configure_mat.dita

<?xml version="1.0" encoding="UTF-8"?>
<!--
    Copyright (c) 2008, 2023 SAP AG and IBM Corporation.
    All rights reserved. This program and the accompanying materials
    are made available under the terms of the Eclipse Public License 2.0
    which accompanies this distribution, and is available at
    https://www.eclipse.org/legal/epl-2.0/
    
    SPDX-License-Identifier: EPL-2.0
   
    Contributors:
        SAP AG - initial API and implementation
        IBM Corporation - preference page and OQL
 -->
<!DOCTYPE task PUBLIC "-//OASIS//DTD DITA Task//EN" "task.dtd" >
<task id="task_configure_mat" xml:lang="en-us">
	<title>Memory Analyzer Configuration</title>
	<prolog>
		<copyright>
			<copyryear year=""></copyryear>
			<copyrholder>
				Copyright (c) 2008, 2023 SAP AG and others.
			    All rights reserved. This program and the accompanying materials
			    are made available under the terms of the Eclipse Public License 2.0
			    which accompanies this distribution, and is available at
			    https://www.eclipse.org/legal/epl-2.0/
			</copyrholder>
		</copyright>
	</prolog>

	<taskbody>
		<context>
			<p>
				Well, analyzing big heap dumps can also require more heap space.
				Give it
				some more memory (possible by running on a 64-bit machine):
			</p>

			<codeph>MemoryAnalyzer.exe -vmargs -Xmx4g</codeph>

			<p>Alternatively, edit the MemoryAnalyzer.ini to
				contain:</p>

			<p>
				<codeblock>-vmargs
-Xmx4g</codeblock>
			</p>

			<p>
				For more details, check out the section
				<xref format="html" scope="peer"
					href="http://help.eclipse.org/topic/org.eclipse.platform.doc.user/tasks/running_eclipse.htm">Running Eclipse</xref>
				in the Help Center. It also contains more details if you are running
				on Mac OS X.
			</p>

			<p>If you are running the Memory Analyzer inside your Eclipse SDK,
				you need to edit the eclipse.ini file.</p>

			<p>The memory intensive parts are the parsing of the dump and building
				of the dominator tree. Try parsing the heap
				dump from the command line, perhaps on a bigger machine. The dumps
				and index files can then be copied to a more convenient machine. 
				Once the dump has been parsed, it usually can
				be opened with less memory in the GUI.
				As a rough estimate if the number of objects is N and the number of 
				classes C, it might take at least T bytes to parse and build the dominator tree
				where:
				<codeph>T &#8776; N * 28.25 + C * 1000 + P</codeph>
				P is the space used by the DTFJ or HPROF parsers. For a PHD file, the space could be:
				<codeph>P &#8776; C * 1000</codeph>
				Memory Analyzer uses additional memory for caching index files, so
				performance will be better if there is more memory available than the minimum
				required to parse a dump.
				</p>
				<p>Memory Analyzer has an architectural limit of 2<sup>31</sup> - 3 objects,
				a current limit of 2<sup>31</sup> - 9 = 2,147,483,639 objects, but has not been
				tested with that many objects. The current record is a heap dump file of 159Gbytes
				containing 2,041,300,061 objects, which was opened with Memory Analyzer running
				with a 172Gbyte heap.
				Exceeding the limit can result in an exception such as
				<systemoutput>java.lang.OutOfMemoryError: Requested length of new long[2,147,483,640] exceeds limit of 2,147,483,639</systemoutput>.
				See <xref format="dita" href="#task_configure_mat/discard">enable discard</xref>
				for options to work around this limit.</p>
				
