From ac7621e607f2b71d197f3b1c381e0caf2a6111d5 Mon Sep 17 00:00:00 2001
From: Timothy Potter <thelabdude@apache.org>
Date: Mon, 6 Oct 2014 17:50:00 +0000
Subject: [PATCH] SOLR-6592: Avoid waiting for the leader to see the down state
 if that leader is not live.

git-svn-id: https://svn.apache.org/repos/asf/lucene/dev/trunk@1629719 13f79535-47bb-0310-9956-ffa450edef68
---
 .../org/apache/solr/cloud/ZkController.java   | 22 +++++++++++++++++++
 .../apache/solr/cloud/HttpPartitionTest.java  | 11 ++++++++++
 2 files changed, 33 insertions(+)

diff --git a/solr/core/src/java/org/apache/solr/cloud/ZkController.java b/solr/core/src/java/org/apache/solr/cloud/ZkController.java
index 88a48aa9943..d149d94b7ce 100644
--- a/solr/core/src/java/org/apache/solr/cloud/ZkController.java
+++ b/solr/core/src/java/org/apache/solr/cloud/ZkController.java
@@ -20,8 +20,11 @@ package org.apache.solr.cloud;
 import java.io.File;
 import java.io.IOException;
 import java.io.UnsupportedEncodingException;
+import java.net.ConnectException;
 import java.net.InetAddress;
 import java.net.NetworkInterface;
+import java.net.SocketException;
+import java.net.SocketTimeoutException;
 import java.net.URLEncoder;
 import java.net.UnknownHostException;
 import java.util.ArrayList;
@@ -41,6 +44,8 @@ import java.util.concurrent.TimeoutException;
 
 import org.apache.commons.io.FileUtils;
 import org.apache.commons.lang.StringUtils;
+import org.apache.http.NoHttpResponseException;
+import org.apache.http.conn.ConnectTimeoutException;
 import org.apache.solr.client.solrj.impl.HttpSolrServer;
 import org.apache.solr.client.solrj.request.CoreAdminRequest.WaitForState;
 import org.apache.solr.common.SolrException;
@@ -1623,6 +1628,23 @@ public final class ZkController {
               server.request(prepCmd);
               break;
             } catch (Exception e) {
+
+              // if the core container is shutdown, don't wait
+              if (cc.isShutDown()) {
+                throw new SolrException(ErrorCode.SERVICE_UNAVAILABLE,
+                    "Core container is shutdown.");
+              }
+
+              Throwable rootCause = SolrException.getRootCause(e);
+              if (rootCause instanceof IOException) {
+                // if there was a communication error talking to the leader, see if the leader is even alive
+                if (!zkStateReader.getClusterState().liveNodesContain(leaderProps.getNodeName())) {
+                  throw new SolrException(ErrorCode.SERVICE_UNAVAILABLE,
+                      "Node "+leaderProps.getNodeName()+" hosting leader for "+
+                          shard+" in "+collection+" is not live!");
+                }
+              }
+
               SolrException.log(log,
                   "There was a problem making a request to the leader", e);
               try {
diff --git a/solr/core/src/test/org/apache/solr/cloud/HttpPartitionTest.java b/solr/core/src/test/org/apache/solr/cloud/HttpPartitionTest.java
index 1027a88d492..c7b002b6c05 100644
--- a/solr/core/src/test/org/apache/solr/cloud/HttpPartitionTest.java
+++ b/solr/core/src/test/org/apache/solr/cloud/HttpPartitionTest.java
@@ -117,6 +117,10 @@ public class HttpPartitionTest extends AbstractFullDistribZkTestBase {
 
     // have the leader lose its Zk session temporarily
     testLeaderZkSessionLoss();
+
+    waitForThingsToLevelOut(30000);
+
+    log.info("HttpParitionTest succeeded ... shutting down now!");
   }
   
   protected void testRf2() throws Exception {
@@ -187,6 +191,8 @@ public class HttpPartitionTest extends AbstractFullDistribZkTestBase {
     // verify all docs received
     assertDocsExistInAllReplicas(notLeaders, testCollectionName, 1, numDocs + 3);
 
+    log.info("testRf2 succeeded ... deleting the "+testCollectionName+" collection");
+
     // try to clean up
     try {
       CollectionAdminRequest req = new CollectionAdminRequest.Delete();
@@ -240,6 +246,9 @@ public class HttpPartitionTest extends AbstractFullDistribZkTestBase {
     sendDoc(4);
     
     assertDocsExistInAllReplicas(notLeaders, testCollectionName, 1, 4);
+
+    log.info("testRf3 succeeded ... deleting the "+testCollectionName+" collection");
+
     // try to clean up
     try {
       CollectionAdminRequest req = new CollectionAdminRequest.Delete();
@@ -334,6 +343,8 @@ public class HttpPartitionTest extends AbstractFullDistribZkTestBase {
     waitToSeeReplicasActive(testCollectionName, "shard1", replicasToCheck, 20);
     assertDocsExistInAllReplicas(participatingReplicas, testCollectionName, 1, 2);
 
+    log.info("testLeaderZkSessionLoss succeeded ... deleting the "+testCollectionName+" collection");
+
     // try to clean up
     try {
       CollectionAdminRequest req = new CollectionAdminRequest.Delete();