[8.x] Non existing synonyms sets do not fail shard recovery (#125659) (#125795)

carlosdelest · web-flow · commit af4a5486dbb4 · 2025-03-28T09:25:23.000+01:00
diff --git a/docs/changelog/125659.yaml b/docs/changelog/125659.yaml
@@ -0,0 +1,6 @@
+pr: 125659
+summary: Non existing synonyms sets do not fail shard recovery for indices
+area: "Analysis"
+type: bug
+issues:
+ - 125603
diff --git a/modules/analysis-common/src/main/java/org/elasticsearch/analysis/common/SynonymTokenFilterFactory.java b/modules/analysis-common/src/main/java/org/elasticsearch/analysis/common/SynonymTokenFilterFactory.java
@@ -74,7 +74,7 @@ public ReaderWithOrigin getRulesReader(SynonymTokenFilterFactory factory, IndexC
                     );
                 } else {
                     reader = new ReaderWithOrigin(
-                        Analysis.getReaderFromIndex(synonymsSet, factory.synonymsManagementAPIService),
+                        Analysis.getReaderFromIndex(synonymsSet, factory.synonymsManagementAPIService, factory.lenient),
                         "[" + synonymsSet + "] synonyms_set in .synonyms index",
                         synonymsSet
                     );
diff --git a/qa/mixed-cluster/build.gradle b/qa/mixed-cluster/build.gradle
@@ -64,9 +64,6 @@ excludeList.add('cluster.desired_nodes/20_dry_run/Test validation works for dry
 // Excluded because they create dot-prefixed indices on older versions
 excludeList.add('indices.resolve_index/20_resolve_system_index/*')
 
-// Can't work until auto-expand replicas is 0-1 for synonyms index
-excludeList.add("synonyms/90_synonyms_reloading_for_synset/Reload analyzers for specific synonym set")
-
 def clusterPath = getPath()
 
 buildParams.bwcVersions.withWireCompatible { bwcVersion, baseName ->
diff --git a/rest-api-spec/build.gradle b/rest-api-spec/build.gradle
@@ -282,4 +282,5 @@ tasks.named("yamlRestTestV7CompatTransform").configure({ task ->
   task.skipTest("indices.create/21_synthetic_source_stored/field param - keep root array", "Synthetic source keep arrays now stores leaf arrays natively")
   task.skipTest("migration/10_get_feature_upgrade_status/Get feature upgrade status", "Moved to plugin")
   task.skipTest("migration/20_post_feature_upgrade/Get feature upgrade status", "Moved to plugin")
+  task.skipTest("synonyms/80_synonyms_from_index/Fail loading synonyms from index if synonyms_set doesn't exist", "Synonyms do no longer fail if the synonyms_set doesn't exist")
 })
diff --git a/rest-api-spec/src/yamlRestTest/resources/rest-api-spec/test/synonyms/110_synonyms_invalid.yml b/rest-api-spec/src/yamlRestTest/resources/rest-api-spec/test/synonyms/110_synonyms_invalid.yml
@@ -315,3 +315,160 @@ setup:
       indices.stats: { index: test_index }
 
