MetaData Builder doesn't properly prevent an alias with the same name as an index (#26804)

Elasticsearch doesn't allow having an index alias named with the same name as an existing index. We currently have logic that tries to prevents that in the `MetaData.Builder#build()` method. Sadly that logic is flawed. Depending on iteration order, we may allow the above to happen (if we encounter the alias before the index).

This commit fixes the above and improves the error message while at it.

Note that we have a lot of protections in place before we end up relying on the metadata builder (validating this when we process APIs). I takes quite an abuse of the cluster to get that far.

core/src/main/java/org/elasticsearch/cluster/metadata/MetaData.java

@@ -22,7 +22,6 @@ package org.elasticsearch.cluster.metadata;
 import com.carrotsearch.hppc.ObjectHashSet;
 import com.carrotsearch.hppc.cursors.ObjectCursor;
 import com.carrotsearch.hppc.cursors.ObjectObjectCursor;
-
 import org.apache.logging.log4j.Logger;
 import org.apache.lucene.util.CollectionUtil;
 import org.elasticsearch.cluster.Diff;
@@ -33,6 +32,7 @@ import org.elasticsearch.cluster.NamedDiffableValueSerializer;
 import org.elasticsearch.cluster.block.ClusterBlock;
 import org.elasticsearch.cluster.block.ClusterBlockLevel;
 import org.elasticsearch.common.Nullable;
+import org.elasticsearch.common.Strings;
 import org.elasticsearch.common.UUIDs;
 import org.elasticsearch.common.collect.HppcMaps;
 import org.elasticsearch.common.collect.ImmutableOpenMap;
@@ -62,9 +62,11 @@ import java.util.Collections;
 import java.util.Comparator;
 import java.util.EnumSet;
 import java.util.HashMap;
+import java.util.HashSet;
 import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
+import java.util.Set;
 import java.util.SortedMap;
 import java.util.TreeMap;
 
@@ -914,55 +916,70 @@ public class MetaData implements Iterable<IndexMetaData>, Diffable<MetaData>, To
             //    while these datastructures aren't even used.
             // 2) The aliasAndIndexLookup can be updated instead of rebuilding it all the time.
 
-            // build all concrete indices arrays:
-            // TODO: I think we can remove these arrays. it isn't worth the effort, for operations on all indices.
-            // When doing an operation across all indices, most of the time is spent on actually going to all shards and
-            // do the required operations, the bottleneck isn't resolving expressions into concrete indices.
-            List<String> allIndicesLst = new ArrayList<>();
+            final Set<String> allIndices = new HashSet<>(indices.size());
+            final List<String> allOpenIndices = new ArrayList<>();
+            final List<String> allClosedIndices = new ArrayList<>();
+            final Set<String> duplicateAliasesIndices = new HashSet<>();
             for (ObjectCursor<IndexMetaData> cursor : indices.values()) {
-                allIndicesLst.add(cursor.value.getIndex().getName());
-            }
-            String[] allIndices = allIndicesLst.toArray(new String[allIndicesLst.size()]);
-
-            List<String> allOpenIndicesLst = new ArrayList<>();
-            List<String> allClosedIndicesLst = new ArrayList<>();
-            for (ObjectCursor<IndexMetaData> cursor : indices.values()) {
-                IndexMetaData indexMetaData = cursor.value;
+                final IndexMetaData indexMetaData = cursor.value;
+                final String name = indexMetaData.getIndex().getName();
+                boolean added = allIndices.add(name);
+                assert added : "double index named [" + name + "]";
                 if (indexMetaData.getState() == IndexMetaData.State.OPEN) {
-                    allOpenIndicesLst.add(indexMetaData.getIndex().getName());
+                    allOpenIndices.add(indexMetaData.getIndex().getName());
                 } else if (indexMetaData.getState() == IndexMetaData.State.CLOSE) {
-                    allClosedIndicesLst.add(indexMetaData.getIndex().getName());
+                    allClosedIndices.add(indexMetaData.getIndex().getName());
+                }
+                indexMetaData.getAliases().keysIt().forEachRemaining(duplicateAliasesIndices::add);
+            }
+            duplicateAliasesIndices.retainAll(allIndices);
+            if (duplicateAliasesIndices.isEmpty() == false) {
+                // iterate again and constructs a helpful message
+                ArrayList<String> duplicates = new ArrayList<>();
+                for (ObjectCursor<IndexMetaData> cursor : indices.values()) {
+                    for (String alias: duplicateAliasesIndices) {
+                        if (cursor.value.getAliases().containsKey(alias)) {
+                            duplicates.add(alias + " (alias of " + cursor.value.getIndex() + ")");
+                        }
+                    }
                 }
+                assert duplicates.size() > 0;
+                throw new IllegalStateException("index and alias names need to be unique, but the following duplicates were found ["
+                    + Strings.collectionToCommaDelimitedString(duplicates)+ "]");
+
             }
-            String[] allOpenIndices = allOpenIndicesLst.toArray(new String[allOpenIndicesLst.size()]);
-            String[] allClosedIndices = allClosedIndicesLst.toArray(new String[allClosedIndicesLst.size()]);
 
