From d137991ccc876988ae8832c316457e525f6bf387 Mon Sep 17 00:00:00 2001
From: tedyu <yuzhihong@gmail.com>
Date: Thu, 18 May 2017 15:50:59 -0700
Subject: [PATCH] HBASE-18058 Zookeeper retry sleep time should have an upper
 limit (Allan Yang)

---
 .../hadoop/hbase/zookeeper/RecoverableZooKeeper.java      | 8 ++++----
 .../java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java    | 3 ++-
 hbase-common/src/main/resources/hbase-default.xml         | 7 +++++++
 3 files changed, 13 insertions(+), 5 deletions(-)

diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/RecoverableZooKeeper.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/RecoverableZooKeeper.java
index e0a950d6f5d..6968ad95acb 100644
--- a/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/RecoverableZooKeeper.java
+++ b/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/RecoverableZooKeeper.java
@@ -100,20 +100,20 @@ public class RecoverableZooKeeper {
   private static final int ID_LENGTH_SIZE =  Bytes.SIZEOF_INT;
 
   public RecoverableZooKeeper(String quorumServers, int sessionTimeout,
-      Watcher watcher, int maxRetries, int retryIntervalMillis)
+      Watcher watcher, int maxRetries, int retryIntervalMillis, int maxSleepTime)
   throws IOException {
-    this(quorumServers, sessionTimeout, watcher, maxRetries, retryIntervalMillis,
+    this(quorumServers, sessionTimeout, watcher, maxRetries, retryIntervalMillis, maxSleepTime,
         null);
   }
 
   @edu.umd.cs.findbugs.annotations.SuppressWarnings(value="DE_MIGHT_IGNORE",
       justification="None. Its always been this way.")
   public RecoverableZooKeeper(String quorumServers, int sessionTimeout,
-      Watcher watcher, int maxRetries, int retryIntervalMillis, String identifier)
+      Watcher watcher, int maxRetries, int retryIntervalMillis, int maxSleepTime, String identifier)
   throws IOException {
     // TODO: Add support for zk 'chroot'; we don't add it to the quorumServers String as we should.
     this.retryCounterFactory =
-      new RetryCounterFactory(maxRetries+1, retryIntervalMillis);
+      new RetryCounterFactory(maxRetries+1, retryIntervalMillis, maxSleepTime);
 
     if (identifier == null || identifier.length() == 0) {
       // the identifier = processID@hostName
diff --git a/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java b/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java
index c678a7cf7a0..08b059e8fb0 100644
--- a/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java
+++ b/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java
@@ -131,10 +131,11 @@ public class ZKUtil {
     int retry = conf.getInt("zookeeper.recovery.retry", 3);
     int retryIntervalMillis =
       conf.getInt("zookeeper.recovery.retry.intervalmill", 1000);
+    int maxSleepTime = conf.getInt("zookeeper.recovery.retry.maxsleeptime", 60000);
     zkDumpConnectionTimeOut = conf.getInt("zookeeper.dump.connection.timeout",
         1000);
     return new RecoverableZooKeeper(ensemble, timeout, watcher,
-        retry, retryIntervalMillis, identifier);
+        retry, retryIntervalMillis, maxSleepTime, identifier);
   }
 
   /**
diff --git a/hbase-common/src/main/resources/hbase-default.xml b/hbase-common/src/main/resources/hbase-default.xml
index a6e37ef61ea..70c638b9134 100644
--- a/hbase-common/src/main/resources/hbase-default.xml
+++ b/hbase-common/src/main/resources/hbase-default.xml
@@ -95,6 +95,13 @@ possible configurations would overwhelm and obscure the important.
   <!--The above are the important configurations for getting hbase up
     and running -->
 
+  <property>
+    <name>zookeeper.recovery.retry.maxsleeptime</name>
+    <value>60000</value>
+    <description>Max sleep time before retry zookeeper operations in milliseconds,
+    a max time is needed here so that sleep time won't grow unboundedly
+    </description>
+  </property>
   <property>
     <name>hbase.local.dir</name>
     <value>${hbase.tmp.dir}/local/</value>