[转帖]Batch mode

batch,mode · 浏览次数 : 0

小编点评

**Memory Analyzer Batch Mode Guide** **Overview** Memory Analyzer allows batch mode processing of heap dumps to analyze memory usage on multiple systems. This guide provides a comprehensive overview of the batch mode workflow, including command-line options, report generation, and other settings. **Command-Line Options** * **-keep_unreachable_objects**: Keep only reachable objects in the heap dump index. * **-remove_unreachable_objects**: Remove all unreachable objects from the index. * **-test_overview**: Test the "Heap Dump Overview" section of the results report. * **-test_problem_suspects**: Perform tests on suspicious objects in the results report. * **-generate_report**: Generate a comprehensive HTML report with detailed analysis. * **-output_directory**: Specify the output directory for the report. * **-results_file**: Specify the output file name for the results report. **Report Generation** The results are generated in an HTML report file named "myheapdump_Leak_Suspects.zip". The report includes the following sections: * **Heap Dump Overview** * **Problem Suspect Analysis** * **All Accumulated Objects by Class** **Additional Settings** * **-depth**: Specify the depth of the dominator tree for testing. * **-threshold**: Set a memory threshold for identifying suspect objects. **Example Command** ``` memory_analyzer --keep_unreachable_objects --test_problem_suspects --generate_report --output_directory=results --results_file=myheapdump_Leak_Suspects.zip ``` **Tips for Best Results** * Use a clean and isolated test environment. * Ensure that the heap dumps are valid and contain representative objects. * Select appropriate threshold values for detecting suspicious objects. * Run multiple test cases to validate the results. **Conclusion** Batch mode in Memory Analyzer provides a powerful tool for comprehensive memory analysis and troubleshooting. By understanding the command-line options and report generation settings, you can customize the analysis to meet your specific requirements.

正文

https://help.eclipse.org/latest/index.jsp?topic=%2Forg.eclipse.mat.ui.help%2Ftasks%2Fbatch.html&cp%3D54_3_14

Memory Analyzer can be operated in batch mode to process heap dumps without user input.

Note: On Windows, use .\mat\ParseHeapDump.bat instead of ./mat/ParseHeapDump.sh
Run ./mat/ParseHeapDump.sh myheapdump.hprof org.eclipse.mat.api:suspects
- For a simple suspects report use org.eclipse.mat.api:suspects See running leak suspect report for more details.
- For an overview report use org.eclipse.mat.api:overview This report is included at the beginning of the leak suspects report and shows:
  - Heap Dump Overview
  - System Properties
  - Thread Overview
  - Top Consumers
  - Class Histogram
- For a report on memory usage by component use org.eclipse.mat.api:top_components See Component report for more details of the report run on each of the top components by memory usage.
- To compare two heap dumps use ./mat/ParseHeapDump.sh myheapdump.hprof -snapshot2=myheapdump2.hprof org.eclipse.mat.api:compare
- To compare two heap dumps for memory leaks with a leak suspects report use ./mat/ParseHeapDump.sh myheapdump.hprof -baseline=myheapdump2.hprof org.eclipse.mat.api:suspects2 See running leak suspect report by snapshot comparison for more details.
- To export a new copy of the heap dump with redactions use a command like
```
./mat/ParseHeapDump.sh
myheapdump.hprof
-output=myheapdump2.hprof
-redact=BASIC
-map=myheapdump2.map
org.eclipse.mat.hprof:export
					
```
  See export heap dump for more details.
- To run a single query on a heap dump and produce an HTML report use ./mat/ParseHeapDump.sh myheapdump.hprof -command=histogram org.eclipse.mat.api:query See Query Report for more details.

Choose additional options to control the parse and report generation.

Command line options controlling the parse.

