Fix pre-sorting of shards in the can_match phase (#53397)

This commit fixes a bug on sorted queries with a primary sort field
that uses different types in the requested indices. In this scenario
the returned min/max values to sort the shards are not comparable so
we should avoid the sorting rather than throwing an obscure exception.

Author: jimczi

server/src/main/java/org/elasticsearch/action/search/CanMatchPreFilterSearchPhase.java

@@ -31,11 +31,9 @@
 import org.elasticsearch.search.sort.SortOrder;
 import org.elasticsearch.transport.Transport;
 
-import java.util.Arrays;
 import java.util.Comparator;
 import java.util.List;
 import java.util.Map;
-import java.util.Objects;
 import java.util.Set;
 import java.util.concurrent.Executor;
 import java.util.function.BiFunction;
@@ -127,7 +125,18 @@ private static List<SearchShardIterator> sortShards(GroupShardsIterator<SearchSh
     }
 
     private static boolean shouldSortShards(MinAndMax<?>[] minAndMaxes) {
-        return Arrays.stream(minAndMaxes).anyMatch(Objects::nonNull);
+        Class<?> clazz = null;
+        for (MinAndMax<?> minAndMax : minAndMaxes) {
+            if (clazz == null) {
+                clazz = minAndMax == null ? null : minAndMax.getMin().getClass();
+            } else if (minAndMax != null && clazz != minAndMax.getMin().getClass()) {
+                // we don't support sort values that mix different types (e.g.: long/double, numeric/keyword).
+                // TODO: we could fail the request because there is a high probability
+                //  that the merging of topdocs will fail later for the same reason ?
+                return false;
+            }
+        }
+        return clazz != null;
     }
 
     private static Comparator<Integer> shardComparator(GroupShardsIterator<SearchShardIterator> shardsIts,

server/src/test/java/org/elasticsearch/action/search/CanMatchPreFilterSearchPhaseTests.java

@@ -18,6 +18,7 @@
  */
 package org.elasticsearch.action.search;
 
+import org.apache.lucene.util.BytesRef;
 import org.elasticsearch.Version;
 import org.elasticsearch.action.ActionListener;
 import org.elasticsearch.action.OriginalIndices;
@@ -53,6 +54,8 @@
 import java.util.concurrent.atomic.AtomicReference;
 import java.util.stream.IntStream;
 
+import static org.hamcrest.Matchers.equalTo;
+
 public class CanMatchPreFilterSearchPhaseTests extends ESTestCase {
 
     public void testFilterShards() throws InterruptedException {
@@ -349,4 +352,76 @@ public void run() {
             }
         }
     }
+
+    public void testInvalidSortShards() throws InterruptedException {
+        final TransportSearchAction.SearchTimeProvider timeProvider =
+            new TransportSearchAction.SearchTimeProvider(0, System.nanoTime(), System::nanoTime);
+
+        Map<String, Transport.Connection> lookup = new ConcurrentHashMap<>();
+        DiscoveryNode primaryNode = new DiscoveryNode("node_1", buildNewFakeTransportAddress(), Version.CURRENT);
+        DiscoveryNode replicaNode = new DiscoveryNode("node_2", buildNewFakeTransportAddress(), Version.CURRENT);
+        lookup.put("node1", new SearchAsyncActionTests.MockConnection(primaryNode));
+        lookup.put("node2", new SearchAsyncActionTests.MockConnection(replicaNode));
+
+        for (SortOrder order : SortOrder.values()) {
+            int numShards = randomIntBetween(2, 20);
+            List<ShardId> shardIds = new ArrayList<>();
+            Set<ShardId> shardToSkip = new HashSet<>();
+
+            SearchTransportService searchTransportService = new SearchTransportService(null, null) {
+                @Override
+                public void sendCanMatch(Transport.Connection connection, ShardSearchRequest request, SearchTask task,
+                                         ActionListener<SearchService.CanMatchResponse> listener) {
+                    final MinAndMax<?> minMax;
+                    if (request.shardId().id() == numShards-1) {
+                        minMax = MinAndMax.newMinMax(new BytesRef("bar"), new BytesRef("baz"));
+                    } else {
+                        Long min = randomLong();
+                        Long max = randomLongBetween(min, Long.MAX_VALUE);
+                        minMax = MinAndMax.newMinMax(min, max);
+                    }
+                    boolean canMatch = frequently();
+                    synchronized (shardIds) {
+                        shardIds.add(request.shardId());
+                        if (canMatch == false) {
+                            shardToSkip.add(request.shardId());
+                        }
+                    }
+                    new Thread(() -> listener.onResponse(new SearchService.CanMatchResponse(canMatch, minMax))).start();
+                }
+            };
+
+            AtomicReference<GroupShardsIterator<SearchShardIterator>> result = new AtomicReference<>();
+            CountDownLatch latch = new CountDownLatch(1);
+            GroupShardsIterator<SearchShardIterator> shardsIter = SearchAsyncActionTests.getShardsIter("logs",
+                new OriginalIndices(new String[]{"logs"}, SearchRequest.DEFAULT_INDICES_OPTIONS),
+                numShards, randomBoolean(), primaryNode, replicaNode);
+            final SearchRequest searchRequest = new SearchRequest();
+            searchRequest.source(new SearchSourceBuilder().sort(SortBuilders.fieldSort("timestamp").order(order)));
+            searchRequest.allowPartialSearchResults(true);
+
+            CanMatchPreFilterSearchPhase canMatchPhase = new CanMatchPreFilterSearchPhase(logger,
+                searchTransportService,
+                (clusterAlias, node) -> lookup.get(node),
+                Collections.singletonMap("_na_", new AliasFilter(null, Strings.EMPTY_ARRAY)),
+                Collections.emptyMap(), Collections.emptyMap(), EsExecutors.newDirectExecutorService(),
+                searchRequest, null, shardsIter, timeProvider, 0, null,
+                (iter) -> new SearchPhase("test") {
+                    @Override
+                    public void run() {
+                        result.set(iter);
+                        latch.countDown();
+                    }
+                }, SearchResponse.Clusters.EMPTY);
+
+            canMatchPhase.start();
+            latch.await();
+            int shardId = 0;
+            for (SearchShardIterator i : result.get()) {
+                assertThat(i.shardId().id(), equalTo(shardId++));
+                assertEquals(shardToSkip.contains(i.shardId()), i.skip());
+            }
+            assertThat(result.get().size(), equalTo(numShards));
+        }
+    }
 }