			<p>The preference page is opened via a menu option.
				<menucascade>
					<uicontrol>Window</uicontrol>
					<uicontrol>Preferences</uicontrol>
				</menucascade>
			</p>
			<p id="preferences">The preferences page allows various aspects of Memory Analyzer to be controlled.
			Use the <xref format="html" scope="peer" href="javascript:executeCommand(%22org.eclipse.ui.window.preferences(preferencePageId=org.eclipse.mat.ui.Preferences)%22)">
			<b>Preferences</b></xref> dialog pages to set how you want Eclipse Memory Analyzer to operate. The menu path is 
				<menucascade>
					<uicontrol>Window</uicontrol>
					<uicontrol>Preferences</uicontrol>
				</menucascade>
			on Windows and Linux -
			on macOS the Preferences can be found in the "App" menu.
			Some of the preferences can also be controlled from the command line in
			<xref type="task" href="batch.dita">batch mode</xref>.
			<image href="../mimes/preferences.png">
				<alt>preferences page showing general options with DTFJ subpage</alt>
			</image>
			<dl>
			<dlentry>
				<dt>Keep unreachable objects</dt>
				<dd>Objects that appear unreachable from GC roots are not discarded in an early stage of 
				Memory Analyzer processing, but are retained for further analysis.</dd>
			</dlentry>
			<dlentry>
				<dt>Hide the getting started wizard</dt>
				<dd>Controls whether to display a wizard for leak suspects, top components after opening
				a snapshot.</dd>
			</dlentry>
			<dlentry>
				<dt>Hide popup query help</dt>
				<dd>Do not display the help panel underneath the query wizard, unless F1 or the help button
				is pressed.</dd>
			</dlentry>
			<dlentry>
				<dt>Hide the Welcome screen on launch</dt>
				<dd>The welcome page has the 'Overview' and 'Tutorials' tabs and the 'Workbench' button.
				The welcome page is closed by going to the workbench.
				Selecting this option means that when Memory Analyzer is started the workbench is displayed
				first. 
				The welcome page can be reopened from the workbench by:
					<menucascade>
						<uicontrol>Help</uicontrol>
						<uicontrol>Welcome</uicontrol>
					</menucascade>
				</dd>
			</dlentry>
			<dlentry>
				<dt>Bytes Display</dt>
				<dd>
					<p>
						There is a option (from MAT 1.5 onwards) to display bytes in B, KB, MB, GB, or Smart formats.
						The default is to always display in Bytes format to match the previous MAT behavior and not cause any confusion.
						The option can be changed in the Eclipse preference dialog or using <option>-Dbytes_display=(bytes|kilobytes|megabytes|gigabytes|smart)</option>.
					</p>
					<image href="../mimes/size_units_cfg.png">
						<alt>Preferences dialog to configure size units</alt>
					</image>
					<dl>
						<dlentry>
							<dt>Bytes (B)</dt>
							<dd>Memory counted in single bytes</dd>
						</dlentry>
						<dlentry>
							<dt>Kilobytes (KB)</dt>
							<dd>Memory counted in units of 1,024 bytes</dd>
						</dlentry>
						<dlentry>
							<dt>Megabytes (MB)</dt>
							<dd>Memory counted in units of 1,048,576 bytes</dd>
						</dlentry>
						<dlentry>
							<dt>Gigabytes (GB)</dt>
							<dd>Memory counted in units of 1,073,741,824‬ bytes</dd>
						</dlentry>
						<dlentry>
							<dt>Smart</dt>
							<dd>In Smart mode, if the value is a gigabyte or more, display in gigabytes; similarly for megabytes and kilobytes; otherwise, display in bytes.
								<image href="../mimes/size_units_sample.png">
									<alt>A histogram showing entries with different size units</alt>
								</image>
							</dd>
						</dlentry>
					</dl>
				</dd>
			</dlentry>
			<dlentry id="expand">
				<dt>Expanded entries in tables and trees</dt>
				<dd>There can be many rows in the tables and trees produced by queries.
				This controls the number of new entries displayed after a double click
				on the totals row, or for the <menucascade><uicontrol>Next <option>N</option>n</uicontrol></menucascade> menu item
				or for expanding a tree entry
				or for the number of rows for a table or tree in a report,
				if not set by the <xref format="html" scope="peer" href="../doc/org/eclipse/mat/report/Params.Rendering.html#LIMIT"><parmname>limit</parmname></xref>.
				</dd>
			</dlentry>
			<dlentry id="discard">
				<dt>Enable discard (experimental)</dt>
				<dd>
					<p>
					Sometimes a heap dump is generated with more objects than Memory Analyzer
					can handle, either from lack of heap to run Memory Analyzer itself, or because the number
					exceeds the Memory Analyzer limit of 2,147,483,639 objects.
					This option controls some experimental settings to help analyze
					such huge dumps, by purposely discarding objects in the original
					heap dump.
					</p>
					<p>The discarded objects are counted in the 
					<xref format="dita" href="../reference/inspections/unreachable_objects.dita">
					unreachable objects histogram</xref> together with any unreachable objects
					discarded by Memory Analyzer after parsing but before building the dominator tree.
					</p>
					<p>The discarded objects are not indexed, so are not part of the dominator tree
					nor are they counted as part of the retained size. As they are not indexed they
					will not appear in the <xref format="dita" href="../reference/querymatrix.dita#ref_querymatrix/list_objects">list objects</xref>
					query.
					Most other queries will not operate on unindexed (discarded) objects.
					Sometimes it is possible to view them. The inspector view can display
					unindexed objects in the attributes and statics tabs.
					It is possible to 'Go Into' an unindexed object from the inspector view, and
					'Go Back' to the previous object.
					Unindexed objects do not have GC root information, so in that
					position in the inspector view is displayed a 
					<image href="../mimes/message_warning.png">
						<alt>yellow warning triangle icon</alt>
					</image>
					warning triangle
					and the message <msgph>Unindexed</msgph>.
					Names resolvers can use the values of unindexed objects to find a displayable
					name for an object. This will be displayed after the object address
					in the inspector view and in the 
					<xref format="dita" href="../reference/tipsandtricks.dita#ref_tips/value_tab">Inspector View value tab</xref>.					
					</p>
					<p>Some trial and error may be required to find suitable values of the options.
					<ol>
					<li>If an OutOfMemoryError still occurs on a parse then more objects need to be
					discarded, either by increasing the discard percentage or by increasing the
					number of types of discarded objects by changing the pattern.</li>
					<li>If the leak is
					not apparent from the snapshot with many discarded objects then examine
