Return 429 status code on read_only_allow_delete index block (#50166)

We consider index level read_only_allow_delete blocks temporary since
the DiskThresholdMonitor can automatically release those when an index
is no longer allocated on nodes above high threshold.

The rest status has therefore been changed to 429 when encountering this
index block to signal retryability to clients.

Related to #49393

Author: gaobinlong

modules/reindex/src/test/java/org/elasticsearch/index/reindex/DeleteByQueryBasicTests.java

@@ -21,12 +21,18 @@
 
 import org.elasticsearch.action.admin.indices.alias.Alias;
 import org.elasticsearch.action.index.IndexRequestBuilder;
+import org.elasticsearch.cluster.ClusterInfoService;
+import org.elasticsearch.cluster.InternalClusterInfoService;
+import org.elasticsearch.cluster.routing.allocation.DiskThresholdSettings;
 import org.elasticsearch.common.settings.Settings;
+import org.elasticsearch.common.unit.TimeValue;
 import org.elasticsearch.index.IndexNotFoundException;
 import org.elasticsearch.index.query.QueryBuilders;
 import org.elasticsearch.plugins.Plugin;
 import org.elasticsearch.search.sort.SortOrder;
 import org.elasticsearch.test.InternalSettingsPlugin;
+import org.elasticsearch.test.InternalTestCluster;
+import org.elasticsearch.threadpool.ThreadPool;
 
 import java.util.ArrayList;
 import java.util.Collection;
@@ -214,18 +220,65 @@ public void testDeleteByQueryOnReadOnlyIndex() throws Exception {
         }
         indexRandom(true, true, true, builders);
 
-        String block = randomFrom(SETTING_READ_ONLY, SETTING_READ_ONLY_ALLOW_DELETE);
         try {
-            enableIndexBlock("test", block);
+            enableIndexBlock("test", SETTING_READ_ONLY);
             assertThat(deleteByQuery().source("test").filter(QueryBuilders.matchAllQuery()).refresh(true).get(),
-                    matcher().deleted(0).failures(docs));
+                matcher().deleted(0).failures(docs));
         } finally {
-            disableIndexBlock("test", block);
+            disableIndexBlock("test", SETTING_READ_ONLY);
         }
 
         assertHitCount(client().prepareSearch("test").setSize(0).get(), docs);
     }
 
+    public void testDeleteByQueryOnReadOnlyAllowDeleteIndex() throws Exception {
+        createIndex("test");
+
+        final int docs = randomIntBetween(1, 50);
+        List<IndexRequestBuilder> builders = new ArrayList<>();
+        for (int i = 0; i < docs; i++) {
+            builders.add(client().prepareIndex("test").setId(Integer.toString(i)).setSource("field", 1));
+        }
+        indexRandom(true, true, true, builders);
+
+        // Because the index level read_only_allow_delete block can be automatically released by disk allocation decider,
+        // so we should test both case of disk allocation decider is enabled and disabled
+        boolean diskAllocationDeciderEnabled = randomBoolean();
+        try {
+            // When a read_only_allow_delete block is set on the index,
+            // it will trigger a retry policy in the delete by query request because the rest status of the block is 429
+            enableIndexBlock("test", SETTING_READ_ONLY_ALLOW_DELETE);
+            if (diskAllocationDeciderEnabled) {
+                InternalTestCluster internalTestCluster = internalCluster();
+                InternalClusterInfoService infoService = (InternalClusterInfoService) internalTestCluster
+                    .getInstance(ClusterInfoService.class, internalTestCluster.getMasterName());
+                ThreadPool threadPool = internalTestCluster.getInstance(ThreadPool.class, internalTestCluster.getMasterName());
+                // Refresh the cluster info after a random delay to check the disk threshold and release the block on the index
+                threadPool.schedule(infoService::refresh, TimeValue.timeValueMillis(randomIntBetween(1, 100)), ThreadPool.Names.MANAGEMENT);
+                // The delete by query request will be executed successfully because the block will be released
+                assertThat(deleteByQuery().source("test").filter(QueryBuilders.matchAllQuery()).refresh(true).get(),
+                    matcher().deleted(docs));
+            } else {
+                // Disable the disk allocation decider to ensure the read_only_allow_delete block cannot be released
+                setDiskAllocationDeciderEnabled(false);
+                // The delete by query request will not be executed successfully because the block cannot be released
+                assertThat(deleteByQuery().source("test").filter(QueryBuilders.matchAllQuery()).refresh(true)
+                        .setMaxRetries(2).setRetryBackoffInitialTime(TimeValue.timeValueMillis(50)).get(),
+                    matcher().deleted(0).failures(docs));
+            }
+        } finally {
+            disableIndexBlock("test", SETTING_READ_ONLY_ALLOW_DELETE);
+            if (diskAllocationDeciderEnabled == false) {
+                setDiskAllocationDeciderEnabled(true);
+            }
+        }
+        if (diskAllocationDeciderEnabled) {
+            assertHitCount(client().prepareSearch("test").setSize(0).get(), 0);
+        } else {
+            assertHitCount(client().prepareSearch("test").setSize(0).get(), docs);
+        }
+    }
+
     public void testSlices() throws Exception {
         indexRandom(true,
                 client().prepareIndex("test").setId("1").setSource("foo", "a"),
@@ -315,4 +368,12 @@ public void testMissingSources() {
         assertThat(response, matcher().deleted(0).slices(hasSize(0)));
     }
 
