libdb/docs/gsg_db_rep/JAVA/exampledoloop.html
2012-11-14 16:35:20 -05:00

658 lines
25 KiB
HTML
Raw Permalink Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" />
<title>Example Processing Loop</title>
<link rel="stylesheet" href="gettingStarted.css" type="text/css" />
<meta name="generator" content="DocBook XSL Stylesheets V1.73.2" />
<link rel="start" href="index.html" title="Getting Started with Replicated Berkeley DB Applications" />
<link rel="up" href="fwrkmasterreplica.html" title="Chapter 4. Replica versus Master Processes" />
<link rel="prev" href="processingloop.html" title="Processing Loop" />
<link rel="next" href="addfeatures.html" title="Chapter 5. Additional Features" />
</head>
<body>
<div xmlns="" class="navheader">
<div class="libver">
<p>Library Version 11.2.5.3</p>
</div>
<table width="100%" summary="Navigation header">
<tr>
<th colspan="3" align="center">Example Processing Loop</th>
</tr>
<tr>
<td width="20%" align="left"><a accesskey="p" href="processingloop.html">Prev</a> </td>
<th width="60%" align="center">Chapter 4. Replica versus Master Processes</th>
<td width="20%" align="right"> <a accesskey="n" href="addfeatures.html">Next</a></td>
</tr>
</table>
<hr />
</div>
<div class="sect1" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h2 class="title" style="clear: both"><a id="exampledoloop"></a>Example Processing Loop</h2>
</div>
</div>
</div>
<div class="toc">
<dl>
<dt>
<span class="sect2">
<a href="exampledoloop.html#runningit">Running It</a>
</span>
</dt>
</dl>
</div>
<p>
In this section we take the example
processing loop that we presented in the
previous section and we flesh it out to
provide a more complete example. We do this
by updating the
<code class="function">doloop()</code>
function that our original transaction
application used
<span>(see <a class="xref" href="simpleprogramlisting.html#doloop_java" title="Method: SimpleTxn.doloop()">Method: SimpleTxn.doloop()</a>)</span>
to fully support our replicated application.
</p>
<p>
In the following example code, code that we
add to the original example is presented in
<strong class="userinput"><code>bold</code></strong>.
</p>
<p>
To begin, we must implement a way to track whether our
application is running as a master or a client. There are many
ways to do this, but in this case what we will do is extend
<code class="classname">com.sleepycat.db.Environment</code> to carry
the information. We do this by creating the
<code class="classname">RepQuoteEnvironment</code> class.
</p>
<pre class="programlisting">package db.repquote;
import com.sleepycat.db.DatabaseException;
import com.sleepycat.db.Environment;
import com.sleepycat.db.EnvironmentConfig;
public class RepQuoteEnvironment extends Environment
{
private boolean isMaster;
public RepQuoteEnvironment(final java.io.File host,
EnvironmentConfig config)
throws DatabaseException, java.io.FileNotFoundException
{
super(host, config);
isMaster = false;
}
boolean getIsMaster()
{
return isMaster;
}
public void setIsMaster(boolean isMaster)
{
this.isMaster = isMaster;
}
}</pre>
<p>
Next, we go to <code class="filename">RepQuoteExampleGSG.java</code> and
we include the
<code class="classname">RepQuoteEnvironment</code> class
as well as the
<code class="classname">EventHandler</code> class. We then
cause our <code class="classname">RepQuoteExampleGSG</code>
class to implement
<code class="classname">EventHandler</code>. We also
change our environment handle to be an instance
of <code class="classname">RepQuoteEnvironment</code>
instead of <code class="classname">Environment</code>.
</p>
<p>
Note that we also import the
<code class="classname">com.sleepycat.db.ReplicationHandleDeadException</code>
class. We will discuss what that exception is
used for a little later in this example.
