apache
/
zookeeper
tükrözi: https://gitee.com/apache/zookeeper.git


			
				
					
						
						
							123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911
							<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<META http-equiv="Content-Type" content="text/html; charset=UTF-8">
<meta content="Apache Forrest" name="Generator">
<meta name="Forrest-version" content="0.8">
<meta name="Forrest-skin-name" content="pelt">
<title></title>
<link type="text/css" href="skin/basic.css" rel="stylesheet">
<link media="screen" type="text/css" href="skin/screen.css" rel="stylesheet">
<link media="print" type="text/css" href="skin/print.css" rel="stylesheet">
<link type="text/css" href="skin/profile.css" rel="stylesheet">
<script src="skin/getBlank.js" language="javascript" type="text/javascript"></script><script src="skin/getMenu.js" language="javascript" type="text/javascript"></script><script src="skin/fontsize.js" language="javascript" type="text/javascript"></script>
<link rel="shortcut icon" href="images/favicon.ico">
</head>
<body onload="init()">
<script type="text/javascript">ndeSetTextSize();</script>
<div id="top">
<!--+
    |breadtrail
    +-->
<div class="breadtrail">
<a href="http://www.apache.org/">Apache</a> &gt; <a href="http://hadoop.apache.org/">Hadoop</a> &gt; <a href="http://hadoop.apache.org/zookeeper/">ZooKeeper</a><script src="skin/breadcrumbs.js" language="JavaScript" type="text/javascript"></script>
</div>
<!--+
    |header
    +-->
<div class="header">
<!--+
    |start group logo
    +-->
<div class="grouplogo">
<a href="http://hadoop.apache.org/"><img class="logoImage" alt="Hadoop" src="images/hadoop-logo.jpg" title="Apache Hadoop"></a>
</div>
<!--+
    |end group logo
    +-->
<!--+
    |start Project Logo
    +-->
<div class="projectlogo">
<a href="http://hadoop.apache.org/zookeeper/"><img class="logoImage" alt="ZooKeeper" src="images/zookeeper_small.gif" title="The Hadoop database"></a>
</div>
<!--+
    |end Project Logo
    +-->
<!--+
    |start Search
    +-->
<div class="searchbox">
<form action="http://www.google.com/search" method="get" class="roundtopsmall">
<input value="hadoop.apache.org" name="sitesearch" type="hidden"><input onFocus="getBlank (this, 'Search the site with google');" size="25" name="q" id="query" type="text" value="Search the site with google">&nbsp; 
                    <input name="Search" value="Search" type="submit">
</form>
</div>
<!--+
    |end search
    +-->
<!--+
    |start Tabs
    +-->
<ul id="tabs">
<li>
<a class="unselected" href="http://hadoop.apache.org/zookeeper/">Project</a>
</li>
<li>
<a class="unselected" href="http://wiki.apache.org/hadoop/ZooKeeper">Wiki</a>
</li>
<li class="current">
<a class="selected" href="index.html">ZooKeeper Documentation</a>
</li>
</ul>
<!--+
    |end Tabs
    +-->
</div>
</div>
<div id="main">
<div id="publishedStrip">
<!--+
    |start Subtabs
    +-->
<div id="level2tabs"></div>
<!--+
    |end Endtabs
    +-->
<script type="text/javascript"><!--
document.write("Last Published: " + document.lastModified);
//  --></script>
</div>
<!--+
    |breadtrail
    +-->
<div class="breadtrail">

