Support real-time gets on hollow shards

server/src/main/java/org/elasticsearch/action/get/TransportGetAction.java

@@ -11,6 +11,7 @@
 
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
+import org.apache.lucene.store.AlreadyClosedException;
 import org.elasticsearch.ElasticsearchException;
 import org.elasticsearch.ExceptionsHelper;
 import org.elasticsearch.action.ActionListener;
@@ -230,7 +231,10 @@ private void getFromTranslog(
         final var retryingListener = listener.delegateResponse((l, e) -> {
             final var cause = ExceptionsHelper.unwrapCause(e);
             logger.debug("get_from_translog failed", cause);
-            if (cause instanceof ShardNotFoundException || cause instanceof IndexNotFoundException) {
+            if (cause instanceof ShardNotFoundException
+                || cause instanceof IndexNotFoundException
+                || cause instanceof AlreadyClosedException) {
+                // TODO AlreadyClosedException the engine reset should be fixed by ES-10826
                 logger.debug("retrying get_from_translog");
                 observer.waitForNextChange(new ClusterStateObserver.Listener() {
                     @Override
@@ -245,7 +249,13 @@ public void onClusterServiceClose() {
 
                     @Override
                     public void onTimeout(TimeValue timeout) {
-                        l.onFailure(new ElasticsearchException("Timed out retrying get_from_translog", cause));
+                        // TODO AlreadyClosedException the engine reset should be fixed by ES-10826
+                        if (cause instanceof AlreadyClosedException) {
+                            // Do an additional retry just in case AlreadyClosedException didn't generate a cluster update
+                            tryGetFromTranslog(request, indexShard, node, l);
+                        } else {
+                            l.onFailure(new ElasticsearchException("Timed out retrying get_from_translog", cause));
+                        }
                     }
                 });
             } else {

server/src/main/java/org/elasticsearch/action/get/TransportGetFromTranslogAction.java

@@ -81,7 +81,7 @@ protected void doExecute(Task task, Request request, ActionListener<Response> li
                 if (engine == null) {
                     throw new AlreadyClosedException("engine closed");
                 }
-                segmentGeneration = ((InternalEngine) engine).getLastUnsafeSegmentGenerationForGets();
+                segmentGeneration = engine.getLastUnsafeSegmentGenerationForGets();
             }
             return new Response(result, indexShard.getOperationPrimaryTerm(), segmentGeneration);
         });

server/src/main/java/org/elasticsearch/action/get/TransportShardMultiGetAction.java

@@ -11,6 +11,7 @@
 
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
+import org.apache.lucene.store.AlreadyClosedException;
 import org.elasticsearch.ElasticsearchException;
 import org.elasticsearch.ExceptionsHelper;
 import org.elasticsearch.action.ActionListener;
@@ -211,7 +212,10 @@ private void shardMultiGetFromTranslog(
         final var retryingListener = listener.delegateResponse((l, e) -> {
             final var cause = ExceptionsHelper.unwrapCause(e);
             logger.debug("mget_from_translog[shard] failed", cause);
-            if (cause instanceof ShardNotFoundException || cause instanceof IndexNotFoundException) {
+            if (cause instanceof ShardNotFoundException
+                || cause instanceof IndexNotFoundException
+                || cause instanceof AlreadyClosedException) {
+                // TODO AlreadyClosedException the engine reset should be fixed by ES-10826
                 logger.debug("retrying mget_from_translog[shard]");
                 observer.waitForNextChange(new ClusterStateObserver.Listener() {
                     @Override
@@ -226,7 +230,13 @@ public void onClusterServiceClose() {
 
                     @Override
                     public void onTimeout(TimeValue timeout) {
-                        l.onFailure(new ElasticsearchException("Timed out retrying mget_from_translog[shard]", cause));
+                        // TODO AlreadyClosedException the engine reset should be fixed by ES-10826
+                        if (cause instanceof AlreadyClosedException) {
+                            // Do an additional retry just in case AlreadyClosedException didn't generate a cluster update
+                            tryShardMultiGetFromTranslog(request, indexShard, node, l);
+                        } else {
+                            l.onFailure(new ElasticsearchException("Timed out retrying mget_from_translog[shard]", cause));
+                        }
                     }
                 });
             } else {

server/src/main/java/org/elasticsearch/action/get/TransportShardMultiGetFomTranslogAction.java

@@ -23,7 +23,6 @@
 import org.elasticsearch.common.io.stream.StreamOutput;
 import org.elasticsearch.index.IndexService;
 import org.elasticsearch.index.engine.Engine;
-import org.elasticsearch.index.engine.InternalEngine;
 import org.elasticsearch.index.shard.IndexShard;
 import org.elasticsearch.index.shard.ShardId;
 import org.elasticsearch.indices.IndicesService;
@@ -102,7 +101,7 @@ protected void doExecute(Task task, Request request, ActionListener<Response> li
                 if (engine == null) {
                     throw new AlreadyClosedException("engine closed");
                 }
-                segmentGeneration = ((InternalEngine) engine).getLastUnsafeSegmentGenerationForGets();
+                segmentGeneration = engine.getLastUnsafeSegmentGenerationForGets();
             }
             return new Response(multiGetShardResponse, indexShard.getOperationPrimaryTerm(), segmentGeneration);
         });

server/src/main/java/org/elasticsearch/index/engine/Engine.java

@@ -2346,4 +2346,8 @@ public record FlushResult(boolean flushPerformed, long generation) {
     public void prepareForEngineReset() throws IOException {
         throw new UnsupportedOperationException("does not support engine reset");
     }
+
+    public long getLastUnsafeSegmentGenerationForGets() {
+        throw new UnsupportedOperationException("Doesn't support getting the latest segment generation");
+    }
 }