Enhance the DSL error store to capture more error info (#101475)

This enhances the DSL error store to caputre more information about each
error, namely: the initial encounter timestamp, the current occurrence
timestamp and the retry count.

This also introduces a new setting,
`data_streams.lifecycle.signalling.error_retry_interval` to control when
we emit an `error` log entry. It defaults to 10 retry counts, namely, if
the error for an index remains the same for 10 consecutive DSL runs, we
log a message at level `error`.

This also exposes all the new error information as part of the
`_lifecycle/explain` API.

Author: andreidan

modules/data-streams/src/internalClusterTest/java/org/elasticsearch/datastreams/lifecycle/DataStreamLifecycleServiceIT.java

@@ -27,6 +27,7 @@
 import org.elasticsearch.action.datastreams.CreateDataStreamAction;
 import org.elasticsearch.action.datastreams.GetDataStreamAction;
 import org.elasticsearch.action.datastreams.ModifyDataStreamsAction;
+import org.elasticsearch.action.datastreams.lifecycle.ErrorEntry;
 import org.elasticsearch.action.datastreams.lifecycle.ExplainIndexDataStreamLifecycle;
 import org.elasticsearch.action.index.IndexRequest;
 import org.elasticsearch.cluster.metadata.ComposableIndexTemplate;
@@ -405,7 +406,7 @@ public void testErrorRecordingOnRollover() throws Exception {
 
         assertBusy(() -> {
             String writeIndexName = getBackingIndices(dataStreamName).get(1);
-            String writeIndexRolloverError = null;
+            ErrorEntry writeIndexRolloverError = null;
             Iterable<DataStreamLifecycleService> lifecycleServices = internalCluster().getInstances(DataStreamLifecycleService.class);
 
             for (DataStreamLifecycleService lifecycleService : lifecycleServices) {
@@ -416,7 +417,7 @@ public void testErrorRecordingOnRollover() throws Exception {
             }
 
             assertThat(writeIndexRolloverError, is(notNullValue()));
-            assertThat(writeIndexRolloverError, containsString("maximum normal shards open"));
+            assertThat(writeIndexRolloverError.error(), containsString("maximum normal shards open"));
         });
 
         // let's reset the cluster max shards per node limit to allow rollover to proceed and check the error store is empty
@@ -497,7 +498,7 @@ public void testErrorRecordingOnRetention() throws Exception {
                 String writeIndex = backingIndices.get(1).getName();
                 assertThat(writeIndex, backingIndexEqualTo(dataStreamName, 2));
 
-                String recordedRetentionExecutionError = null;
+                ErrorEntry recordedRetentionExecutionError = null;
                 Iterable<DataStreamLifecycleService> lifecycleServices = internalCluster().getInstances(DataStreamLifecycleService.class);
 
                 for (DataStreamLifecycleService lifecycleService : lifecycleServices) {
@@ -508,7 +509,7 @@ public void testErrorRecordingOnRetention() throws Exception {
                 }
 
                 assertThat(recordedRetentionExecutionError, is(notNullValue()));
-                assertThat(recordedRetentionExecutionError, containsString("blocked by: [FORBIDDEN/5/index read-only (api)"));
+                assertThat(recordedRetentionExecutionError.error(), containsString("blocked by: [FORBIDDEN/5/index read-only (api)"));
             });
 
             // let's mark the index as writeable and make sure it's deleted and the error store is empty

modules/data-streams/src/internalClusterTest/java/org/elasticsearch/datastreams/lifecycle/ExplainDataStreamLifecycleIT.java

@@ -49,6 +49,7 @@
 import static org.elasticsearch.indices.ShardLimitValidator.SETTING_CLUSTER_MAX_SHARDS_PER_NODE;
 import static org.hamcrest.Matchers.containsString;
 import static org.hamcrest.Matchers.equalTo;
+import static org.hamcrest.Matchers.greaterThanOrEqualTo;
 import static org.hamcrest.Matchers.is;
 import static org.hamcrest.Matchers.notNullValue;
 import static org.hamcrest.Matchers.nullValue;
@@ -252,7 +253,9 @@ public void testExplainLifecycleForIndicesWithErrors() throws Exception {
                 // index has not been rolled over yet
                 assertThat(explainIndex.getGenerationTime(System::currentTimeMillis), nullValue());
 
-                assertThat(explainIndex.getError(), containsString("maximum normal shards open"));
+                assertThat(explainIndex.getError(), notNullValue());
+                assertThat(explainIndex.getError().error(), containsString("maximum normal shards open"));
+                assertThat(explainIndex.getError().retryCount(), greaterThanOrEqualTo(1));
             }
         });
 

modules/data-streams/src/main/java/org/elasticsearch/datastreams/DataStreamsPlugin.java