             &nbsp;
           </div>
<!--+
    |start Menu, mainarea
    +-->
<!--+
    |start Menu
    +-->
<div id="menu">
<div onclick="SwitchMenu('menu_selected_1.1', 'skin/')" id="menu_selected_1.1Title" class="menutitle" style="background-image: url('skin/images/chapter_open.gif');">Documentation</div>
<div id="menu_selected_1.1" class="selectedmenuitemgroup" style="display: block;">
<div class="menuitem">
<a href="index.html">Welcome</a>
</div>
<div class="menuitem">
<a href="zookeeperOver.html">Zookeeper Overview</a>
</div>
<div class="menuitem">
<a href="zookeeperStarted.html">Getting Started</a>
</div>
<div class="menuitem">
<a href="zookeeperProgrammers.html">Programmer's Guide</a>
</div>
<div class="menupage">
<div class="menupagetitle">Recipes</div>
</div>
<div class="menuitem">
<a href="zookeeperAdmin.html">Administrator's Guide</a>
</div>
<div class="menuitem">
<a href="api/index.html">API Docs</a>
</div>
<div class="menuitem">
<a href="http://wiki.apache.org/hadoop/ZooKeeper">Wiki</a>
</div>
<div class="menuitem">
<a href="http://wiki.apache.org/hadoop/ZooKeeper/FAQ">FAQ</a>
</div>
<div class="menuitem">
<a href="http://hadoop.apache.org/zookeeper/mailing_lists.html">Mailing Lists</a>
</div>
<div class="menuitem">
<a href="zookeeperOtherInfo.html">Other Info</a>
</div>
</div>
<div id="credit"></div>
<div id="roundbottom">
<img style="display: none" class="corner" height="15" width="15" alt="" src="skin/images/rc-b-l-15-1body-2menu-3menu.png"></div>
<!--+
  |alternative credits
  +-->
<div id="credit2"></div>
</div>
<!--+
    |end Menu
    +-->
<!--+
    |start content
    +-->
<div id="content">
<div title="Portable Document Format" class="pdflink">
<a class="dida" href="recipes.pdf"><img alt="PDF -icon" src="skin/images/pdfdoc.gif" class="skin"><br>
        PDF</a>
</div>
<div id="minitoc-area">
<ul class="minitoc">
<li>
<a href="#A+Guide+to+Creating+Higher-level+Constructs+with+ZooKeeper">A Guide to Creating Higher-level Constructs with ZooKeeper</a>
<ul class="minitoc">
<li>
<a href="#sc_outOfTheBox">Out of the Box Applications: Name Service, Configuration, Group
    Membership</a>
</li>
<li>
<a href="#sc_recipes_eventHandles">Barriers</a>
<ul class="minitoc">
<li>
<a href="#sc_doubleBarriers">Double Barriers</a>
</li>
</ul>
</li>
<li>
<a href="#sc_recipes_Queues">Queues</a>
<ul class="minitoc">
<li>
<a href="#sc_recipes_priorityQueues">Priority Queues</a>
</li>
</ul>
</li>
<li>
<a href="#sc_recipes_Locks">Locks</a>
<ul class="minitoc">
<li>
<a href="#Shared+Locks">Shared Locks</a>
</li>
<li>
<a href="#sc_recoverableSharedLocks">Recoverable Shared Locks</a>
</li>
</ul>
</li>
<li>
<a href="#sc_recipes_twoPhasedCommit">Two-phased Commit</a>
</li>
<li>
<a href="#sc_leaderElection">Leader Election</a>
</li>
</ul>
</li>
</ul>
</div>
  
<title>ZooKeeper Recipes and Solutions</title>

  

