready
-        mockEISServer.enqueueAuthorizeAllModelsResponse();
-    }
-
     public void testGetDefaultEndpoints() throws IOException {
+        mockEISServer.enqueueAuthorizeAllModelsResponse();
         var allModels = getAllModels();
+
+        mockEISServer.enqueueAuthorizeAllModelsResponse();
         var chatCompletionModels = getModels("_all", TaskType.CHAT_COMPLETION);
 
         assertThat(allModels, hasSize(7));

x-pack/plugin/inference/qa/inference-service-tests/src/javaRestTest/java/org/elasticsearch/xpack/inference/InferenceGetServicesIT.java

@@ -13,7 +13,6 @@
 import org.elasticsearch.common.Strings;
 import org.elasticsearch.inference.TaskType;
 import org.junit.Before;
-import org.junit.BeforeClass;
 
 import java.io.IOException;
 import java.util.List;
@@ -32,21 +31,6 @@ public void setUp() throws Exception {
         mockEISServer.enqueueAuthorizeAllModelsResponse();
     }
 
-    /**
-     * This is done before the class because I've run into issues where another class that extends {@link BaseMockEISAuthServerTest}
-     * results in an authorization response not being queued up for the new Elasticsearch Node in time. When the node starts up, it
-     * retrieves authorization. If the request isn't queued up when that happens the tests will fail. From my testing locally it seems
-     * like the base class's static functionality to queue a response is only done once and not for each subclass.
-     *
-     * My understanding is that the @Before will be run after the node starts up and wouldn't be sufficient to handle
-     * this scenario. That is why this needs to be @BeforeClass.
-     */
-    @BeforeClass
-    public static void init() {
-        // Ensure the mock EIS server has an authorized response ready
-        mockEISServer.enqueueAuthorizeAllModelsResponse();
-    }
-
     public void testGetServicesWithoutTaskType() throws IOException {
         assertThat(
             allProviders(),

x-pack/plugin/inference/qa/inference-service-tests/src/javaRestTest/java/org/elasticsearch/xpack/inference/InferenceGetServicesWithoutEisIT.java