</p>
<pre class="programlisting">package db.repquote;
import java.io.FileNotFoundException;
import java.io.BufferedReader;
import java.io.InputStreamReader;
import java.io.IOException;
import java.io.UnsupportedEncodingException;
import java.lang.Thread;
import java.lang.InterruptedException;
import com.sleepycat.db.Cursor;
import com.sleepycat.db.Database;
import com.sleepycat.db.DatabaseConfig;
import com.sleepycat.db.DatabaseEntry;
import com.sleepycat.db.DatabaseException;
import com.sleepycat.db.DatabaseType;
import com.sleepycat.db.EnvironmentConfig;
<strong class="userinput"><code>import com.sleepycat.db.EventHandler;</code></strong>
import com.sleepycat.db.LockMode;
import com.sleepycat.db.OperationStatus;
<strong class="userinput"><code>import com.sleepycat.db.ReplicationHandleDeadException;</code></strong>
import com.sleepycat.db.ReplicationHostAddress;
import com.sleepycat.db.ReplicationManagerSiteConfig;
import db.repquote.RepConfig;
<strong class="userinput"><code>import db.repquote.RepQuoteEnvironment</code></strong>
public class RepQuoteExampleGSG <strong class="userinput"><code> implements EventHandler</code></strong>
{
private RepConfig repConfig;
private <strong class="userinput"><code>RepQuoteEnvironment</code></strong> dbenv; </pre>
<p>
That done, we can skip the
<span><code class="methodname">main()</code> method and
our class constructor, because they do not change.</span>
Instead, we skip down to our
<span><code class="methodname">init()</code> method
where we take care of opening our environment and setting
the event handler. </span>
</p>
<p>
To update our <code class="methodname">init()</code> method, we only need
to do a couple of things. First, we identify the current class as
the event handler. Then, when we open our environment, we
instantiate a <code class="classname">RepQuoteEnvironment</code>
class instead of an <code class="classname">Environment</code>
class.
</p>
<pre class="programlisting"> public int init(RepConfig config)
throws DatabaseException
{
int ret = 0;
repConfig = config;
EnvironmentConfig envConfig = new EnvironmentConfig();
envConfig.setErrorStream(System.err);
envConfig.setErrorPrefix(RepConfig.progname);
envConfig.addReplicationManagerSite(repConfig.getThisHost());
for (ReplicationHostAddress host =
repConfig.getFirstOtherHost(); host != null;
host = repConfig.getNextOtherHost()){
ReplicationManagerSiteConfig repmgrRemoteSiteConfig =
new ReplicationManagerSiteConfig(host.host, host.port);
repmgrRemoteSiteConfig.setBootstrapHelper(true);
envConfig.addReplicationManagerSite(
repmgrRemoteSiteConfig);
}
envConfig.setReplicationPriority(repConfig.priority);
envConfig.setReplicationManagerAckPolicy(
ReplicationManagerAckPolicy.ALL);
envConfig.setCacheSize(RepConfig.CACHESIZE);
envConfig.setTxnNoSync(true);
<strong class="userinput"><code>envConfig.setEventHandler(this);</code></strong>
envConfig.setAllowCreate(true);
envConfig.setRunRecovery(true);
envConfig.setThreaded(true);
envConfig.setInitializeReplication(true);
envConfig.setInitializeLocking(true);
envConfig.setInitializeLogging(true);
envConfig.setInitializeCache(true);
envConfig.setTransactional(true);
envConfig.setVerboseReplication(appConfig.verbose);
try {
dbenv = new <strong class="userinput"><code>RepQuoteEnvironment</code></strong>(repConfig.getHome(),
envConfig);
} catch(FileNotFoundException e) {
System.err.println("FileNotFound exception: " + e.toString());
System.err.println(
"Ensure that the environment directory is pre-created.");
ret = 1;
}
// start Replication Manager
dbenv.replicationManagerStart(3, repConfig.startPolicy);
return ret;
} </pre>
<p>
That done, we need to implement the methods
required for responding to replication events.