  
<a name="N1000A"></a><a name="A+Guide+to+Creating+Higher-level+Constructs+with+ZooKeeper"></a>
<h2 class="h3">A Guide to Creating Higher-level Constructs with ZooKeeper</h2>
<div class="section">
<p>In this article, you'll find guidelines for using
    ZooKeeper to implement higher order functions. All of them are conventions
    implemented at the client and do not require special support from
    ZooKeeper. Hopfully the community will capture these conventions in client-side libraries 
    to ease their use and to encourage standardization.</p>
<p>One of the most interesting things about ZooKeeper is that even
    though ZooKeeper uses <em>asynchronous</em> notifications, you
    can use it to build <em>synchronous</em> consistency
    primitives, such as queues and locks. As you will see, this is possible
    because ZooKeeper imposes an overall order on updates, and has mechanisms
    to expose this ordering.</p>
<p>Note that the recipes below attempt to employ best practices. In
    particular, they avoid polling, timers or anything else that would result
    in a "herd effect", causing bursts of traffic and limiting
    scalability.</p>
<p>There are many useful functions that can be imagined that aren't
    included here - revocable read-write priority locks, as just one example.
    And some of the constructs mentioned here - locks, in particular -
    illustrate certain points, even though you may find other constructs, such
    as event handles or queues, a more practical means of performing the same
    function. In general, the examples in this section are designed to
    stimulate thought.</p>
<a name="N10022"></a><a name="sc_outOfTheBox"></a>
<h3 class="h4">Out of the Box Applications: Name Service, Configuration, Group
    Membership</h3>
<p>Name service and configuration are two of the primary applications
    of ZooKeeper. These two functions are provided directly by the ZooKeeper
    API.</p>
<p>Another function directly provided by ZooKeeper is <em>group
    membership</em>. The group is represented by a node. Members of the
    group create ephemeral nodes under the group node. Nodes of the members
    that fail abnormally will be removed automatically when ZooKeeper detects
    the failure.</p>
<a name="N10032"></a><a name="sc_recipes_eventHandles"></a>
<h3 class="h4">Barriers</h3>
<p>Distributed systems use <em>barriers</em> to block
    processing of a set of nodes until a condition is met at which time all
    the nodes are allowed to proceed. Barriers are implemented in ZooKeeper by
    designating a barrier node. The barrier is in place if the barrier node
    exists. Here's the pseudo code:</p>
<ol>
      
<li>
        
<p>Client calls the ZooKeeper API's <strong>exists()</strong> function on the barrier node, with
        <em>watch</em> set to true.</p>
      
</li>

      
<li>
        
<p>If <strong>exists()</strong> returns false, the
        barrier is gone and the client proceeds</p>
      
</li>

      
<li>
        
<p>Else, if <strong>exists()</strong> returns true,
        the clients wait for a watch event from ZooKeeper for the barrier
        node.</p>
      
</li>

      
<li>
        
<p>When the watch event is triggered, the client reissues the
        <strong>exists( )</strong> call, again waiting until
        the barrier node is removed.</p>
      
</li>
    
</ol>
<p>
<remark>[tbd: maybe an illustration would be nice for each of the
    recipes?]</remark>
</p>
<a name="N1006C"></a><a name="sc_doubleBarriers"></a>
<h4>Double Barriers</h4>
<p>Double barriers enable clients to synchronize the beginning and
      the end of a computation. When enough processes have joined the barrier,
      processes start their computation and leave the barrier once they have
      finished. This recipe shows how to use a ZooKeeper node as a
      barrier.</p>
<p>The pseudo code in this recipe represents the barrier node as
      <em>b</em>. Every client process <em>p</em>
      registers with the barrier node on entry and unregisters when it is
      ready to leave. A node registers with the barrier node via the <strong>Enter</strong> procedure below, it waits until
      <em>x</em> client process register before proceeding with
      the computation. (The <em>x</em> here is up to you to
      determine for your system.)</p>
<p>
<table class="ForrestTable" cellspacing="1" cellpadding="4">
          
              
<tr>
                
<td><strong>Enter</strong></td>

                <td><strong>Leave</strong></td>
              
</tr>

              
<tr>
                
<td>
<ol>
                    
<li>
                      
<p>Create a name <em><em>n</em> =
                      <em>b</em>+&ldquo;/&rdquo;+<em>p</em></em>
</p>
                    
</li>

                    
<li>
                      
<p>Set watch: <strong>exists(<em>b</em> + &lsquo;&lsquo;/ready&rsquo;&rsquo;,
                      true)</strong>
</p>
                    
</li>

                    
<li>
                      
<p>Create child: <strong>create(
                      <em>n</em>, EPHEMERAL)</strong>
</p>
                    
</li>

                    
<li>
                      
<p>
<strong>L = getChildren(b,
                      false)</strong>
</p>
                    
</li>

                    
<li>
                      
<p>if fewer children in L than<em>
                      x</em>, wait for watch event <remark>[tbd: how do
                      you wait?]</remark>
</p>
                    