+    /** Enables or disables the cluster disk allocation decider **/
+    private void setDiskAllocationDeciderEnabled(boolean value) {
+        Settings settings = value ? Settings.builder().putNull(
+            DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_DISK_THRESHOLD_ENABLED_SETTING.getKey()).build() :
+            Settings.builder().put(
+                DiskThresholdSettings.CLUSTER_ROUTING_ALLOCATION_DISK_THRESHOLD_ENABLED_SETTING.getKey(), value).build();
+        assertAcked(client().admin().cluster().prepareUpdateSettings().setTransientSettings(settings).get());
+    }
 }

server/src/main/java/org/elasticsearch/cluster/block/ClusterBlockException.java

@@ -110,13 +110,23 @@ private static String buildMessageForIndexBlocks(Map<String, Set<ClusterBlock>>
     @Override
     public RestStatus status() {
         RestStatus status = null;
+        boolean onlyRetryableBlocks = true;
         for (ClusterBlock block : blocks) {
-            if (status == null) {
-                status = block.status();
-            } else if (status.getStatus() < block.status().getStatus()) {
-                status = block.status();
+            boolean isRetryableBlock = block.status() == RestStatus.TOO_MANY_REQUESTS;
+            if (isRetryableBlock == false) {
+                if (status == null) {
+                    status = block.status();
+                } else if (status.getStatus() < block.status().getStatus()) {
+                    status = block.status();
+                }
             }
+            onlyRetryableBlocks = onlyRetryableBlocks && isRetryableBlock;
         }
+        // return retryable status if there are only retryable blocks
+        if (onlyRetryableBlocks) {
+            return RestStatus.TOO_MANY_REQUESTS;
+        }
+        // return status which has the maximum code of all status except the retryable blocks'
         return status;
     }
 }

server/src/main/java/org/elasticsearch/cluster/metadata/IndexMetaData.java

@@ -95,7 +95,7 @@ public class IndexMetaData implements Diffable<IndexMetaData>, ToXContentFragmen
             RestStatus.FORBIDDEN, EnumSet.of(ClusterBlockLevel.METADATA_WRITE, ClusterBlockLevel.METADATA_READ));
     public static final ClusterBlock INDEX_READ_ONLY_ALLOW_DELETE_BLOCK =
         new ClusterBlock(12, "index read-only / allow delete (api)", false, false,
-            true, RestStatus.FORBIDDEN, EnumSet.of(ClusterBlockLevel.METADATA_WRITE, ClusterBlockLevel.WRITE));
+            true, RestStatus.TOO_MANY_REQUESTS, EnumSet.of(ClusterBlockLevel.METADATA_WRITE, ClusterBlockLevel.WRITE));
 
     public enum State {
         OPEN((byte) 0),

server/src/test/java/org/elasticsearch/action/admin/indices/delete/DeleteIndexBlocksIT.java

@@ -72,7 +72,7 @@ public void testClusterBlockMessageHasIndexName() {
             client().admin().indices().prepareUpdateSettings("test").setSettings(settings).get();
             ClusterBlockException e = expectThrows(ClusterBlockException.class, () ->
                 client().prepareIndex().setIndex("test").setId("1").setSource("foo", "bar").get());
-            assertEquals("index [test] blocked by: [FORBIDDEN/12/index read-only / allow delete (api)];", e.getMessage());
+            assertEquals("index [test] blocked by: [TOO_MANY_REQUESTS/12/index read-only / allow delete (api)];", e.getMessage());
         } finally {
             assertAcked(client().admin().indices().prepareUpdateSettings("test")
                 .setSettings(Settings.builder().putNull(IndexMetaData.SETTING_READ_ONLY_ALLOW_DELETE).build()).get());

test/framework/src/main/java/org/elasticsearch/test/hamcrest/ElasticsearchAssertions.java

@@ -43,6 +43,7 @@
 import org.elasticsearch.action.support.master.AcknowledgedResponse;
 import org.elasticsearch.cluster.block.ClusterBlock;
 import org.elasticsearch.cluster.block.ClusterBlockException;
+import org.elasticsearch.cluster.metadata.IndexMetaData;
 import org.elasticsearch.cluster.metadata.IndexTemplateMetaData;
 import org.elasticsearch.common.Nullable;
 import org.elasticsearch.common.bytes.BytesReference;
@@ -150,7 +151,9 @@ public static void assertBlocked(BroadcastResponse replicatedBroadcastResponse)
             assertNotNull("expected the cause of failure to be a ClusterBlockException but got " + exception.getCause().getMessage(),
                     clusterBlockException);
             assertThat(clusterBlockException.blocks().size(), greaterThan(0));
-            assertThat(clusterBlockException.status(), CoreMatchers.equalTo(RestStatus.FORBIDDEN));
+
+            RestStatus status = checkRetryableBlock(clusterBlockException.blocks()) ? RestStatus.TOO_MANY_REQUESTS : RestStatus.FORBIDDEN;
+            assertThat(clusterBlockException.status(), CoreMatchers.equalTo(status));
         }
     }
 