These methods are required because we are now implementing
<code class="classname">com.sleepycat.db.EventHandler</code>. While
we are required to provide an implementation for all of these
methods, for our simple application we are really only interested
in these because they allow us to track whether a replication
instance is a master.
</p>
<pre class="programlisting"> public void handleRepClientEvent()
{
dbenv.setIsMaster(false);
}
public void handleRepConnectBrokenEvent()
{
// Ignored for now.
}
public void handleRepConnectEstablishedEvent()
{
// Ignored for now.
}
public void handleRepConnectTryFailedEvent()
{
// Ignored for now.
}
public void handleRepMasterEvent()
{
dbenv.setIsMaster(true);
}
public void handleRepNewMasterEvent(int envId)
{
// Ignored for now.
}
public void handleWriteFailedEvent(int errorCode)
{
System.err.println("Write to stable storage failed!" +
"Operating system error code:" + errorCode);
System.err.println("Continuing....");
}
public void handleRepStartupDoneEvent()
{
// Ignored for now.
}
public void handleRepPermFailedEvent()
{
// Ignored for now.
}
public void handleRepLocalSiteRemovedEvent()
{
// Ignored for now.
}
public void handleRepSiteAddedEvent()
{
// Ignored for now.
}
public void handleRepSiteRemovedEvent()
{
// Ignored for now.
}
public void handleRepElectedEvent()
{
// Safely ignored for Replication Manager applications.
}
public void handleRepElectionFailedEvent()
{
// Safely ignored for Replication Manager applications that do
// not manage their own master selection.
}
public void handleRepJoinFailureEvent()
{
// Safely ignored since this application did not turn off AUTOINIT.
}
public void handleRepMasterFailureEvent()
{
// Safely ignored for Replication Manager applications that do
// not manage their own master selection.
}
public void handleRepDupmasterEvent()
{
// Safely ignored for Replication Manager applications that do
// not manage their own master selection.
}
public void handlePanicEvent()
{
System.err.println("Panic encountered!");
System.err.println("Shutting down.");
System.err.println("You should restart, running recovery.");
try {
terminate();
} catch (DatabaseException dbe) {
System.err.println("Caught an exception during " +
"termination in handlePanicEvent: " + dbe.toString());
}
System.exit(-1);
} </pre>
<p>
That done, we need to update our
<code class="function">doloop()</code>
<span>method.</span>
</p>
<p>
We begin by updating our <code class="classname">DatabaseConfig</code>
instance to determine which options to use, depending on whether the
application is running as a master.
</p>
<pre class="programlisting"> public int doloop()
throws DatabaseException
{
Database db = null;
for (;;)
{
if (db == null) {
DatabaseConfig dbconf = new DatabaseConfig();
dbconf.setType(DatabaseType.BTREE);
<strong class="userinput"><code>if (dbenv.getIsMaster()) {
dbconf.setAllowCreate(true);
}</code></strong>
dbconf.setTransactional(true); </pre>
<p>
When we open the database, we modify our error handling to
account for the case where the database does not yet exist. This can
happen if our code is running as a replica and the Replication Manager has not
yet had a chance to create the databases for us. Recall that replicas never
write to their own databases directly, and so they cannot
create databases on their own.
</p>
<p>
If we detect that the database does not yet exist, we simply
close the database handle, sleep for a short period of time
and then continue processing. This gives the Replication Manager a chance to
create the database so that our replica can continue
operations.
</p>
<pre class="programlisting"> try {
db = dbenv.openDatabase
(null, RepConfig.progname, null, dbconf);
} catch (java.io.FileNotFoundException e) {
<strong class="userinput"><code>System.err.println("no stock database available yet.");
if (db != null) {
db.close(true);
db = null;
}
try {
Thread.sleep(RepConfig.SLEEPTIME);
} catch (InterruptedException ie) {}
continue;</code></strong>
}
} </pre>
<p>
Next we modify our prompt, so that if the local process is running
as a replica, we can tell from the shell that the prompt is for a
read-only process.