</li>

                    
<li>
                      
<p>else <strong>create(b + &lsquo;&lsquo;/ready&rsquo;&rsquo;,
                      REGULAR)</strong>
</p>
                    
</li>
                  
</ol>
</td>

                <td>
<ol>
                    
<li>
                      
<p>
<strong>L = getChildren(b,
                      false)</strong>
</p>
                    
</li>

                    
<li>
                      
<p>if no children, exit</p>
                    
</li>

                    
<li>
                      
<p>if <em>p</em> is only process node in
                      L, delete(n) and exit</p>
                    
</li>

                    
<li>
                      
<p>if <em>p</em> is the lowest process
                      node in L, wait on highest process node in P</p>
                    
</li>

                    
<li>
                      
<p>else <strong>delete(<em>n</em>) </strong>if
                      still exists and wait on lowest process node in L</p>
                    
</li>

                    
<li>
                      
<p>goto 1</p>
                    
</li>
                  
</ol>
</td>
              
</tr>
            
        
</table>On entering, all processes watch on a ready node and
      create an ephemeral node as a child of the barrier node. Each process
      but the last enters the barrier and waits for the ready node to appear
      at line 5. The process that creates the xth node, the last process, will
      see x nodes in the list of children and create the ready node, waking up
      the other processes. Note that waiting processes wake up only when it is
      time to exit, so waiting is efficient.</p>
<p>On exit, you can't use a flag such as <em>ready</em>
      because you are watching for process nodes to go away. By using
      ephemeral nodes, processes that fail after the barrier has been entered
      do not prevent correct processes from finishing. When processes are
      ready to leave, they need to delete their process nodes and wait for all
      other processes to do the same.</p>
<p>Processes exit when there are no process nodes left as children of
      <em>b</em>. However, as an efficiency, you can use the
      lowest process node as the ready flag. All other processes that are
      ready to exit watch for the lowest existing process node to go away, and
      the owner of the lowest process watches for any other process node
      (picking the highest for simplicity) to go away. This means that only a
      single process wakes up on each node deletion except for the last node,
      which wakes up everyone when it is removed.</p>
<a name="N10120"></a><a name="sc_recipes_Queues"></a>
<h3 class="h4">Queues</h3>
<p>Distributed queues are a common data structure. To implement a
    distributed queue in ZooKeeper, first designate a znode to hold the queue,
    the queue node. The distributed clients put something into the queue by
    calling create() with a pathname ending in "queue-", with the
    <em>sequence</em> and <em>ephemeral</em> flags in
    the create() call set to true. Because the <em>sequence</em>
    flag is set, the new pathnames will have the form
    _path-to-queue-node_/queue-X, where X is a monotonic increasing number. A
    client that wants to be remove from the queue calls ZooKeeper's <strong>getChildren( )</strong> function, with
    <em>watch</em> set to true on the queue node, and begins
    processing nodes with the lowest number. The client does not need to issue
    another <strong>getChildren( )</strong> until it exhausts
    the list obtained from the first <strong>getChildren(
    )</strong> call. If there are are no children in the queue node, the
    reader waits for a watch notification to check to queue again.</p>
<a name="N1013E"></a><a name="sc_recipes_priorityQueues"></a>
<h4>Priority Queues</h4>
<p>To implement a priority queue, you need only make two simple
      changes to the generic <a href="#sc_recipes_Queues">queue
      recipe</a> . First, to add to a queue, the pathname ends with
      "queue-YY" where YY is the priority of the element with lower numbers
      representing higher priority (just like UNIX). Second, when removing
      from the queue a client uses an up-to-date children list meaning that
      the client will invalidate previously obtained children lists if a watch
      notification triggers for the queue node.</p>
<a name="N1014D"></a><a name="sc_recipes_Locks"></a>
<h3 class="h4">Locks</h3>
<p>Fully distributed locks that are globally synchronous, meaning at
    any snapshot in time no two clients think they hold the same lock. These
    can be implemented using ZooKeeeper. As with priority queues, first define
    a lock node.</p>
<p>Clients wishing to obtain a lock do the following:</p>
<ol>
      