					the unreachable objects histogram to see if any key objects have been discarded
					and modify the discard pattern. Ideally only primitive arrays or objects which
					just have primitive fields or refer to primitive arrays should be discarded.
					This avoids disrupting the object graph too much.</li>
					<li>If the types of the discarded objects look reasonable then try changing
					which objects have been discarded, either by varying the offset (if the
					discard percentage is not <userinput>100</userinput>) or the discard seed.</li>
					<li>If the available heap to run Memory Analyzer is very small then try
					discarding all of the ordinary objects by choosing <userinput>100</userinput>
					for the discard percentage and <userinput>.*</userinput> to match all
					object types. The resulting snapshot will not be useful for finding leaks,
					but the unreachable object histogram will show the types of the objects
					taking most of the heap space, and might give a hint as to the problem.</li>
					</ol>
					See <xref format="dita" href="batch.dita#task_batch/discard">
					batch mode discard options</xref> for controlling discard in batch
					mode.
					See <xref format="dita" href="../reference/oqlsyntaxfrom.dita#ref_oqlsyntaxfrom/unindexed">
					OQL FROM clauses</xref> for how to use OQL with unindexed objects. 
					</p>
					<note>This feature is experimental and is subject to change
					or removal. Please respond on the 
					<xref format="html" scope="peer"
						href="http://www.eclipse.org/forums/index.php/f/186/">
					Memory Analyzer forum</xref> if you try it and
					find it useful or have suggestions.</note>
			<dl>
			<dlentry>
				<dt>Discard percentage</dt>
				<dd>
					A number between <userinput>0</userinput> and <userinput>100</userinput>, treated as a percentage.
					Approximately this percentage of ordinary objects 
					matching the discard pattern will be
					discarded by the HPROF or DTFJ parsers.
				</dd>
			</dlentry>
			<dlentry>
				<dt>Discard pattern</dt>
				<dd>
					Only objects with a class name matching this regular expression
					will be discarded. It is best to chose objects of a type which
					does not link to other objects, such as primitive arrays, or
					objects which just link to other such objects. This avoids breaking
					the object graph too much, and gives a hope that the leak
					analysis will find the problem.
					The default pattern of
					<userinput>char\[\]|java\.lang\.String</userinput> discards
					character arrays and Strings. For Java 9 heap dumps with
					compact strings based on byte arrays then 
					<userinput>byte\[\]|java\.lang\.String</userinput> may be 
					more appropriate.
				</dd>
			</dlentry>
			<dlentry>
				<dt>Discard offset</dt>
				<dd>
					A number between <userinput>0</userinput> and <userinput>99</userinput>.
					This allows different objects to be discarded from two different
					parses of the same heap dump. This might be useful to allow objects
					which have been discarded from one snapshot to be found in the
					others. For example with a discard percentage of <userinput>25</userinput>, then offsets
					chosen from <userinput>0</userinput>,<userinput>25</userinput>,<userinput>50</userinput>,<userinput>75</userinput> in successive parses will discard 
					4 distinct sets of objects.
					With a discard percentage of <userinput>80</userinput>, and offsets of 
					<userinput>0</userinput>,
					<userinput>20</userinput>,
					<userinput>40</userinput>,
					<userinput>60</userinput>,
					<userinput>80</userinput>
					in successive parses, will have each object present in one of 
					those five snapshots.
				</dd>
			</dlentry>
			<dlentry>
				<dt>Discard seed</dt>
				<dd>
					This controls the starting point of the random number generator
					used to decide whether to discard particular objects. If after 
					discarding objects the leak cannot be found then rerunning the
					parse with a different seed would discard some different objects
					and might show the leak better.
				</dd>
			</dlentry>
			</dl>
				</dd>
			</dlentry>
			<dlentry id="dtfj_preferences">
				<dt>
					<menucascade>
						<uicontrol>Memory Analyzer</uicontrol>
						<uicontrol>DTFJ Parser</uicontrol>
						<uicontrol>Stack frames</uicontrol>
					</menucascade>
				</dt>
				<dd>
					Normally stack frames are just shown in the Thread overview and stacks query.
					This option allows stack frames to be treated as objects which are then visible like any other Java object in a Memory Analyzer view.
					<dl>
					<dlentry>
					<dt>Normal
					</dt>
					<dd>Stack frames are just shown in the Thread overview and stacks query.
					</dd>
					</dlentry>
					<dlentry>
					<dt>Only stack frames as pseudo-objects
					</dt>
					<dd>
					Stack frames are of type <codeph>&lt;stack frame&gt;</codeph>, size 0, with method names and source file
and line numbers via fields and a name resolver.
					<p>This is useful when looking at paths to and from objects via local variables as the stack frames are visible in
					the paths to GC roots queries.</p>
					</dd>
					</dlentry>
					<dlentry>
					<dt>Stack frames as pseudo-objects and running methods as pseudo-classes
					</dt>
					<dd>Stack frames of are of type <codeph>packageName.className.methodName(Signature)ReturnType</codeph> extending <codeph>&lt;method&gt;</codeph> representing the method being executed,
of size the stack frame size, with source file and line number via fields and a name resolver,
and those methods are pseudo-classes of type <codeph>&lt;method type&gt;</codeph> of size 0.
					<p>This can be useful to find out which methods are currently running and how much stack space they take up.
					To examine running methods then take the histogram view, filter by '\(', then sort by instances or instance size.
					</p>
					</dd>
					</dlentry>
					<dlentry>
					<dt>Stack frames as pseudo-objects and all methods as pseudo-classes
					</dt>
					<dd>Stack frames of are of type <codeph>packageName.className.methodName(Signature)ReturnType</codeph> extending <codeph>&lt;method&gt;</codeph>codeph> representing the method being executed,
of size the stack frame size, with source file and line number via fields and a name resolver,
and all methods are pseudo-classes of type <codeph>&lt;method type&gt;</codeph> of size based on the JITted and byte code sizes. The method sizes are then not part of the class size.
					<p>This can be useful to find out which methods have large JITted or byte code sizes. They can be viewed by going to the histogram view,
					then selecting <codeph>&lt;method type&gt;</codeph> and listing objects.
					</p>
					</dd>
					</dlentry>
					</dl>
					