</p>
<pre class="programlisting"> BufferedReader stdin =
new BufferedReader(new InputStreamReader(System.in));
// listen for input, and add it to the database.
System.out.print("QUOTESERVER");
<strong class="userinput"><code>if (!dbenv.getIsMaster())
System.out.print("(read-only)");
System.out.print("&gt; ");</code></strong>
System.out.flush();
String nextline = null;
try {
nextline = stdin.readLine();
} catch (IOException ioe) {
System.err.println("Unable to get data from stdin");
break;
}
String[] words = nextline.split("\\s"); </pre>
<p>
When we collect data from the prompt, there is a case that says
if no data is entered then show the entire stocks database.
This display is performed by our
<code class="function">print_stocks()</code>
<span>method</span>
(which has not
required a modification since we first introduced it in
<a class="xref" href="simpleprogramlisting.html#printstocks_c" title="Method: SimpleTxn.printStocks()">
<span>Method: SimpleTxn.printStocks()</span>
</a>).
</p>
<p>
When we call
<span><code class="function">printStocks()</code>, </span>
we check for a dead replication handle. Dead
replication handles happen whenever a replication election
results in a previously committed transaction becoming
invalid. This is an error scenario caused by a new master having a
slightly older version of the data than the original
master and so all replicas must modify their database(s) to
reflect that of the new master. In this situation, some
number of previously committed transactions may have to be
unrolled. From the replica's perspective, the database
handles should all be closed and then opened again.
</p>
<pre class="programlisting"> // A blank line causes the DB to be dumped to stdout.
if (words.length == 0 ||
(words.length == 1 &amp;&amp; words[0].length() == 0)) {
try {
printStocks(db);
<strong class="userinput"><code>} catch (DeadlockException de) {
continue;
// Dead replication handles are caused by an election
// resulting in a previously committing read becoming
// invalid. Close the db handle and reopen.
} catch (ReplicationHandleDeadException rhde) {
db.close(true); // close no sync.
db = null;
continue;</code></strong>
} catch (DatabaseException e) {
System.err.println("Got db exception reading " +
"replication DB: " + e.toString());
break;
}
continue;
}
if (words.length == 1 &amp;&amp;
(words[0].compareToIgnoreCase("quit") == 0 ||
words[0].compareToIgnoreCase("exit") == 0)) {
break;
} else if (words.length != 2) {
System.err.println("Format: TICKER VALUE");
continue;
} </pre>
<p>
That done, we need to add a little error checking to our
command prompt to make sure the user is not attempting to
modify the database at a replica. Remember, replicas must never
modify their local databases on their own. This guards against
that happening due to user input at the prompt.
</p>
<pre class="programlisting"> <strong class="userinput"><code>if (!dbenv.getIsMaster()) {
System.err.println("Can't update client.");
continue;
}</code></strong>
DatabaseEntry key = new DatabaseEntry(words[0].getBytes());
DatabaseEntry data = new DatabaseEntry(words[1].getBytes());
db.put(null, key, data);
}
if (db != null)
db.close(true);
return 0;
} </pre>
<p>
With that completed, we are all done updating our application
for replication.
The only remaining
<span>method, <code class="function">printStocks()</code>,</span>
is unmodified from when we
originally introduced it. For details on that function, see
<a class="xref" href="simpleprogramlisting.html#printstocks_c" title="Method: SimpleTxn.printStocks()">
<span>Method: SimpleTxn.printStocks()</span>
</a>.
</p>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="runningit"></a>Running It</h3>
</div>
</div>
</div>
<p>
To run our replicated application, we need to make
sure each participating environment has its own unique
home directory. We can do this by running
each site on a separate networked machine, but that
is not strictly necessary; multiple instances of this
code can run on the same machine provided the
environment home restriction is observed.