<li>
        
<p>Call <strong>create( )</strong> with a pathname
        of "_locknode_/lock-" and the <em>sequence</em> and
        <em>ephemeral</em> flags set.</p>
      
</li>

      
<li>
        
<p>Call <strong>getChildren( )</strong> on the lock
        node <em>without</em> setting the watch flag (this is
        important to avoid the herd effect).</p>
      
</li>

      
<li>
        
<p>If the pathname created in step <strong>1</strong> has the lowest sequence number suffix, the
        client has the lock and the client exits the protocol.</p>
      
</li>

      
<li>
        
<p>The client calls <strong>exists( )</strong> with
        the watch flag set on the path in the lock directory with the next
        lowest sequence number.</p>
      
</li>

      
<li>
        
<p>if <strong>exists( )</strong> returns false, go
        to step <strong>2</strong>. Otherwise, wait for a
        notification for the pathname from the previous step before going to
        step <strong>2</strong>.</p>
      
</li>
    
</ol>
<p>The unlock protocol is very simple: clients wishing to release a
    lock simply delete the node they created in step 1.</p>
<p>Here are a few things to notice:</p>
<ul>
      
<li>
        
<p>The removal of a node will only cause one client to wake up
        since each node is watched by exactly one client. In this way, you
        avoid the herd effect.</p>
      
</li>
    
</ul>
<ul>
      
<li>
        
<p>There is no polling or timeouts.</p>
      
</li>
    
</ul>
<ul>
      
<li>
        
<p>Because of the way you implement locking, it is easy to see the
        amount of lock contention, break locks, debug locking problems,
        etc.</p>
      
</li>
    
</ul>
<a name="N101B9"></a><a name="Shared+Locks"></a>
<h4>Shared Locks</h4>
<p>You can implement shared locks by with a few changes to the lock
      protocol:</p>
<table class="ForrestTable" cellspacing="1" cellpadding="4">
        
            
<tr>
              
<td><strong>Obtaining a read
              lock:</strong></td>

              <td><strong>Obtaining a write
              lock:</strong></td>
            
</tr>

            
<tr>
              
<td>
<ol>
                  
<li>
                    
<p>Call <strong>create( )</strong> to
                    create a node with pathname
                    "<span class="codefrag parameter">_locknode_/read-</span>". This is the
                    lock node use later in the protocol. Make sure to set both
                    the <em>sequence</em> and
                    <em>ephemeral</em> flags.</p>
                  
</li>

                  
<li>
                    
<p>Call <strong>getChildren( )</strong>
                    on the lock node <em>without</em> setting the
                    <em>watch</em> flag - this is important, as it
                    avoids the herd effect.</p>
                  
</li>

                  
<li>
                    
<p>If there are no children with a pathname starting
                    with "<span class="codefrag parameter">write-</span>" and having a lower
                    sequence number than the node created in step <strong>1</strong>, the client has the lock and can
                    exit the protocol. </p>
                  
</li>

                  
<li>
                    
<p>Otherwise, call <strong>exists(
                    )</strong>, with <em>watch</em> flag, set on
                    the node in lock directory with pathname staring with
                    "<span class="codefrag parameter">write-</span>" having the next lowest
                    sequence number.</p>
                  
</li>

                  
<li>
                    
<p>If <strong>exists( )</strong>
                    returns <em>false</em>, goto step <strong>2</strong>.</p>
                  
</li>

                  
<li>
                    
<p>Otherwise, wait for a notification for the pathname
                    from the previous step before going to step <strong>2</strong>
</p>
                  
</li>
                
</ol>
</td>

              <td>
<ol>
                  
<li>
                    
<p>Call <strong>create( )</strong> to
                    create a node with pathname
                    "<span class="codefrag parameter">_locknode_/write-</span>". This is the
                    lock node spoken of later in the protocol. Make sure to
                    set both <em>sequence</em> and
                    <em>ephemeral</em> flags.</p>
                  
