Fix NPE in RestoreService when shard size is unavailable

Signed-off-by: Sotaro Hikita <[email protected]>

Author: lawofcycles

CHANGELOG.md

@@ -35,6 +35,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Fix issue with updating core with a patch number other than 0 ([#19377](https://github.com/opensearch-project/OpenSearch/pull/19377))
 - [Java Agent] Allow JRT protocol URLs in protection domain extraction ([#19683](https://github.com/opensearch-project/OpenSearch/pull/19683))
 - Fix potential concurrent modification exception when updating allocation filters ([#19701])(https://github.com/opensearch-project/OpenSearch/pull/19701))
+- Fix NullPointerException when restoring remote snapshot with missing shard size information ([#19684](https://github.com/opensearch-project/OpenSearch/pull/19684))
 
 ### Dependencies
 - Update to Gradle 9.1 ([#19575](https://github.com/opensearch-project/OpenSearch/pull/19575))

server/src/internalClusterTest/java/org/opensearch/snapshots/SearchableSnapshotIT.java

@@ -1054,4 +1054,61 @@ private void assertSearchableSnapshotIndexDirectoryExistence(String nodeName, In
             assertTrue("index cache path should " + (exists ? "exist" : "not exist"), Files.exists(indexPath) == exists);
         }, 30, TimeUnit.SECONDS);
     }
+
+    public void testRestoreRemoteSnapshotWithNullShardSizes() throws Exception {
+        final String snapshotName1 = "test-snap-1";
+        final String snapshotName2 = "test-snap-2";
+        final String repoName = "test-repo";
+        final String indexName1 = "test-idx-1";
+        final String indexName2 = "test-idx-2";
+        final Client client = client();
+
+        // Setup cluster with delayed ClusterInfo updates to simulate null shard sizes
+        client.admin()
+            .cluster()
+            .prepareUpdateSettings()
+            .setPersistentSettings(Settings.builder().put("cluster.info.update.interval", "60m"))
+            .get();
+
+        internalCluster().ensureAtLeastNumDataNodes(2);
+
+        createIndexWithDocsAndEnsureGreen(0, 50, indexName1);
+        createRepositoryWithSettings(null, repoName);
+        takeSnapshot(client, snapshotName1, repoName, indexName1);
+        deleteIndicesAndEnsureGreen(client, indexName1);
+
+        createIndexWithDocsAndEnsureGreen(0, 50, indexName2);
+        takeSnapshot(client, snapshotName2, repoName, indexName2);
+        deleteIndicesAndEnsureGreen(client, indexName2);
+
+        internalCluster().ensureAtLeastNumWarmNodes(2);
+
+        RestoreSnapshotRequest firstRestore = new RestoreSnapshotRequest(repoName, snapshotName1).indices(indexName1)
+            .storageType(RestoreSnapshotRequest.StorageType.REMOTE_SNAPSHOT)
+            .renamePattern("(.+)")
+            .renameReplacement("remote-$1")
+            .waitForCompletion(true);
+
+        client.admin().cluster().restoreSnapshot(firstRestore).get();
+        ensureGreen("remote-" + indexName1);
+
+        // Second restore immediately after - ClusterInfo won't have size data for first restore shards
+        RestoreSnapshotRequest secondRestore = new RestoreSnapshotRequest(repoName, snapshotName2).indices(indexName2)
+            .storageType(RestoreSnapshotRequest.StorageType.REMOTE_SNAPSHOT)
+            .renamePattern("(.+)")
+            .renameReplacement("remote-$1")
+            .waitForCompletion(true);
+
+        client.admin().cluster().restoreSnapshot(secondRestore).get();
+        ensureGreen("remote-" + indexName2);
+
+        assertDocCount("remote-" + indexName1, 50L);
+        assertDocCount("remote-" + indexName2, 50L);
+
+        client.admin()
+            .cluster()
+            .prepareUpdateSettings()
+            .setPersistentSettings(Settings.builder().putNull("cluster.info.update.interval"))
+            .get();
+    }
 }

server/src/main/java/org/opensearch/snapshots/RestoreService.java

@@ -909,13 +909,29 @@ private void validateSearchableSnapshotRestorable(long totalRestorableRemoteInde
                             .routingTable()
                             .allShardsSatisfyingPredicate(isRemoteSnapshotShard);
 
-                        long totalRestoredRemoteIndexesSize = shardsIterator.getShardRoutings()
-                            .stream()
-                            .map(clusterInfo::getShardSize)
-                            .mapToLong(Long::longValue)
-                            .sum();
+                        long totalRestoredRemoteIndicesSize = 0;
+                        int missingSizeCount = 0;
+                        List<ShardRouting> routings = shardsIterator.getShardRoutings();
+
+                        for (ShardRouting shardRouting : routings) {
+                            Long shardSize = clusterInfo.getShardSize(shardRouting);
+                            if (shardSize != null) {
+                                totalRestoredRemoteIndicesSize += shardSize;
+                            } else {
+                                missingSizeCount++;
+                            }
+                        }
+
+                        if (missingSizeCount > 0) {
+                            logger.warn(
+                                "Size information unavailable for {} out of {} remote snapshot shards. "
+                                    + "File cache validation will use available data only.",
+                                missingSizeCount,
+                                routings.size()
+                            );
+                        }
 
-                        if (totalRestoredRemoteIndexesSize + totalRestorableRemoteIndexesSize > remoteDataToFileCacheRatio
+                        if (totalRestoredRemoteIndicesSize + totalRestorableRemoteIndexesSize > remoteDataToFileCacheRatio
                             * totalNodeFileCacheSize) {
                             throw new SnapshotRestoreException(
                                 snapshot,