             // build all indices map
             SortedMap<String, AliasOrIndex> aliasAndIndexLookup = new TreeMap<>();
             for (ObjectCursor<IndexMetaData> cursor : indices.values()) {
                 IndexMetaData indexMetaData = cursor.value;
-                aliasAndIndexLookup.put(indexMetaData.getIndex().getName(), new AliasOrIndex.Index(indexMetaData));
+                AliasOrIndex existing = aliasAndIndexLookup.put(indexMetaData.getIndex().getName(), new AliasOrIndex.Index(indexMetaData));
+                assert existing == null : "duplicate for " + indexMetaData.getIndex();
 
                 for (ObjectObjectCursor<String, AliasMetaData> aliasCursor : indexMetaData.getAliases()) {
                     AliasMetaData aliasMetaData = aliasCursor.value;
-                    AliasOrIndex aliasOrIndex = aliasAndIndexLookup.get(aliasMetaData.getAlias());
-                    if (aliasOrIndex == null) {
-                        aliasOrIndex = new AliasOrIndex.Alias(aliasMetaData, indexMetaData);
-                        aliasAndIndexLookup.put(aliasMetaData.getAlias(), aliasOrIndex);
-                    } else if (aliasOrIndex instanceof AliasOrIndex.Alias) {
-                        AliasOrIndex.Alias alias = (AliasOrIndex.Alias) aliasOrIndex;
-                        alias.addIndex(indexMetaData);
-                    } else if (aliasOrIndex instanceof AliasOrIndex.Index) {
-                        AliasOrIndex.Index index = (AliasOrIndex.Index) aliasOrIndex;
-                        throw new IllegalStateException("index and alias names need to be unique, but alias [" + aliasMetaData.getAlias() + "] and index " + index.getIndex().getIndex() + " have the same name");
-                    } else {
-                        throw new IllegalStateException("unexpected alias [" + aliasMetaData.getAlias() + "][" + aliasOrIndex + "]");
-                    }
+                    aliasAndIndexLookup.compute(aliasMetaData.getAlias(), (aliasName, alias) -> {
+                        if (alias == null) {
+                            return new AliasOrIndex.Alias(aliasMetaData, indexMetaData);
+                        } else {
+                            assert alias instanceof AliasOrIndex.Alias : alias.getClass().getName();
+                            ((AliasOrIndex.Alias) alias).addIndex(indexMetaData);
+                            return alias;
+                        }
+                    });
                 }
             }
             aliasAndIndexLookup = Collections.unmodifiableSortedMap(aliasAndIndexLookup);
+            // build all concrete indices arrays:
+            // TODO: I think we can remove these arrays. it isn't worth the effort, for operations on all indices.
+            // When doing an operation across all indices, most of the time is spent on actually going to all shards and
+            // do the required operations, the bottleneck isn't resolving expressions into concrete indices.
+            String[] allIndicesArray = allIndices.toArray(new String[allIndices.size()]);
+            String[] allOpenIndicesArray = allOpenIndices.toArray(new String[allOpenIndices.size()]);
+            String[] allClosedIndicesArray = allClosedIndices.toArray(new String[allClosedIndices.size()]);
+
             return new MetaData(clusterUUID, version, transientSettings, persistentSettings, indices.build(), templates.build(),
-                                customs.build(), allIndices, allOpenIndices, allClosedIndices, aliasAndIndexLookup);
+                                customs.build(), allIndicesArray, allOpenIndicesArray, allClosedIndicesArray, aliasAndIndexLookup);
         }
 