</li>

                  
<li>
                    
<p>Call <strong>getChildren( )
                    </strong> on the lock node <em>without</em>
                    setting the <em>watch</em> flag - this is
                    important, as it avoids the herd effect.</p>
                  
</li>

                  
<li>
                    
<p>If there are no children with a lower sequence
                    number than the node created in step <strong>1</strong>, the client has the lock and the
                    client exits the protocol.</p>
                  
</li>

                  
<li>
                    
<p>Call <strong>exists( ),</strong>
                    with <em>watch</em> flag set, on the node with
                    the pathname that has the next lowest sequence
                    number.</p>
                  
</li>

                  
<li>
                    
<p>If <strong>exists( )</strong>
                    returns <em>false</em>, goto step <strong>2</strong>. Otherwise, wait for a
                    notification for the pathname from the previous step
                    before going to step <strong>2</strong>.</p>
                  
</li>
                
</ol>
</td>
            
</tr>
          
      
</table>
<p>
<div class="note">
<div class="label">Note</div>
<div class="content">
          
<p>It might appear that this recipe creates a herd effect: when
          there is a large group of clients waiting for a read lock, and all
          getting notified more or less simultaneously when the
          "<span class="codefrag parameter">write-</span>" node with the lowest sequence number
          is deleted. In fact. that's valid behavior: as all those waiting
          reader clients should be released since they have the lock. The herd
          effect refers to releasing a "herd" when in fact only a single or a
          small number of machines can proceed. <remark>[tbd: myabe helpful to
          indicate which step this refers to?]</remark>
</p>
        
</div>
</div>
</p>
<a name="N10288"></a><a name="sc_recoverableSharedLocks"></a>
<h4>Recoverable Shared Locks</h4>
<p>With minor modifications to the Shared Lock protocol, you make
      shared locks revocable by modifying the shared lock protocol:</p>
<p>In step <strong>1</strong>, of both obtain reader
      and writer lock protocols, call <strong>getData(
      )</strong> with <em>watch</em> set, immediately after the
      call to <strong>create( )</strong>. If the client
      subsequently receives notification for the node it created in step
      <strong>1</strong>, it does another <strong>getData( )</strong> on that node, with
      <em>watch</em> set and looks for the string "unlock", which
      signals to the client that it must release the lock. This is because,
      according to this shared lock protocol, you can request the client with
      the lock give up the lock by calling <strong>setData()
      </strong> on the lock node, writing "unlock" to that node.</p>
<p>Note that this protocol requires the lock holder to consent to
      releasing the lock. Such consent is important, especially if the lock
      holder needs to do some processing before releasing the lock. Of course
      you can always implement <em>Revocable Shared Locks with Freaking
      Laser Beams</em> by stipulating in your protocol that the revoker
      is allowed to delete the lock node if after some length of time the lock
      isn't deleted by the lock holder.</p>
<a name="N102B4"></a><a name="sc_recipes_twoPhasedCommit"></a>
<h3 class="h4">Two-phased Commit</h3>
<p>A two-phase commit protocol is an algorithm that lets all clients in
    a distributed system agree either to commit a transaction or abort.</p>
<p>In ZooKeeper, you can implement a two-phased commit by having a
    coordinator create a transaction node, say "/app/Tx", and one child node
    per participating site, say "/app/Tx/s_i". When coordinator creates the
    child node, it leaves the content undefined. Once each site involved in
    the transaction receives the transaction from the coordinator, the site
    reads each child node and sets a watch. Each site then processes the query
    and votes "commit" or "abort" by writing to its respective node. Once the
    write completes, the other sites are notified, and as soon as all sites
    have all votes, they can decide either "abort" or "commit". Note that a
    node can decide "abort" earlier if some site votes for "abort".</p>
<p>An interesting aspect of this implementation is that the only role
    of the coordinator is to decide upon the group of sites, to create the
    ZooKeeper nodes, and to propagate the transaction to the corresponding
    sites. In fact, even propagating the transaction can be done through
    ZooKeeper by writing it in the transaction node.</p>
<p>There are two important drawbacks of the approach described above.
    One is the message complexity, which is O(n&sup2;). The second is the
    impossibility of detecting failures of sites through ephemeral nodes. To
    detect the failure of a site using ephemeral nodes, it is necessary that
    the site create the node.</p>
<p>To solve the first problem, you can have only the coordinator
    notified of changes to the transaction nodes, and then notify the sites
    once coordinator reaches a decision. Note that this approach is scalable,
    but it's is slower too, as it requires all communication to go through the
    coordinator.</p>
<p>To address the second problem, you can have the coordinator
    propagate the transaction to the sites, and have each site creating its
    own ephemeral node.</p>
<a name="N102CD"></a><a name="sc_leaderElection"></a>
<h3 class="h4">Leader Election</h3>
<p>A simple way of doing leader election with ZooKeeper is to use the
    <strong>SEQUENCE|EPHEMERAL</strong> flags when creating
    znodes that represent "proposals" of clients. The idea is to have a znode,
    say "/election", such that each znode creates a child znode "/election/n_"
    with both flags SEQUENCE|EPHEMERAL. With the sequence flag, ZooKeeper
    automatically appends a sequence number that is greater that any one
    previously appended to a child of "/election". The process that created
    the znode with the smallest appended sequence number is the leader.
    </p>
<p>That's not all, though. It is important to watch for failures of the
    leader, so that a new client arises as the new leader in the case the
    current leader fails. A trivial solution is to have all application
    processes watching upon the current smallest znode, and checking if they
    are the new leader when the smallest znode goes away (note that the
    smallest znode will go away if the leader fails because the node is
    ephemeral). But this causes a herd effect: upon of failure of the current
    leader, all other processes receive a notification, and execute
    getChildren on "/election" to obtain the current list of children of
    "/election". If the number of clients is large, it causes a spike on the
    number of operations that ZooKeeper servers have to process. To avoid the
    herd effect, it is sufficient to watch for the next znode down on the
    sequence of znodes. If a client receives a notification that the znode it
    is watching is gone, then it becomes the new leader in the case that there
    is no smaller znode. Note that this avoids the herd effect by not having
    all clients watching the same znode. </p>
<p>Here's the pseudo code:</p>
<p>Let ELECTION be a path of choice of the application. To volunteer to
    be a leader: </p>
<ol>
      