				</dd>
			</dlentry>
			<dlentry>
				<dt>
					<menucascade>
						<uicontrol>Memory Analyzer</uicontrol>
						<uicontrol>DTFJ Parser</uicontrol>
						<uicontrol>Suppress Class native memory size calculations (IBM system dumps)</uicontrol>
					</menucascade>
				</dt>
				<dd>
					By default, when MAT parses IBM system dumps, the size of classes includes some of the amount of native memory in the Java process (but outside of the Java heap) which is related to those classes such as native memory for bytecode and JIT compiled code for the class methods. Check this option to disable this calculation and only report Java heap usage.
				</dd>
			</dlentry>
					<dlentry id="hprof_preferences">
						<dt>
							<menucascade>
								<uicontrol>Memory Analyzer</uicontrol>
								<uicontrol>HPROF Parser</uicontrol>
								<uicontrol>Parser Strictness</uicontrol>
							</menucascade>
						</dt>
						<dd>
							By default, the HPROF parser runs in strict mode. This means that
							any deviation in the file from the
							<xref format="html" scope="external"
								href="http://java.net/downloads/heap-snapshot/hprof-binary-format.html">HPROF specification</xref>
							is treated as a severe error, an exception is thrown, and dump
							processing stops. There is one exception to this rule which is
							that we can reliably workaround the observed issue in
							<xref format="html" scope="external"
								href="https://bugs.eclipse.org/bugs/show_bug.cgi?id=404679">bug #404679</xref>
							.