@@ -166,7 +169,8 @@ public static void assertBlocked(final ActionRequestBuilder builder, @Nullable f
             fail("Request executed with success but a ClusterBlockException was expected");
         } catch (ClusterBlockException e) {
             assertThat(e.blocks().size(), greaterThan(0));
-            assertThat(e.status(), equalTo(RestStatus.FORBIDDEN));
+            RestStatus status = checkRetryableBlock(e.blocks()) ? RestStatus.TOO_MANY_REQUESTS : RestStatus.FORBIDDEN;
+            assertThat(e.status(), equalTo(status));
 
             if (expectedBlockId != null) {
                 boolean found = false;
@@ -191,6 +195,16 @@ public static void assertBlocked(final ActionRequestBuilder builder, @Nullable f
         assertBlocked(builder, expectedBlock != null ? expectedBlock.id() : null);
     }
 
+    private static boolean checkRetryableBlock(Set<ClusterBlock> clusterBlocks){
+        // check only retryable blocks exist in the set
+        for (ClusterBlock clusterBlock : clusterBlocks) {
+            if (clusterBlock.id() != IndexMetaData.INDEX_READ_ONLY_ALLOW_DELETE_BLOCK.id()) {
+                return false;
+            }
+        }
+        return true;
+    }
+
     public static String formatShardStatus(BroadcastResponse response) {
         StringBuilder msg = new StringBuilder();
         msg.append(" Total shards: ").append(response.getTotalShards())

test/framework/src/test/java/org/elasticsearch/test/hamcrest/ElasticsearchAssertionsTests.java

@@ -19,6 +19,11 @@
 
 package org.elasticsearch.test.hamcrest;
 
+import org.elasticsearch.action.support.DefaultShardOperationFailedException;
+import org.elasticsearch.action.support.broadcast.BroadcastResponse;
+import org.elasticsearch.cluster.block.ClusterBlock;
+import org.elasticsearch.cluster.block.ClusterBlockException;
+import org.elasticsearch.cluster.metadata.IndexMetaData;
 import org.elasticsearch.common.bytes.BytesReference;
 import org.elasticsearch.common.xcontent.XContentBuilder;
 import org.elasticsearch.common.xcontent.XContentParser;
@@ -27,7 +32,12 @@
 import org.elasticsearch.test.RandomObjects;
 
 import java.io.IOException;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.Set;
 
+import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertBlocked;
 import static org.elasticsearch.test.hamcrest.ElasticsearchAssertions.assertToXContentEquivalent;
 import static org.hamcrest.Matchers.containsString;
 
@@ -188,4 +198,24 @@ public void testAssertXContentEquivalentErrors() throws IOException {
             assertThat(error.getMessage(), containsString("expected [1] more entries"));
         }
     }
+
+    public void testAssertBlocked() {
+        Map<String, Set<ClusterBlock>> indexLevelBlocks = new HashMap<>();
+
+        indexLevelBlocks.put("test", Set.of(IndexMetaData.INDEX_READ_ONLY_BLOCK));
+        assertBlocked(new BroadcastResponse(1, 0, 1, List.of(new DefaultShardOperationFailedException("test", 0,
+            new ClusterBlockException(indexLevelBlocks)))));
+
+        indexLevelBlocks.put("test", Set.of(IndexMetaData.INDEX_READ_ONLY_ALLOW_DELETE_BLOCK));
+        assertBlocked(new BroadcastResponse(1, 0, 1, List.of(new DefaultShardOperationFailedException("test", 0,
+            new ClusterBlockException(indexLevelBlocks)))));
+
+        indexLevelBlocks.put("test", Set.of(IndexMetaData.INDEX_READ_BLOCK, IndexMetaData.INDEX_METADATA_BLOCK));
+        assertBlocked(new BroadcastResponse(1, 0, 1, List.of(new DefaultShardOperationFailedException("test", 0,
+            new ClusterBlockException(indexLevelBlocks)))));
+
+        indexLevelBlocks.put("test", Set.of(IndexMetaData.INDEX_READ_ONLY_BLOCK, IndexMetaData.INDEX_READ_ONLY_ALLOW_DELETE_BLOCK));
+        assertBlocked(new BroadcastResponse(1, 0, 1, List.of(new DefaultShardOperationFailedException("test", 0,
+            new ClusterBlockException(indexLevelBlocks)))));
+    }
 }