   - length: { indices: 0 }
+
+---
+"Load index with non existent synonyms set":
+  - requires:
+      cluster_features: [ index.synonyms_set_lenient_on_non_existing ]
+      reason: "requires synonyms_set_lenient_on_non_existing bug fix"
+  - do:
+      indices.create:
+        index: test_index
+        body:
+          settings:
+            index:
+              number_of_shards: 1
+              number_of_replicas: 0
+            analysis:
+              filter:
+                my_synonym_filter:
+                  type: synonym
+                  synonyms_set: set1
+                  updateable: true
+              analyzer:
+                my_analyzer:
+                  type: custom
+                  tokenizer: whitespace
+                  filter: [ lowercase, my_synonym_filter ]
+          mappings:
+            properties:
+              my_field:
+                type: text
+                search_analyzer: my_analyzer
+
+  - match: { acknowledged: true }
+  - match: { shards_acknowledged: true }
+
+  - do:
+      indices.stats: { index: test_index }
+
+  - match: { indices.test_index.health: "green" }
+
+  # Synonyms are not applied
+  - do:
+      indices.analyze:
+        index: test_index
+        body:
+          analyzer: my_analyzer
+          text: foo
+
+  - length: { tokens: 1 }
+  - match: { tokens.0.token: foo }
+
+
+  # Create synonyms set and check synonyms are applied
+  - do:
+      synonyms.put_synonym:
+        id: set1
+        body:
+          synonyms_set:
+            synonyms: "foo => bar, baz"
+
+  # This is to ensure that all index shards (write and read) are available. In serverless this can take some time.
+  - do:
+      cluster.health:
+        index: .synonyms
+        wait_for_status: green
+
+
+  - do:
+      indices.stats: { index: test_index }
+
+  - match: { indices.test_index.health: "green" }
+
+  # Synonyms are applied
+  - do:
+      indices.analyze:
+        index: test_index
+        body:
+          analyzer: my_analyzer
+          text: foo
+
+  - length: { tokens: 2 }
+
+---
+"Load index with non existent synonyms set and lenient set to false":
+  - requires:
+      test_runner_features: [ allowed_warnings ]
+
+  - do:
+      indices.create:
+        index: test_index
+        body:
+          settings:
+            index:
+              number_of_shards: 1
+              number_of_replicas: 0
+            analysis:
+              filter:
+                my_synonym_filter:
+                  type: synonym
+                  synonyms_set: set1
+                  updateable: true
+                  lenient: false
+              analyzer:
+                my_analyzer:
+                  type: custom
+                  tokenizer: whitespace
+                  filter: [ lowercase, my_synonym_filter ]
+          mappings:
+            properties:
+              my_field:
+                type: text
+                search_analyzer: my_analyzer
+
+  - match: { acknowledged: true }
+  - match: { shards_acknowledged: false }
+
+  - do:
+      indices.stats: { index: test_index }
+
+  - length: { indices: 0 }
+
+  # Create synonyms set and check synonyms are applied
+  - do:
+      synonyms.put_synonym:
+        id: set1
+        body:
+          synonyms_set:
+            synonyms: "foo => bar, baz"
+
+  # This is to ensure that all index shards (write and read) are available. In serverless this can take some time.
+  - do:
+      cluster.health:
+        index: .synonyms
+        # BWC tests won't be able to get green - synonyms index is 0-all, so some shards won't be recovered in older nodes
+        wait_for_status: yellow
+
+  - do:
+      # Warning issued in previous versions
+      allowed_warnings:
+        - "The [state] field in the response to the reroute API is deprecated and will be removed in a future version. Specify ?metric=none to adopt the future behaviour."
+      cluster.reroute:
+        retry_failed: true
+
+  - do:
+      cluster.health:
+        index: test_index
+        wait_for_status: green
+
+  # Synonyms are applied
+  - do:
+      indices.analyze:
+        index: test_index
+        body:
+          analyzer: my_analyzer
+          text: foo
+
+  - length: { tokens: 2 }
+
diff --git a/rest-api-spec/src/yamlRestTest/resources/rest-api-spec/test/synonyms/80_synonyms_from_index.yml b/rest-api-spec/src/yamlRestTest/resources/rest-api-spec/test/synonyms/80_synonyms_from_index.yml
@@ -168,34 +168,6 @@ setup:
                 query: hola
   - match: { hits.total.value: 1 }
 
----
-"Fail loading synonyms from index if synonyms_set doesn't exist":
-  - do:
-      indices.create:
-        index: another_index
-        body:
-          settings:
-            index:
-              number_of_shards: 1
-            analysis:
-              filter:
-                my_synonym_filter:
-                  type: synonym
-                  synonyms_set: set_missing
-                  updateable: true
-              analyzer:
-                my_analyzer:
-                  type: custom
-                  tokenizer: standard
-                  filter: [ lowercase, my_synonym_filter ]
-          mappings:
-            properties:
-              my_field:
-                type: text
-                search_analyzer: my_analyzer
-  - match: { acknowledged: true }
-  - match: { shards_acknowledged: false }
-
 ---
 "Load empty synonyms set from index for an analyzer":
   - do:
diff --git a/server/src/main/java/org/elasticsearch/index/IndexFeatures.java b/server/src/main/java/org/elasticsearch/index/IndexFeatures.java
@@ -23,8 +23,10 @@ public Set<NodeFeature> getFeatures() {
 
     public static final NodeFeature LOGSDB_NO_HOST_NAME_FIELD = new NodeFeature("index.logsdb_no_host_name_field");
 
