Merge pull request #26 from unbxd/SSK-2764

[SSK-2764/SOLR-10720] fix: patch clusterstatus
unbxd · Aug 11, 2021 · f84e530 · f84e530
2 parents a68c6e1 + 7fb6b06
commit f84e530
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 37 deletions.
diff --git a/solr/core/src/java/org/apache/solr/cloud/OverseerCollectionMessageHandler.java b/solr/core/src/java/org/apache/solr/cloud/OverseerCollectionMessageHandler.java
@@ -24,7 +24,6 @@
 import java.util.Collection;
 import java.util.Collections;
 import java.util.HashMap;
-import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
 import java.util.Random;
@@ -344,38 +343,6 @@ private void balanceProperty(ClusterState clusterState, ZkNodeProps message, Nam
     inQueue.offer(Utils.toJSON(new ZkNodeProps(propMap)));
   }
 
-  /**
-   * Walks the tree of collection status to verify that any replicas not reporting a "down" status is
-   * on a live node, if any replicas reporting their status as "active" but the node is not live is
-   * marked as "down"; used by CLUSTERSTATUS.
-   * @param liveNodes List of currently live node names.
-   * @param collectionProps Map of collection status information pulled directly from ZooKeeper.
-   */
-
-  @SuppressWarnings("unchecked")
-  protected void crossCheckReplicaStateWithLiveNodes(List<String> liveNodes, NamedList<Object> collectionProps) {
-    Iterator<Map.Entry<String,Object>> colls = collectionProps.iterator();
-    while (colls.hasNext()) {
-      Map.Entry<String,Object> next = colls.next();
-      Map<String,Object> collMap = (Map<String,Object>)next.getValue();
-      Map<String,Object> shards = (Map<String,Object>)collMap.get("shards");
-      for (Object nextShard : shards.values()) {
-        Map<String,Object> shardMap = (Map<String,Object>)nextShard;
-        Map<String,Object> replicas = (Map<String,Object>)shardMap.get("replicas");
-        for (Object nextReplica : replicas.values()) {
-          Map<String,Object> replicaMap = (Map<String,Object>)nextReplica;
-          if (Replica.State.getState((String) replicaMap.get(ZkStateReader.STATE_PROP)) != Replica.State.DOWN) {
-            // not down, so verify the node is live
-            String node_name = (String)replicaMap.get(ZkStateReader.NODE_NAME_PROP);
-            if (!liveNodes.contains(node_name)) {
-              // node is not live, so this replica is actually down
-              replicaMap.put(ZkStateReader.STATE_PROP, Replica.State.DOWN.toString());
-            }
-          }
-        }
-      }
-    }
-  }
 
   /**
    * Get collection status from cluster state.

diff --git a/solr/core/src/java/org/apache/solr/handler/admin/ClusterStatus.java b/solr/core/src/java/org/apache/solr/handler/admin/ClusterStatus.java
@@ -138,10 +138,19 @@ public  void getClusterStatus(NamedList results)
       if (collectionVsAliases.containsKey(name) && !collectionVsAliases.get(name).isEmpty()) {
         collectionStatus.put("aliases", collectionVsAliases.get(name));
       }
-      String configName = zkStateReader.readConfigName(name);
-      collectionStatus.put("configName", configName);
-      collectionProps.add(name, collectionStatus);
-    }
+
+    try {
+        String configName = zkStateReader.readConfigName(name);
+        collectionStatus.put("configName", configName);
+        collectionProps.add(name, collectionStatus);
+      } catch (SolrException e) {
+        if (e.getCause() instanceof KeeperException.NoNodeException)  {
+          // skip this collection because the collection's znode has been deleted
+          // which can happen during aggressive collection removal, see SOLR-10720
+        } else throw e;
+      } 
+
+   }
 
     List<String> liveNodes = zkStateReader.getZkClient().getChildren(ZkStateReader.LIVE_NODES_ZKNODE, null, true);