@@ -0,0 +1,121 @@
+/*
+ * Copyright Elasticsearch B.V. and/or licensed to Elasticsearch B.V. under one
+ * or more contributor license agreements. Licensed under the Elastic License
+ * 2.0; you may not use this file except in compliance with the Elastic License
+ * 2.0.
+ *
+ * this file has been contributed to by a Generative AI
+ */
+
+package org.elasticsearch.xpack.inference;
+
+import org.elasticsearch.client.Request;
+import org.elasticsearch.common.Strings;
+import org.elasticsearch.common.settings.SecureString;
+import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.common.util.concurrent.ThreadContext;
+import org.elasticsearch.inference.TaskType;
+import org.elasticsearch.test.cluster.ElasticsearchCluster;
+import org.elasticsearch.test.cluster.local.distribution.DistributionType;
+import org.elasticsearch.test.rest.ESRestTestCase;
+import org.junit.ClassRule;
+
+import java.io.IOException;
+import java.util.List;
+import java.util.Map;
+
+import static org.elasticsearch.xpack.inference.InferenceBaseRestTest.assertStatusOkOrCreated;
+import static org.hamcrest.Matchers.equalTo;
+import static org.hamcrest.Matchers.hasItem;
+import static org.hamcrest.Matchers.not;
+
+public class InferenceGetServicesWithoutEisIT extends ESRestTestCase {
+
+    @ClassRule
+    public static ElasticsearchCluster cluster = ElasticsearchCluster.local()
+        .distribution(DistributionType.DEFAULT)
+        .setting("xpack.license.self_generated.type", "trial")
+        .setting("xpack.security.enabled", "true")
+        // This plugin is located in the inference/qa/test-service-plugin package, look for TestInferenceServicePlugin
+        .plugin("inference-service-test")
+        .user("x_pack_rest_user", "x-pack-test-password")
+        .build();
+
+    @Override
+    protected String getTestRestCluster() {
+        return cluster.getHttpAddresses();
+    }
+
+    @Override
+    protected Settings restClientSettings() {
+        String token = basicAuthHeaderValue("x_pack_rest_user", new SecureString("x-pack-test-password".toCharArray()));
+        return Settings.builder().put(ThreadContext.PREFIX + ".Authorization", token).build();
+    }
+
+    public void testGetServicesWithoutTaskType() throws IOException {
+        assertThat(allProviders(), not(hasItem("elastic")));
+    }
+
+    private List<String> allProviders() throws IOException {
+        return providers(getAllServices());
+    }
+
+    @SuppressWarnings("unchecked")
+    private List<String> providers(List<Object> services) {
+        return services.stream().map(service -> {
+            var serviceConfig = (Map<String, Object>) service;
+            return (String) serviceConfig.get("service");
+        }).toList();
+    }
+
+    public void testGetServicesWithTextEmbeddingTaskType() throws IOException {
+        var providers = providersFor(TaskType.TEXT_EMBEDDING);
+        assertThat(providers.size(), not(equalTo(0)));
+        assertThat(providers, not(hasItem("elastic")));
+    }
+
+    private List<String> providersFor(TaskType taskType) throws IOException {
+        return providers(getServices(taskType));
+    }
+
+    public void testGetServicesWithRerankTaskType() throws IOException {
+        var providers = providersFor(TaskType.RERANK);
+        assertThat(providers.size(), not(equalTo(0)));
+        assertThat(providersFor(TaskType.RERANK), not(hasItem("elastic")));
+    }
+
+    public void testGetServicesWithCompletionTaskType() throws IOException {
+        var providers = providersFor(TaskType.COMPLETION);
+        assertThat(providers.size(), not(equalTo(0)));
+        assertThat(providersFor(TaskType.COMPLETION), not(hasItem("elastic")));
+    }
+
+    public void testGetServicesWithChatCompletionTaskType() throws IOException {
+        var providers = providersFor(TaskType.CHAT_COMPLETION);
+        assertThat(providers.size(), not(equalTo(0)));
+        assertThat(providersFor(TaskType.CHAT_COMPLETION), not(hasItem("elastic")));
+    }
+
+    public void testGetServicesWithSparseEmbeddingTaskType() throws IOException {
+        var providers = providersFor(TaskType.SPARSE_EMBEDDING);
+        assertThat(providers.size(), not(equalTo(0)));
+        assertThat(providersFor(TaskType.SPARSE_EMBEDDING), not(hasItem("elastic")));
+    }
+
+    private List<Object> getAllServices() throws IOException {
+        var endpoint = Strings.format("_inference/_services");
+        return getInternalAsList(endpoint);
+    }
+
+    private List<Object> getServices(TaskType taskType) throws IOException {
+        var endpoint = Strings.format("_inference/_services/%s", taskType);
+        return getInternalAsList(endpoint);
+    }
+
+    private List<Object> getInternalAsList(String endpoint) throws IOException {
+        var request = new Request("GET", endpoint);
+        var response = client().performRequest(request);
+        assertStatusOkOrCreated(response);
+        return entityAsList(response);
+    }
+}

x-pack/plugin/inference/src/internalClusterTest/java/org/elasticsearch/xpack/inference/integration/InferenceIndicesIT.java

@@ -167,7 +167,7 @@ public void testRetrievingInferenceEndpoint_ThrowsException_WhenIndexNodeIsNotAv
 
         var proxyResponse = sendInferenceProxyRequest(inferenceId);
         var exception = expectThrows(ElasticsearchException.class, () -> proxyResponse.actionGet(TEST_REQUEST_TIMEOUT));
-        assertThat(exception.toString(), containsString("Failed to load inference endpoint with secrets [test-index-id-2]"));
+        assertThat(exception.toString(), containsString("Failed to load inference endpoint [test-index-id-2]"));
 
         var causeException = exception.getCause();
         assertThat(causeException, instanceOf(SearchPhaseExecutionException.class));
@@ -196,7 +196,7 @@ public void testRetrievingInferenceEndpoint_ThrowsException_WhenSecretsIndexNode
         var proxyResponse = sendInferenceProxyRequest(inferenceId);
 
         var exception = expectThrows(ElasticsearchException.class, () -> proxyResponse.actionGet(TEST_REQUEST_TIMEOUT));
-        assertThat(exception.toString(), containsString("Failed to load inference endpoint with secrets [test-secrets-index-id]"));
+        assertThat(exception.toString(), containsString("Failed to load inference endpoint [test-secrets-index-id]"));
 
         var causeException = exception.getCause();