SOLR-3369: shards.tolerant=true is broken for group queries

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1498992 13f79535-47bb-0310-9956-ffa450edef68

solr/CHANGES.txt

@@ -237,6 +237,9 @@ Bug Fixes
 * SOLR-4974: Outgrowth of SOLR-4960 that includes transient cores and pending cores
   (Erick Erickson)
 
+* SOLR-3369: shards.tolerant=true is broken for group queries
+  (Russell Black, Martijn van Groningen, Jabouille jean Charles, Ryan McKinley via shalin)
+
 Optimizations
 ----------------------
 

solr/core/src/java/org/apache/solr/search/grouping/distributed/responseprocessor/SearchGroupShardResponseProcessor.java

@@ -20,8 +20,11 @@ package org.apache.solr.search.grouping.distributed.responseprocessor;
 import org.apache.lucene.search.Sort;
 import org.apache.lucene.search.grouping.SearchGroup;
 import org.apache.lucene.util.BytesRef;
+import org.apache.solr.client.solrj.SolrServerException;
 import org.apache.solr.common.SolrException;
+import org.apache.solr.common.params.ShardParams;
 import org.apache.solr.common.util.NamedList;
+import org.apache.solr.common.util.SimpleOrderedMap;
 import org.apache.solr.handler.component.ResponseBuilder;
 import org.apache.solr.handler.component.ShardRequest;
 import org.apache.solr.handler.component.ShardResponse;
@@ -31,6 +34,8 @@ import org.apache.solr.search.grouping.distributed.command.Pair;
 import org.apache.solr.search.grouping.distributed.shardresultserializer.SearchGroupsResultTransformer;
 
 import java.io.IOException;
+import java.io.PrintWriter;
+import java.io.StringWriter;
 import java.util.*;
 
 /**
@@ -61,7 +66,38 @@ public class SearchGroupShardResponseProcessor implements ShardResponseProcessor
     try {
       int maxElapsedTime = 0;
       int hitCountDuringFirstPhase = 0;
+
+      NamedList<Object> shardInfo = null;
+      if (rb.req.getParams().getBool(ShardParams.SHARDS_INFO, false)) {
+        shardInfo = new SimpleOrderedMap<Object>();
+        rb.rsp.getValues().add(ShardParams.SHARDS_INFO + ".firstPhase", shardInfo);
+      }
+
       for (ShardResponse srsp : shardRequest.responses) {
+        if (shardInfo != null) {
+          SimpleOrderedMap<Object> nl = new SimpleOrderedMap<Object>();
+
+          if (srsp.getException() != null) {
+            Throwable t = srsp.getException();
+            if (t instanceof SolrServerException) {
+              t = ((SolrServerException) t).getCause();
+            }
+            nl.add("error", t.toString());
+            StringWriter trace = new StringWriter();
+            t.printStackTrace(new PrintWriter(trace));
+            nl.add("trace", trace.toString());
+          } else {
+            nl.add("numFound", (Integer) srsp.getSolrResponse().getResponse().get("totalHitCount"));
+          }
+          if (srsp.getSolrResponse() != null) {
+            nl.add("time", srsp.getSolrResponse().getElapsedTime());
+          }
+
+          shardInfo.add(srsp.getShard(), nl);
+        }
+        if (rb.req.getParams().getBool(ShardParams.SHARDS_TOLERANT, false) && srsp.getException() != null) {
+          continue; // continue if there was an error and we're tolerant.  
+        }
         maxElapsedTime = (int) Math.max(maxElapsedTime, srsp.getSolrResponse().getElapsedTime());
         @SuppressWarnings("unchecked")
         NamedList<NamedList> firstPhaseResult = (NamedList<NamedList>) srsp.getSolrResponse().getResponse().get("firstPhase");

solr/core/src/java/org/apache/solr/search/grouping/distributed/responseprocessor/TopGroupsShardResponseProcessor.java

@@ -23,8 +23,11 @@ import org.apache.lucene.search.TopDocs;
 import org.apache.lucene.search.grouping.GroupDocs;
 import org.apache.lucene.search.grouping.TopGroups;
 import org.apache.lucene.util.BytesRef;
+import org.apache.solr.client.solrj.SolrServerException;
 import org.apache.solr.common.SolrException;
+import org.apache.solr.common.params.ShardParams;
 import org.apache.solr.common.util.NamedList;
+import org.apache.solr.common.util.SimpleOrderedMap;
 import org.apache.solr.handler.component.ResponseBuilder;
 import org.apache.solr.handler.component.ShardDoc;
 import org.apache.solr.handler.component.ShardRequest;
@@ -35,6 +38,8 @@ import org.apache.solr.search.grouping.distributed.command.QueryCommandResult;
 import org.apache.solr.search.grouping.distributed.shardresultserializer.TopGroupsResultTransformer;
 
 import java.io.IOException;
+import java.io.PrintWriter;
+import java.io.StringWriter;
 import java.util.ArrayList;
 import java.util.HashMap;
 import java.util.List;
@@ -76,18 +81,66 @@ public class TopGroupsShardResponseProcessor implements ShardResponseProcessor {
     }
 