+    private static final NodeFeature SYNONYMS_SET_LENIENT_ON_NON_EXISTING = new NodeFeature("index.synonyms_set_lenient_on_non_existing");
+
     @Override
     public Set<NodeFeature> getTestFeatures() {
-        return Set.of(LOGSDB_NO_HOST_NAME_FIELD);
+        return Set.of(LOGSDB_NO_HOST_NAME_FIELD, SYNONYMS_SET_LENIENT_ON_NON_EXISTING);
     }
 }
diff --git a/server/src/main/java/org/elasticsearch/index/analysis/Analysis.java b/server/src/main/java/org/elasticsearch/index/analysis/Analysis.java
@@ -48,6 +48,7 @@
 import org.apache.lucene.analysis.th.ThaiAnalyzer;
 import org.apache.lucene.analysis.tr.TurkishAnalyzer;
 import org.apache.lucene.analysis.util.CSVUtil;
+import org.elasticsearch.ResourceNotFoundException;
 import org.elasticsearch.action.support.PlainActionFuture;
 import org.elasticsearch.common.Strings;
 import org.elasticsearch.common.logging.DeprecationCategory;
@@ -367,10 +368,39 @@ public static Reader getReaderFromFile(Environment env, String filePath, String
         }
     }
 
-    public static Reader getReaderFromIndex(String synonymsSet, SynonymsManagementAPIService synonymsManagementAPIService) {
+    public static Reader getReaderFromIndex(
+        String synonymsSet,
+        SynonymsManagementAPIService synonymsManagementAPIService,
+        boolean ignoreMissing
+    ) {
         final PlainActionFuture<PagedResult<SynonymRule>> synonymsLoadingFuture = new PlainActionFuture<>();
         synonymsManagementAPIService.getSynonymSetRules(synonymsSet, synonymsLoadingFuture);
-        PagedResult<SynonymRule> results = synonymsLoadingFuture.actionGet();
+
+        PagedResult<SynonymRule> results;
+
+        try {
+            results = synonymsLoadingFuture.actionGet();
+        } catch (Exception e) {
+            if (ignoreMissing == false) {
+                throw e;
+            }
+
+            boolean notFound = e instanceof ResourceNotFoundException;
+            String message = String.format(
+                Locale.ROOT,
+                "Synonyms set %s %s. Synonyms will not be applied to search results on indices that use this synonym set",
+                synonymsSet,
+                notFound ? "not found" : "could not be loaded"
+            );
+
+            if (notFound) {
+                logger.warn(message);
+            } else {
+                logger.error(message, e);
+            }
+
+            results = new PagedResult<>(0, new SynonymRule[0]);
+        }
 
         SynonymRule[] synonymRules = results.pageResults();
         StringBuilder sb = new StringBuilder();

Original file line number	Diff line number	Diff line change
`@@ -23,8 +23,10 @@ public Set<NodeFeature> getFeatures() {`
`23`	`23`
`24`	`24`	`public static final NodeFeature LOGSDB_NO_HOST_NAME_FIELD = new NodeFeature("index.logsdb_no_host_name_field");`
`25`	`25`
	`26`	`+ private static final NodeFeature SYNONYMS_SET_LENIENT_ON_NON_EXISTING = new NodeFeature("index.synonyms_set_lenient_on_non_existing");`
	`27`	`+`
`26`	`28`	`@Override`
`27`	`29`	`public Set<NodeFeature> getTestFeatures() {`
`28`		`- return Set.of(LOGSDB_NO_HOST_NAME_FIELD);`
	`30`	`+ return Set.of(LOGSDB_NO_HOST_NAME_FIELD, SYNONYMS_SET_LENIENT_ON_NON_EXISTING);`
`29`	`31`	`}`
`30`	`32`	`}`