lockingsubsystem.html [plain text]
<?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>The Locking Subsystem</title>
<link rel="stylesheet" href="gettingStarted.css" type="text/css" />
<meta name="generator" content="DocBook XSL Stylesheets V1.62.4" />
<link rel="home" href="index.html" title="Getting Started with Berkeley DB Transaction Processing" />
<link rel="up" href="txnconcurrency.html" title="Chapter 4. Concurrency" />
<link rel="previous" href="blocking_deadlocks.html" title="Locks, Blocks, and Deadlocks" />
<link rel="next" href="isolation.html" title="Isolation" />
</head>
<body>
<div class="navheader">
<table width="100%" summary="Navigation header">
<tr>
<th colspan="3" align="center">The Locking Subsystem</th>
</tr>
<tr>
<td width="20%" align="left"><a accesskey="p" href="blocking_deadlocks.html">Prev</a> </td>
<th width="60%" align="center">Chapter 4. Concurrency</th>
<td width="20%" align="right"> <a accesskey="n" href="isolation.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="lockingsubsystem"></a>The Locking Subsystem</h2>
</div>
</div>
<div></div>
</div>
<p>
In order to allow concurrent operations, DB provides the locking
subsystem. This subsystem provides inter- and intra- process
concurrency mechanisms. It is extensively used by DB concurrent
applications, but it can also be generally used for non-DB
resources.
</p>
<p>
This section describes the locking subsystem as it is used to
protect DB resources. In particular, issues on configuration are
examined here. For information on using the locking subsystem to
manage non-DB resources, see the
<i class="citetitle">Berkeley DB Programmer's Reference Guide</i>.
</p>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="configuringlock"></a>Configuring the Locking Subsystem</h3>
</div>
</div>
<div></div>
</div>
<p>
You initialize the locking subsystem by specifying
<span>
<tt class="literal">DB_INIT_LOCK</tt> to the
<tt class="methodname">DbEnv::open()</tt>
method.
</span>
</p>
<p>
Before opening your environment, you can configure various
maximum values for your locking subsystem. Note that these
limits can only be configured before the environment is
opened. Also, these methods configure the entire environment,
not just a specific environment handle.
</p>
<p>
Finally, each bullet below identifies the
<tt class="filename">DB_CONFIG</tt> file parameter that can be used
to specify the specific locking limit. If used, these
<tt class="filename">DB_CONFIG</tt> file parameters override any
value that you might specify using the environment handle.
</p>
<p>
The limits that you can configure are as follows:
</p>
<div class="itemizedlist">
<ul type="disc">
<li>
<p>
The maximum number of lockers
supported by the environment. This value is used by
the environment when it is opened to estimate the amount
of space that it should allocate for various internal
data structures. By default, 1,000 lockers are
supported.
</p>
<p>
To configure this value, use the
<span>
<tt class="methodname">DbEnv::set_lk_max_lockers()</tt>
method.
</span>
</p>
<p>
As an alternative to this method, you can configure this
value using the <tt class="filename">DB_CONFIG</tt> file's
<tt class="literal">set_lk_max_lockers</tt> parameter.
</p>
</li>
<li>
<p>
The maximum number of locks supported by the environment.
By default, 1,000 locks are supported.
</p>
<p>
To configure this value, use the
<span>
<tt class="methodname">DbEnv::set_lk_max_locks()</tt>
method.
</span>
</p>
<p>
As an alternative to this method, you can configure this
value using the <tt class="filename">DB_CONFIG</tt> file's
<tt class="literal">set_lk_max_locks</tt> parameter.
</p>
</li>
<li>
<p>
The maximum number of locked objects supported by the environment.
By default, 1,000 objects can be locked.
</p>
<p>
To configure this value, use the
<span>
<tt class="methodname">DbEnv::set_lk_max_objects()</tt>
method.
</span>
</p>
<p>
As an alternative to this method, you can configure this
value using the <tt class="filename">DB_CONFIG</tt> file's
<tt class="literal">set_lk_max_objects</tt> parameter.
</p>
</li>
</ul>
</div>
<p>
For a definition of lockers, locks, and locked objects, see
<a href="blocking_deadlocks.html#lockresources">Lock Resources</a>.
</p>
<p>
For example, to configure the maximum number of locks that your
environment can use:
</p>
<pre class="programlisting">#include "db_cxx.h"
...
int main(void)
{
u_int32_t env_flags = DB_CREATE | // If the environment does not
// exist, create it.