							<p>This
								is a change in behavior from previous
								releases when a
								warning
								was
								shown in the error log and processing
								continued. This
								default
								change was made to alert the user to a
								potential problem
								either
								with the file itself or a bug in the JVM
								or in MAT. You may
								choose
								to change the strictness of the parser:</p>
							<dl>
								<dlentry>
									<dt>Strict
									</dt>
									<dd>Default. Any deviation from the HPROF specification causes
										an exception to be thrown and processing to stop (with the
										exception of
										<xref format="html" scope="external"
								href="https://bugs.eclipse.org/bugs/show_bug.cgi?id=404679">bug #404679</xref>
										.) This option may be specified on the
										command line with -DhprofStrictnessStop=true
									</dd>
								</dlentry>
								<dlentry>
									<dt>Warning
									</dt>
									<dd>
										Choose this option to revert to the old behavior where a
										warning is printed to the error log and processing continues.
										This option is not recommended because the warning is probably
										a sign of a problem. Please open a bug report instead. This
										option may be specified on the
										command line with
										-DhprofStrictnessWarning=true
									</dd>
								</dlentry>
							</dl>

						</dd>
					</dlentry>
			<dlentry>
				<dt>
					<menucascade>
						<uicontrol>Memory Analyzer</uicontrol>
						<uicontrol>HPROF Parser</uicontrol>
						<uicontrol>Include additional references from class objects 
						found in a HPROF file.</uicontrol>
					</menucascade>
				</dt>
				<dd>
					Class objects have hidden references to the other objects.
					These include the signers and the protection domain.
					Memory Analyzer can include them via pseudo-fields named
					<parmname>&lt;signers&gt;</parmname> and <parmname>&lt;protectionDomain&gt;</parmname>.
					Sometimes it might be useful to see these references, and this
					option allows that.
				</dd>
			</dlentry>
			<dlentry>
				<dt>
					<menucascade>
						<uicontrol>Memory Analyzer</uicontrol>
						<uicontrol>HPROF Parser</uicontrol>
						<uicontrol>Stack frames</uicontrol>
					</menucascade>
				</dt>
				<dd>
					Normally stack frames are just shown in the Thread overview and stacks query.
					This option allows stack frames to be treated as objects which are then visible like any other Java object in a Memory Analyzer view.
					<dl>
					<dlentry>
					<dt>Normal
					</dt>
					<dd>Stack frames are just shown in the Thread overview and stacks query.
					</dd>
					</dlentry>
					<dlentry>
					<dt>Only stack frames as pseudo-objects
					</dt>
					<dd>
					Stack frames are of type <codeph>&lt;stack frame&gt;</codeph>, size 0, with method names and source file
and line numbers via fields and a name resolver.
					<p>This is useful when looking at paths to and from objects via local variables as the stack frames are visible in
					the paths to GC roots queries.</p>
					</dd>
					</dlentry>
					<dlentry>
					<dt>Stack frames as pseudo-objects and running methods as pseudo-classes
					</dt>
					<dd>Stack frames of are of type <codeph>packageName.className.methodName(Signature)ReturnType</codeph> extending <codeph>&lt;method&gt;</codeph> representing the method being executed,
of size an estimate of the stack frame size, with source file and line number via fields and a name resolver,
and those methods are pseudo-classes of type <codeph>&lt;method type&gt;</codeph> of size 0.
The size of a stack frame is estimated as 8 bytes times the maximum number of object references from a frame of that type.
					<p>This can be useful to find out which methods are currently running and how much stack space they take up.
					To examine running methods then take the histogram view, filter by '\(', then sort by instances or instance size.
					</p>
					</dd>
					</dlentry>
					</dl>
					
				</dd>
			</dlentry>
			<dlentry>
				<dt>
					<menucascade>
						<uicontrol>General</uicontrol>
						<uicontrol>Appearance</uicontrol>
						<uicontrol>Colors and Fonts</uicontrol>
						<uicontrol>Memory Analyzer</uicontrol>
						<uicontrol>OQL comment color</uicontrol>
					</menucascade>
				</dt>
				<dd>Modify the color of comments in the Object Query Language (OQL) studio.</dd>
			</dlentry>
			<dlentry>
				<dt>
					<menucascade>
						<uicontrol>General</uicontrol>
						<uicontrol>Appearance</uicontrol>
						<uicontrol>Colors and Fonts</uicontrol>
						<uicontrol>Memory Analyzer</uicontrol>
						<uicontrol>OQL keyword color</uicontrol>
					</menucascade>
				</dt>
				<dd>Modify the color of keywords in the Object Query Language (OQL) studio.</dd>
			</dlentry>
			</dl>
			</p>
		</context>
	</taskbody>
</task>