Mailing-List: contact commits-help@zookeeper.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@<zookeeper.apache.org>
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: svn commit: r1481322 - in /zookeeper/branches/branch-3.4: CHANGES.txt
 src/java/main/org/apache/zookeeper/server/quorum/LearnerHandler.java
 src/java/main/org/apache/zookeeper/server/quorum/QuorumPeer.java
Date: Sat, 11 May 2013 13:32:38 -0000
To: commits@zookeeper.apache.org
From: fpj@apache.org
Message-Id: <20130511133239.22CE52388847@eris.apache.org>

Author: fpj
Date: Sat May 11 13:32:38 2013
New Revision: 1481322

URL: http://svn.apache.org/r1481322
Log:
ZOOKEEPER-1697:  large snapshots can cause continuous quorum failure
  (phunt via fpj)


Modified:
    zookeeper/branches/branch-3.4/CHANGES.txt
    zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/LearnerHandler.java
    zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/QuorumPeer.java

Modified: zookeeper/branches/branch-3.4/CHANGES.txt
URL: http://svn.apache.org/viewvc/zookeeper/branches/branch-3.4/CHANGES.txt?rev=1481322&r1=1481321&r2=1481322&view=diff
==============================================================================
--- zookeeper/branches/branch-3.4/CHANGES.txt (original)
+++ zookeeper/branches/branch-3.4/CHANGES.txt Sat May 11 13:32:38 2013
@@ -60,6 +60,9 @@ BUGFIXES:
   ZOOKEEPER-1633. Introduce a protocol version to connection initiation
   message (Alexander Shraer via michim)
 
+  ZOOKEEPER-1697:  large snapshots can cause continuous quorum failure
+  (phunt via fpj)
+
 IMPROVEMENTS:
 
   ZOOKEEPER-1564. Allow JUnit test build with IBM Java

Modified: zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/LearnerHandler.java
URL: http://svn.apache.org/viewvc/zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/LearnerHandler.java?rev=1481322&r1=1481321&r2=1481322&view=diff
==============================================================================
--- zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/LearnerHandler.java (original)
+++ zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/LearnerHandler.java Sat May 11 13:32:38 2013
@@ -64,7 +64,11 @@ public class LearnerHandler extends Thre
 
     final Leader leader;
 
-    long tickOfLastAck;
+    /** Deadline for receiving the next ack. If we are bootstrapping then
+     * it's based on the initLimit, if we are done bootstrapping it's based
+     * on the syncLimit. Once the deadline is past this learner should
+     * be considered no longer "sync'd" with the leader. */
+    volatile long tickOfNextAckDeadline;
     
     /**
      * ZooKeeper server identifier of this learner
@@ -104,7 +108,7 @@ public class LearnerHandler extends Thre
     public String toString() {
         StringBuilder sb = new StringBuilder();
         sb.append("LearnerHandler ").append(sock);
-        sb.append(" tickOfLastAck:").append(tickOfLastAck());
+        sb.append(" tickOfNextAckDeadline:").append(tickOfNextAckDeadline());
         sb.append(" synced?:").append(synced());
         sb.append(" queuedPacketLength:").append(queuedPackets.size());
         return sb.toString();
@@ -232,7 +236,10 @@ public class LearnerHandler extends Thre
      */
     @Override
     public void run() {
-        try {            
+        try {
+            tickOfNextAckDeadline = leader.self.tick
+                    + leader.self.initLimit + leader.self.syncLimit;
+
             ia = BinaryInputArchive.getArchive(new BufferedInputStream(sock
                     .getInputStream()));
             bufferedOutput = new BufferedOutputStream(sock.getOutputStream());
@@ -452,6 +459,7 @@ public class LearnerHandler extends Thre
                 LOG.error("Next packet was supposed to be an ACK");
                 return;
             }
+            LOG.info("Received NEWLEADER-ACK message from " + getSid());
             leader.waitForNewLeaderAck(getSid(), qp.getZxid(), getLearnerType());
             
             // now that the ack has been processed expect the syncLimit
@@ -482,7 +490,7 @@ public class LearnerHandler extends Thre
                 if (LOG.isTraceEnabled()) {
                     ZooTrace.logQuorumPacket(LOG, traceMask, 'i', qp);
                 }
-                tickOfLastAck = leader.self.tick;
+                tickOfNextAckDeadline = leader.self.tick + leader.self.syncLimit;
 
 
                 ByteBuffer bb;
@@ -597,8 +605,8 @@ public class LearnerHandler extends Thre
         leader.removeLearnerHandler(this);
     }
 
-    public long tickOfLastAck() {
-        return tickOfLastAck;
+    public long tickOfNextAckDeadline() {
+        return tickOfNextAckDeadline;
     }
 
     /**
@@ -620,6 +628,6 @@ public class LearnerHandler extends Thre
 
     public boolean synced() {
         return isAlive()
-        && tickOfLastAck >= leader.self.tick - leader.self.syncLimit;
+        && leader.self.tick <= tickOfNextAckDeadline;
     }
 }

Modified: zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/QuorumPeer.java
URL: http://svn.apache.org/viewvc/zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/QuorumPeer.java?rev=1481322&r1=1481321&r2=1481322&view=diff
==============================================================================
--- zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/QuorumPeer.java (original)
+++ zookeeper/branches/branch-3.4/src/java/main/org/apache/zookeeper/server/quorum/QuorumPeer.java Sat May 11 13:32:38 2013
@@ -246,7 +246,7 @@ public class QuorumPeer extends Thread i
     /**
      * The current tick
      */
-    protected int tick;
+    protected volatile int tick;
 
     /**
      * @deprecated As of release 3.4.0, this class has been deprecated, since