DB_INIT_LOCK | // Initialize locking
DB_INIT_LOG | // Initialize logging
DB_INIT_MPOOL | // Initialize the cache
DB_THREAD | // Free-thread the env handle.
DB_INIT_TXN; // Initialize transactions
std::string envHome("/export1/testEnv");
DbEnv myEnv(0);
try {
// Configure max locks
myEnv.set_lk_max_locks(envp, 5000);
myEnv.open(envHome.c_str(), env_flags, 0);
} catch(DbException &e) {
std::cerr << "Error opening database environment: "
<< envHome << std::endl;
std::cerr << e.what() << std::endl;
return (EXIT_FAILURE);
}
try {
myEnv.close(0);
} catch(DbException &e) {
std::cerr << "Error closing database environment: "
<< envHome << std::endl;
std::cerr << e.what() << std::endl;
return (EXIT_FAILURE);
}
return (EXIT_SUCCESS);
} </pre>
</div>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="configdeadlkdetect"></a>Configuring Deadlock Detection</h3>
</div>
</div>
<div></div>
</div>
<p>
In order for DB to know that a deadlock has occurred,
some mechanism must be used to perform deadlock
detection. There are three ways that deadlock detection can
occur:
</p>
<div class="orderedlist">
<ol type="1">
<li>
<p>
Allow DB to internally detect deadlocks as they
occur.
</p>
<p>
To do this, you use
<span><tt class="methodname">DbEnv::set_lk_detect()</tt>.</span>
This method causes DB to walk its internal lock table
looking for a deadlock whenever a lock request
is blocked. This method also identifies how DB decides which lock
requests are rejected when deadlocks are detected. For example,
DB can decide to reject the lock request for the transaction
that has the most number of locks, the least number of locks,
holds the oldest lock, holds the most number of write locks, and
so forth (see the API reference documentation for a complete
list of the lock detection policies).
</p>
<p>
You can call this method at any time during your application's
lifetime, but typically it is used before you open your environment.
</p>
<p>
Note that how you want DB to decide which thread of control should break a deadlock is
extremely dependent on the nature of your application. It is not unusual for some performance
testing to be required in order to make this determination. That said, a transaction that is
holding the maximum number of locks is usually indicative of the transaction that has performed
the most amount of work. Frequently you will not want a transaction that has performed a lot of
work to abandon its efforts and start all over again. It is not therefore uncommon for
application developers to initially select the transaction with the <span class="emphasis"><em>minimum</em></span>
number of write locks to break the deadlock.
</p>
<p>
Using this mechanism for deadlock detection means
that your application will never have to wait on a
lock before discovering that a deadlock has
occurred. However, walking the lock table every
time a lock request is blocked can be expensive
from a performance perspective.
</p>
</li>
<li>
<p>
Use a dedicated thread or external process to perform
deadlock detection. Note that this thread must be
performing no other database operations beyond deadlock
detection.
</p>
<p>
To externally perform lock detection, you can use
either the
<tt class="methodname">DbEnv::lock_detect()</tt>
method, or use the
<span><b class="command">db_deadlock</b></span> command line
utility. This method (or command) causes DB to walk the
lock table looking for deadlocks.
</p>
<p>
Note that like
<span><tt class="methodname">DbEnv::set_lk_detect()</tt>,</span>
you also use this method (or command line utility)
to identify which lock requests are rejected in the
event that a deadlock is detected.
</p>
<p>
Applications that perform deadlock detection in
this way typically run deadlock detection between every few
seconds and a minute. This means that your
application may have to wait to be notified of a
deadlock, but you also save the overhead of walking
the lock table every time a lock request is blocked.
</p>
</li>
<li>
<p>
Lock timeouts.
</p>
<p>
You can configure your locking subsystem such that
it times out any lock that is not released within a
specified amount of time. To do this, use the
<span><tt class="methodname">DbEnv::set_timeout()</tt></span>
method.
Note that lock timeouts are only checked when a
lock request is blocked or when deadlock
detection is otherwise performed. Therefore, a lock can have timed out and still be held for
some length of time until DB has a reason to examine its locking tables.
</p>
<p>
Be aware that extremely long-lived transactions, or
operations that hold locks for a long time, may be
inappropriately timed out before the transaction or
operation has a chance to complete. You should
therefore use this mechanism only if you know your
application will hold locks for very short periods
of time.
</p>
</li>
</ol>
</div>
<p>
For example, to configure your application such that DB
checks the lock table for deadlocks every time a lock
request is blocked:
</p>
<pre class="programlisting">#include "db_cxx.h"
...
int main(void)
{
u_int32_t env_flags = DB_CREATE | // If the environment does not
// exist, create it.