@@ -141,6 +141,7 @@ public List<Setting<?>> getSettings() {
         pluginSettings.add(DataStreamLifecycleService.DATA_STREAM_LIFECYCLE_POLL_INTERVAL_SETTING);
         pluginSettings.add(DataStreamLifecycleService.DATA_STREAM_MERGE_POLICY_TARGET_FLOOR_SEGMENT_SETTING);
         pluginSettings.add(DataStreamLifecycleService.DATA_STREAM_MERGE_POLICY_TARGET_FACTOR_SETTING);
+        pluginSettings.add(DataStreamLifecycleService.DATA_STREAM_SIGNALLING_ERROR_RETRY_INTERVAL_SETTING);
         return pluginSettings;
     }
 
@@ -155,7 +156,7 @@ public Collection<?> createComponents(PluginServices services) {
         );
         this.updateTimeSeriesRangeService.set(updateTimeSeriesRangeService);
         components.add(this.updateTimeSeriesRangeService.get());
-        errorStoreInitialisationService.set(new DataStreamLifecycleErrorStore());
+        errorStoreInitialisationService.set(new DataStreamLifecycleErrorStore(services.threadPool()::absoluteTimeInMillis));
         dataLifecycleInitialisationService.set(
             new DataStreamLifecycleService(
                 settings,

modules/data-streams/src/main/java/org/elasticsearch/datastreams/lifecycle/DataStreamLifecycleErrorStore.java

@@ -9,12 +9,14 @@
 package org.elasticsearch.datastreams.lifecycle;
 
 import org.elasticsearch.ElasticsearchException;
+import org.elasticsearch.action.datastreams.lifecycle.ErrorEntry;
 import org.elasticsearch.common.Strings;
 import org.elasticsearch.core.Nullable;
 
 import java.util.List;
 import java.util.concurrent.ConcurrentHashMap;
 import java.util.concurrent.ConcurrentMap;
+import java.util.function.LongSupplier;
 
 import static org.elasticsearch.xcontent.ToXContent.EMPTY_PARAMS;
 
@@ -26,7 +28,12 @@
 public class DataStreamLifecycleErrorStore {
 
     public static final int MAX_ERROR_MESSAGE_LENGTH = 1000;
-    private final ConcurrentMap<String, String> indexNameToError = new ConcurrentHashMap<>();
+    private final ConcurrentMap<String, ErrorEntry> indexNameToError = new ConcurrentHashMap<>();
+    private final LongSupplier nowSupplier;
+
+    public DataStreamLifecycleErrorStore(LongSupplier nowSupplier) {
+        this.nowSupplier = nowSupplier;
+    }
 
     /**
      * Records a string representation of the provided exception for the provided index.
@@ -35,13 +42,24 @@ public class DataStreamLifecycleErrorStore {
      * Returns the previously recorded error for the provided index, or null otherwise.
      */
     @Nullable
-    public String recordError(String indexName, Exception e) {
+    public ErrorEntry recordError(String indexName, Exception e) {
         String exceptionToString = Strings.toString((builder, params) -> {
             ElasticsearchException.generateThrowableXContent(builder, EMPTY_PARAMS, e);
             return builder;
         });
-        String recordedError = Strings.substring(exceptionToString, 0, MAX_ERROR_MESSAGE_LENGTH);
-        return indexNameToError.put(indexName, recordedError);
+        String newError = Strings.substring(exceptionToString, 0, MAX_ERROR_MESSAGE_LENGTH);
+        ErrorEntry existingError = indexNameToError.get(indexName);
+        long recordedTimestamp = nowSupplier.getAsLong();
+        if (existingError == null) {
+            indexNameToError.put(indexName, new ErrorEntry(recordedTimestamp, newError, recordedTimestamp, 0));
+        } else {
+            if (existingError.error().equals(newError)) {
+                indexNameToError.put(indexName, ErrorEntry.incrementRetryCount(existingError, nowSupplier));
+            } else {
+                indexNameToError.put(indexName, new ErrorEntry(recordedTimestamp, newError, recordedTimestamp, 0));
+            }
+        }
+        return existingError;
     }
 
     /**
@@ -62,7 +80,7 @@ public void clearStore() {
      * Retrieves the recorded error for the provided index.
      */
     @Nullable
-    public String getError(String indexName) {
+    public ErrorEntry getError(String indexName) {
         return indexNameToError.get(indexName);
     }