         public static String toXContent(MetaData metaData) throws IOException {

core/src/test/java/org/elasticsearch/cluster/metadata/MetaDataTests.java

@@ -26,7 +26,6 @@ import org.elasticsearch.common.bytes.BytesReference;
 import org.elasticsearch.common.io.stream.BytesStreamOutput;
 import org.elasticsearch.common.io.stream.NamedWriteableAwareStreamInput;
 import org.elasticsearch.common.io.stream.NamedWriteableRegistry;
-import org.elasticsearch.common.io.stream.StreamInput;
 import org.elasticsearch.common.settings.Settings;
 import org.elasticsearch.common.xcontent.ToXContent;
 import org.elasticsearch.common.xcontent.XContentBuilder;
@@ -36,9 +35,14 @@ import org.elasticsearch.index.Index;
 import org.elasticsearch.test.ESTestCase;
 
 import java.io.IOException;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.Map;
+import java.util.Set;
 
 import static org.hamcrest.Matchers.equalTo;
 import static org.hamcrest.Matchers.is;
+import static org.hamcrest.Matchers.startsWith;
 
 public class MetaDataTests extends ESTestCase {
 
@@ -52,7 +56,42 @@ public class MetaDataTests extends ESTestCase {
             MetaData.builder().put(builder).build();
             fail("exception should have been thrown");
         } catch (IllegalStateException e) {
-            assertThat(e.getMessage(), equalTo("index and alias names need to be unique, but alias [index] and index [index] have the same name"));
+            assertThat(e.getMessage(), equalTo("index and alias names need to be unique, but the following duplicates were found [index (alias of [index])]"));
+        }
+    }
+
+    public void testAliasCollidingWithAnExistingIndex() {
+        int indexCount = randomIntBetween(10, 100);
+        Set<String> indices = new HashSet<>(indexCount);
+        for (int i = 0; i < indexCount; i++) {
+            indices.add(randomAlphaOfLength(10));
+        }
+        Map<String, Set<String>> aliasToIndices = new HashMap<>();
+        for (String alias: randomSubsetOf(randomIntBetween(1, 10), indices)) {
+            aliasToIndices.put(alias, new HashSet<>(randomSubsetOf(randomIntBetween(1, 3), indices)));
+        }
+        int properAliases = randomIntBetween(0, 3);
+        for (int i = 0; i < properAliases; i++) {
+            aliasToIndices.put(randomAlphaOfLength(5), new HashSet<>(randomSubsetOf(randomIntBetween(1, 3), indices)));
+        }
+        MetaData.Builder metaDataBuilder = MetaData.builder();
+        for (String index : indices) {
+            IndexMetaData.Builder indexBuilder = IndexMetaData.builder(index)
+                .settings(Settings.builder().put(IndexMetaData.SETTING_VERSION_CREATED, Version.CURRENT))
+                .numberOfShards(1)
+                .numberOfReplicas(0);
+            aliasToIndices.forEach((key, value) -> {
+                if (value.contains(index)) {
+                    indexBuilder.putAlias(AliasMetaData.builder(key).build());
+                }
+            });
+            metaDataBuilder.put(indexBuilder);
+        }
+        try {
+            metaDataBuilder.build();
+            fail("exception should have been thrown");
+        } catch (IllegalStateException e) {
+            assertThat(e.getMessage(), startsWith("index and alias names need to be unique"));
         }
     }