DB_INIT_LOCK | // Initialize locking
DB_INIT_LOG | // Initialize logging
DB_INIT_MPOOL | // Initialize the cache
DB_THREAD | // Free-thread the env handle
DB_INIT_TXN; // Initialize transactions
std::string envHome("/export1/testEnv");
DbEnv myEnv(0);
try {
// Configure db to perform deadlock detection internally, and to
// choose the transaction that has performed the least amount
// of writing to break the deadlock in the event that one
// is detected.
myEnv.set_lk_detect(DB_LOCK_MINWRITE);
myEnv.open(envHome.c_str(), env_flags, 0);
// From here, you open your databases, proceed with your
// database operations, and respond to deadlocks as
// is normal (omitted for brevity).
...</pre>
<p>
Finally, the following command line call causes
deadlock detection to be run against the
environment contained in <tt class="literal">/export/dbenv</tt>. The
transaction with the youngest lock is chosen to break the
deadlock:
</p>
<pre class="programlisting">> /usr/local/db_install/bin/db_deadlock -h /export/dbenv -a y</pre>
<p>
For more information, see the
<a href="http://www.oracle.com/technology/documentation/berkeley-db/db/utility/db_deadlock.html" target="_top">
<tt class="literal">db_deadlock</tt> reference documentation.
</a>
</p>
</div>
<div class="sect2" lang="en" xml:lang="en">
<div class="titlepage">
<div>
<div>
<h3 class="title"><a id="deadlockresolve"></a>Resolving Deadlocks</h3>
</div>
</div>
<div></div>
</div>
<p>
When DB determines that a deadlock has occurred, it will
select a thread of control to resolve the deadlock and then
<span>
throws <tt class="literal">DbDeadlockException</tt> in that
thread.
</span>
If a deadlock is detected, the thread must:
</p>
<div class="orderedlist">
<ol type="1">
<li>
<p>
Cease all read and write operations.
</p>
</li>
<li>
<p>
Close all open cursors.
</p>
</li>
<li>
<p>
Abort the transaction.
</p>
</li>
<li>
<p>
Optionally retry the operation. If your application
retries deadlocked operations, the new attempt must
be made using a new transaction.
</p>
</li>
</ol>
</div>
<div class="note" style="margin-left: 0.5in; margin-right: 0.5in;">
<h3 class="title">Note</h3>
<p>
If a thread has deadlocked, it may not make any
additional database calls using the handle that has
deadlocked.
</p>
</div>
<p>
For example:
</p>
<pre class="programlisting">// retry_count is a counter used to identify how many times
// we've retried this operation. To avoid the potential for
// endless looping, we won't retry more than MAX_DEADLOCK_RETRIES
// times.
// txn is a transaction handle.
// key and data are DBT handles. Their usage is not shown here.
while (retry_count < MAX_DEADLOCK_RETRIES) {
try {
envp->txn_begin(NULL, txn, 0);
dbp->put(txn, &key, &data, 0);
txn->commit(0);
return (EXIT_SUCCESS);
} catch (DbDeadlockException &de) {
try {
// Abort the transaction and increment the
// retry counter
txn->abort();
retry_count++;
// If we've retried too many times, log it and exit
if (retry_count >= MAX_DEADLOCK_RETRIES) {
envp->errx("Exceeded retry limit. Giving up.");
return (EXIT_FAILURE);
}
} catch (DbException &ae) {
envp->err(ae.get_errno(), "txn abort failed.");
return (EXIT_FAILURE);
}
} catch (DbException &e) {
try {
// For a generic error, log it and abort.
envp->err(e.get_errno(), "Error putting data.");
txn->abort();
} catch (DbException &ae) {
envp->err(ae.get_errno(), "txn abort failed.");
return (EXIT_FAILURE);
}
}
} </pre>
</div>
</div>
<div class="navfooter">
<hr />
<table width="100%" summary="Navigation footer">
<tr>
<td width="40%" align="left"><a accesskey="p" href="blocking_deadlocks.html">Prev</a> </td>
<td width="20%" align="center">
<a accesskey="u" href="txnconcurrency.html">Up</a>
</td>
<td width="40%" align="right"> <a accesskey="n" href="isolation.html">Next</a></td>
</tr>
<tr>
<td width="40%" align="left" valign="top">Locks, Blocks, and Deadlocks </td>
<td width="20%" align="center">
<a accesskey="h" href="index.html">Home</a>
</td>
<td width="40%" align="right" valign="top"> Isolation</td>
</tr>
</table>
</div>
</body>
</html>