</p>
<p>
To run a process, make sure the environment home exists and
then start the process using the <code class="literal">-h</code>
option to specify that directory. You must also use the
<code class="literal">-l</code> or <code class="literal">-L</code> option to
identify the local host and port that this process will use
to listen for replication messages (-L means that this is a
group creator), and the <code class="literal">-r</code> option to
identify the other processes in the replication group.
Finally, use the <code class="literal">-p</code> option to specify a
priority. The process that you designate to have the
highest priority will become the master.
</p>
<pre class="programlisting">&gt; mkdir env1
&gt; java db.repquote_gsg.RepQuoteExampleGSG -h env1 -L localhost:8080 \
-p 10
No stock database yet available.
No stock database yet available. </pre>
<p>
Now, start another process. This time, change the environment
home to something else, use the <code class="literal">-l</code> flag to at
least change the port number the process is listening on, and
use the <code class="literal">-r</code> option to identify the host and
port of the other replication process:
</p>
<pre class="programlisting">&gt; mkdir env2
&gt; java db.repquote_gsg.RepQuoteExampleGSG -h env2 -l localhost:8081 \
-r localhost:8080 -p 20</pre>
<p>
After a short pause, the second process should display the master
prompt:
</p>
<pre class="programlisting">
QUOTESERVER &gt; </pre>
<p>
And the first process should
display the read-only prompt:
</p>
<pre class="programlisting">
QUOTESERVER (read-only)&gt; </pre>
<p>
Now go to the master process and give it a couple of stocks and stock
prices:
</p>
<pre class="programlisting">QUOTESERVER&gt; FAKECO 9.87
QUOTESERVER&gt; NOINC .23
QUOTESERVER&gt; </pre>
<p>
Then, go to the replica and hit <strong class="userinput"><code>return</code></strong> at the prompt to
see the new values:
</p>
<pre class="programlisting">QUOTESERVER (read-only)&gt;
Symbol Price
====== =====
FAKECO 9.87
NOINC .23
QUOTESERVER (read-only)&gt; </pre>
<p>
Doing the same at the master results in the same thing:
</p>
<pre class="programlisting">QUOTESERVER&gt;
Symbol Price
====== =====
FAKECO 9.87
NOINC .23
QUOTESERVER&gt; </pre>
<p>
You can change a stock by simply entering the stock value and
new price at the master's prompt:
</p>
<pre class="programlisting">QUOTESERVER&gt; FAKECO 10.01
QUOTESERVER&gt; </pre>
<p>
Then, go to either the master or the replica to see the updated
database. On the master:
</p>
<pre class="programlisting">QUOTESERVER&gt;
Symbol Price
====== =====
FAKECO 10.01
NOINC .23
QUOTESERVER&gt; </pre>
<p>
And on the replica:
</p>
<pre class="programlisting">QUOTESERVER (read-only)&gt;
Symbol Price
====== =====
FAKECO 10.01
NOINC .23
QUOTESERVER (read-only)&gt; </pre>
<p>
Finally, to quit the applications, simply type
<code class="literal">quit</code> at both prompts. On the replica:
</p>
<pre class="programlisting">QUOTESERVER (read-only)&gt; quit
&gt; </pre>
<p>
And on the master as well:
</p>
<pre class="programlisting">QUOTESERVER&gt; quit
&gt; </pre>
</div>
</div>
<div class="navfooter">
<hr />
<table width="100%" summary="Navigation footer">
<tr>
<td width="40%" align="left"><a accesskey="p" href="processingloop.html">Prev</a> </td>
<td width="20%" align="center">
<a accesskey="u" href="fwrkmasterreplica.html">Up</a>
</td>
<td width="40%" align="right"> <a accesskey="n" href="addfeatures.html">Next</a></td>
</tr>
<tr>
<td width="40%" align="left" valign="top">Processing Loop </td>
<td width="20%" align="center">
<a accesskey="h" href="index.html">Home</a>
</td>
<td width="40%" align="right" valign="top"> Chapter 5. Additional Features</td>
</tr>
</table>
</div>
</body>
</html>