<li>
        
<p>Create znode z with path "ELECTION/n_" with both SEQUENCE and
        EPHEMERAL flags;</p>
      
</li>

      
<li>
        
<p>Let C be the children of "ELECTION", and i be the sequence
        number of z;</p>
      
</li>

      
<li>
        
<p>Watch for changes on "ELECTION/n_j", where j is the smallest
        sequence number such that j &lt; i and n_j is a znode in C;</p>
      
</li>
    
</ol>
<p>Upon receiving a notification of znode deletion: </p>
<ol>
      
<li>
        
<p>Let C be the new set of children of ELECTION; </p>
      
</li>

      
<li>
        
<p>If z is the smallest node in C, then execute leader
        procedure;</p>
      
</li>

      
<li>
        
<p>Otherwise, watch for changes on "ELECTION/n_j", where j is the
        smallest sequence number such that j &lt; i and n_j is a znode in C;
        </p>
      
</li>
    
</ol>
<p>Note that the znode having no preceding znode on the list of
    children does not imply that the creator of this znode is aware that it is
    the current leader. Applications may consider creating a separate to znode
    to acknowledge that the leader has executed the leader procedure. </p>
</div>

<p align="right">
<font size="-2"></font>
</p>
</div>
<!--+
    |end content
    +-->
<div class="clearboth">&nbsp;</div>
</div>
<div id="footer">
<!--+
    |start bottomstrip
    +-->
<div class="lastmodified">
<script type="text/javascript"><!--
document.write("Last Published: " + document.lastModified);
//  --></script>
</div>
<div class="copyright">
        Copyright &copy;
         2008 <a href="http://www.apache.org/licenses/">The Apache Software Foundation.</a>
</div>
<!--+
    |end bottomstrip
    +-->
</div>
</body>
</html>