     TopGroupsResultTransformer serializer = new TopGroupsResultTransformer(rb);
+
+    NamedList<Object> shardInfo = null;
+    if (rb.req.getParams().getBool(ShardParams.SHARDS_INFO, false)) {
+      shardInfo = new SimpleOrderedMap<Object>();
+      rb.rsp.getValues().add(ShardParams.SHARDS_INFO, shardInfo);
+    }
+
     for (ShardResponse srsp : shardRequest.responses) {
+      SimpleOrderedMap<Object> individualShardInfo = null;
+      if (shardInfo != null) {
+        individualShardInfo = new SimpleOrderedMap<Object>();
+
+        if (srsp.getException() != null) {
+          Throwable t = srsp.getException();
+          if (t instanceof SolrServerException) {
+            t = ((SolrServerException) t).getCause();
+          }
+          individualShardInfo.add("error", t.toString());
+          StringWriter trace = new StringWriter();
+          t.printStackTrace(new PrintWriter(trace));
+          individualShardInfo.add("trace", trace.toString());
+        } else {
+          // summary for successful shard response is added down below
+        }
+        if (srsp.getSolrResponse() != null) {
+          individualShardInfo.add("time", srsp.getSolrResponse().getElapsedTime());
+        }
+
+        shardInfo.add(srsp.getShard(), individualShardInfo);
+      }
+      if (rb.req.getParams().getBool(ShardParams.SHARDS_TOLERANT, false) && srsp.getException() != null) {
+        continue; // continue if there was an error and we're tolerant.  
+      }
       NamedList<NamedList> secondPhaseResult = (NamedList<NamedList>) srsp.getSolrResponse().getResponse().get("secondPhase");
       Map<String, ?> result = serializer.transformToNative(secondPhaseResult, groupSort, sortWithinGroup, srsp.getShard());
+      int numFound = 0;
+      float maxScore = Float.NaN;
       for (String field : commandTopGroups.keySet()) {
         TopGroups<BytesRef> topGroups = (TopGroups<BytesRef>) result.get(field);
         if (topGroups == null) {
           continue;
         }
+        if (individualShardInfo != null) { // keep track of this when shards.info=true
+          numFound += topGroups.totalHitCount;
+          if (Float.isNaN(maxScore) || topGroups.maxScore > maxScore) maxScore = topGroups.maxScore;
+        }
         commandTopGroups.get(field).add(topGroups);
       }
       for (String query : queries) {
-        commandTopDocs.get(query).add((QueryCommandResult) result.get(query));
+        QueryCommandResult queryCommandResult = (QueryCommandResult) result.get(query);
+        if (individualShardInfo != null) { // keep track of this when shards.info=true
+          numFound += queryCommandResult.getMatches();
+          float thisMax = queryCommandResult.getTopDocs().getMaxScore();
+          if (Float.isNaN(maxScore) || thisMax > maxScore) maxScore = thisMax;
+        }
+        commandTopDocs.get(query).add(queryCommandResult);
+      }
+      if (individualShardInfo != null) { // when shards.info=true
+        individualShardInfo.add("numFound", numFound);
+        individualShardInfo.add("maxScore", maxScore);
       }
     }
     try {

solr/core/src/test/org/apache/solr/TestDistributedSearch.java

@@ -360,6 +360,26 @@ public class TestDistributedSearch extends BaseDistributedSearchTestCase {
           ShardParams.SHARDS_INFO,"true",
           ShardParams.SHARDS_TOLERANT,"true");
 
+      queryPartialResults(upShards, upClients,
+          "q", "*:*",
+          "facet", "true",
+          "facet.query", i1 + ":[1 TO 50]",
+          ShardParams.SHARDS_INFO, "true",
+          ShardParams.SHARDS_TOLERANT, "true");
+
+      // test group query
+      queryPartialResults(upShards, upClients,
+          "q", "*:*",
+          "rows", 100,
+          "fl", "id," + i1,
+          "group", "true",
+          "group.query", t1 + ":kings OR " + t1 + ":eggs",
+          "group.limit", 10,
+          "sort", i1 + " asc, id asc",
+          CommonParams.TIME_ALLOWED, 1,
+          ShardParams.SHARDS_INFO, "true",
+          ShardParams.SHARDS_TOLERANT, "true");
+
       // restart the jettys
       for (JettySolrRunner downJetty : downJettys) {
         downJetty.start();