Enable top-k optimization only if balanced and non-sparse

3 files changed, 18 insertions, 11 deletions

diff --git a/container-search/src/main/java/com/yahoo/search/dispatch/InterleavedSearchInvoker.java b/container-search/src/main/java/com/yahoo/search/dispatch/InterleavedSearchInvoker.java
index adf7368faa2..fb04c8299e9 100644
--- a/container-search/src/main/java/com/yahoo/search/dispatch/InterleavedSearchInvoker.java
+++ b/container-search/src/main/java/com/yahoo/search/dispatch/InterleavedSearchInvoker.java
@@ -3,6 +3,7 @@ package com.yahoo.search.dispatch;
 
 import com.yahoo.search.Query;
 import com.yahoo.search.Result;
+import com.yahoo.search.dispatch.searchcluster.Group;
 import com.yahoo.search.dispatch.searchcluster.SearchCluster;
 import com.yahoo.search.result.Coverage;
 import com.yahoo.search.result.ErrorMessage;
@@ -41,9 +42,9 @@ public class InterleavedSearchInvoker extends SearchInvoker implements ResponseM
 
     private final Set<SearchInvoker> invokers;
     private final SearchCluster searchCluster;
+    private final Group group;
     private final LinkedBlockingQueue<SearchInvoker> availableForProcessing;
     private final Set<Integer> alreadyFailedNodes;
-    private final boolean isContentWellBalanced;
     private Query query;
 
     private boolean adaptiveTimeoutCalculated = false;
@@ -60,14 +61,17 @@ public class InterleavedSearchInvoker extends SearchInvoker implements ResponseM
     private boolean timedOut = false;
     private boolean degradedByMatchPhase = false;
 
-    public InterleavedSearchInvoker(Collection<SearchInvoker> invokers, boolean isContentWellBalanced, SearchCluster searchCluster, Set<Integer> alreadyFailedNodes) {
+    public InterleavedSearchInvoker(Collection<SearchInvoker> invokers,
+                                    SearchCluster searchCluster,
+                                    Group group,
+                                    Set<Integer> alreadyFailedNodes) {
         super(Optional.empty());
         this.invokers = Collections.newSetFromMap(new IdentityHashMap<>());
         this.invokers.addAll(invokers);
         this.searchCluster = searchCluster;
+        this.group = group;
         this.availableForProcessing = newQueue();
         this.alreadyFailedNodes = alreadyFailedNodes;
-        this.isContentWellBalanced = isContentWellBalanced;
     }
 
     /**
@@ -85,7 +89,7 @@ public class InterleavedSearchInvoker extends SearchInvoker implements ResponseM
         int originalOffset = query.getOffset();
         int neededHits = originalHits + originalOffset;
         int q = neededHits;
-        if (isContentWellBalanced) {
+        if (group.isBalanced() && !group.isSparse()) {
             Double topkProbabilityOverrride = query.properties().getDouble(Dispatcher.topKProbability);
             q = (topkProbabilityOverrride != null)
                     ? searchCluster.estimateHitsToFetch(neededHits, invokers.size(), topkProbabilityOverrride)
diff --git a/container-search/src/main/java/com/yahoo/search/dispatch/InvokerFactory.java b/container-search/src/main/java/com/yahoo/search/dispatch/InvokerFactory.java
index bd66ca88622..e602afadcfb 100644
--- a/container-search/src/main/java/com/yahoo/search/dispatch/InvokerFactory.java
+++ b/container-search/src/main/java/com/yahoo/search/dispatch/InvokerFactory.java
@@ -89,7 +89,7 @@ public abstract class InvokerFactory {
         if (invokers.size() == 1 && failed == null) {
             return Optional.of(invokers.get(0));
         } else {
-            return Optional.of(new InterleavedSearchInvoker(invokers, group.isBalanced(), searchCluster, failed));
+            return Optional.of(new InterleavedSearchInvoker(invokers, searchCluster, group, failed));
         }
     }
 
diff --git a/container-search/src/main/java/com/yahoo/search/dispatch/searchcluster/Group.java b/container-search/src/main/java/com/yahoo/search/dispatch/searchcluster/Group.java
index 7be5cde0b14..727fb64faef 100644
--- a/container-search/src/main/java/com/yahoo/search/dispatch/searchcluster/Group.java
+++ b/container-search/src/main/java/com/yahoo/search/dispatch/searchcluster/Group.java
@@ -61,7 +61,7 @@ public class Group {
         return (int) nodes.stream().filter(node -> node.isWorking() == Boolean.TRUE).count();
     }
 
-    void aggregateNodeValues() {
+    public void aggregateNodeValues() {
         long activeDocs = nodes.stream().filter(node -> node.isWorking() == Boolean.TRUE).mapToLong(Node::getActiveDocuments).sum();
         activeDocuments.set(activeDocs);
         isBlockingWrites.set(nodes.stream().anyMatch(Node::isBlockingWrites));
@@ -69,13 +69,13 @@ public class Group {
         if (numWorkingNodes > 0) {
             long average = activeDocs / numWorkingNodes;
             long skew = nodes.stream().filter(node -> node.isWorking() == Boolean.TRUE).mapToLong(node -> Math.abs(node.getActiveDocuments() - average)).sum();
-            boolean skewIsLow = skew <= activeDocs * maxContentSkew;
-            if (!isBalanced.get() || skewIsLow != isBalanced.get()) {
+            boolean balanced = skew <= activeDocs * maxContentSkew;
+            if (!isBalanced.get() || balanced != isBalanced.get()) {
                 if (!isSparse())
-                    log.info("Content is " + (skewIsLow ? "" : "not ") + "well balanced. Current deviation = " +
+                    log.info("Content is " + (balanced ? "" : "not ") + "well balanced. Current deviation = " +
                              skew * 100 / activeDocs + " %. activeDocs = " + activeDocs + ", skew = " + skew +
                              ", average = " + average);
-                isBalanced.set(skewIsLow);
+                isBalanced.set(balanced);
             }
         } else {
             isBalanced.set(true);
@@ -92,7 +92,10 @@ public class Group {
     public boolean isBalanced() { return isBalanced.get(); }
 
     /** Returns whether this group has too few documents per node to expect it to be balanced */
-    public boolean isSparse() { return activeDocuments.get() / nodes.size() < minDocsPerNodeToRequireLowSkew; }
+    public boolean isSparse() {
+        if (nodes.isEmpty()) return false;
+        return activeDocuments.get() / nodes.size() < minDocsPerNodeToRequireLowSkew;
+    }
 
     public boolean fullCoverageStatusChanged(boolean hasFullCoverageNow) {
         boolean previousState = hasFullCoverage.getAndSet(hasFullCoverageNow);