SOLR-17158 Terminate distributed processing faster when query limit i…

…s reached and partial results are not needed (#2666)

solr/CHANGES.txt

@@ -111,6 +111,12 @@ New Features
 
 Improvements
 ---------------------
+* SOLR-17158: Users using query limits (timeAllowed, cpuTimeAllowed) for whom partial results are uninteresting
+  may set partialResults=false. This parameter has been enhanced to reduce time spent processing partial results
+  and omit partialResults from the response. Since this is requested behavior, no exception is thrown and the
+  partialResults response header will always exist if the result was short circuited.
+  (Gus Heck, Andrzej Bialecki, hossman)
+
 * SOLR-17397: SkipExistingDocumentsProcessor now functions correctly with child documents.  (Tim Owens via Eric Pugh)
 
 * SOLR-17180: Deprecate snapshotscli.sh in favour of bin/solr snapshot sub commands.  Now able to manage Snapshots from the CLI.  HDFS module specific snapshot script now ships as part of that module in the modules/hdfs/bin directory. (Eric Pugh)

solr/core/src/java/org/apache/solr/handler/RequestHandlerBase.java

@@ -17,6 +17,7 @@
 package org.apache.solr.handler;
 
 import static org.apache.solr.core.RequestParams.USEPARAM;
+import static org.apache.solr.response.SolrQueryResponse.haveCompleteResults;
 
 import com.codahale.metrics.Counter;
 import com.codahale.metrics.Meter;
@@ -235,7 +236,8 @@ public void handleRequest(SolrQueryRequest req, SolrQueryResponse rsp) {
       rsp.setHttpCaching(httpCaching);
       handleRequestBody(req, rsp);
       // count timeouts
-      if (rsp.isPartialResults()) {
+
+      if (!haveCompleteResults(rsp.getResponseHeader())) {
         metrics.numTimeouts.mark();
         rsp.setHttpCaching(false);
       }