-keep_unreachable_objects means that all objects are retained in the snapshot, even if they are not reachable from an ordinary GC root.
-snapshot_identifier=identifier Selects one snapshot when there is the choice of several in the heap dump file. For HPROF files this might be of the form #1 or #2. For DTFJ system dump files this might be of the form Address space ID.Process ID.Runtime ID. If there are multiple snapshots in a heap dump file and a specific snapshot is not selected then the parse will fail with a org.eclipse.mat.snapshot.MultipleSnapshotsException which will list appropriate possible snapshot identifiers to use with this option.
Note: Experimental
-discard_ratio=percentage. Option to discard a fraction of the objects from huge dumps.
See discard configuration for an explanation of these discard options.
Note: Experimental
-discard_pattern=regex. A regular expression matching the class names of objects to be discarded.
Note: Experimental
-discard_offset=offset value. Controls which particular objects are discarded.
Note: Experimental
-discard_seed=random number generator seed. Controls which particular objects are discarded.

Other report options

Other parameters controlling output include the following:

Argument	Description
-limit=`number of entries`	Controls the number of entries generated in an output table. See org.eclipse.mat.report.Params.Rendering.LIMIT
-format=`type of output`	Controls the type of the output (HTML, CSV, or TXT). See org.eclipse.mat.report.Params.FORMAT
-html.show_table_headings	Controls whether table headings should be shown for tables. See org.eclipse.mat.report.Params.Html.SHOW_TABLE_HEADER
-html.show_totals	Controls whether totals should be shown for tables. See org.eclipse.mat.report.Params.Html.SHOW_TOTALS

Task: Parsing myheapdump.hprof
[Task: Parsing .\Downloads\myheapdump.hprof
[
Subtask: Scanning .\Downloads\myheapdump.hprof
[.................
[INFO] Detected compressed references, because with uncompressed 64-bit references the array at 0x5c4bd4258 would overlap the array at 0x5c4bd4230
[....................
[INFO] Wrote threads call stacks to .\Downloads\myheapdump.threads
[....................
[INFO] Heap .\Downloads\myheapdump.hprof contains 116,716 objects
[....................
Subtask: Extracting objects from .\Downloads\myheapdump.hprof
[...................................................................................Task: Removing unreachable objects
[
Subtask: Searching for unreachable objects
[
Subtask: Marking reachable objects
[.
Subtask: Re-indexing objects
[..
Subtask: Re-indexing classes
[...
Subtask: Writing .\Downloads\myheapdump.idx.index
[....
Subtask: Writing .\Downloads\myheapdump.o2c.index
[.....
Subtask: Writing .\Downloads\myheapdump.a2s.index
[......
Subtask: Re-indexing outbound index
[.......
Subtask: Writing .\Downloads\myheapdump.inbound.index
[........
Subtask: Writing .\Downloads\myheapdump.outbound.index
[.........
Subtask: Writing .\Downloads\myheapdump.o2hprof.index
[.........Task: Calculating Dominator Tree
[
Subtask: Dominator Tree calculation
[
Subtask: Depth-first search
[....................
Subtask: Computing dominators
[..............................................
Subtask: Calculate retained sizes
[............................................................
Subtask: Create dominators index file
[................................................................................]

Subtask: Test 'Heap Dump Overview' of section 'System Overview'
[
Subtask: Test 'System Properties' of section 'System Overview'
[
Subtask: Extracting Key Value Pairs...
[
Subtask: Test 'Thread Overview' of section 'System Overview'
[
Subtask: Searching Threads...
[
Subtask: Test 'Top Consumers' of section 'System Overview'
[
Subtask: Creating histogram
[
Subtask: Grouping by package
[
Subtask: Test 'Biggest Objects (Overview)' of section 'Top Consumers'
[
Subtask: Test 'Biggest Objects' of section 'Top Consumers'
[
Subtask: Test 'Biggest Top-Level Dominator Classes (Overview)' of section 'Top Consumers'
[
Subtask: Test 'Biggest Top-Level Dominator Classes' of section 'Top Consumers'
[
Subtask: Test 'Biggest Top-Level Dominator Class Loaders (Overview)' of section 'Top Consumers'
[
Subtask: Test 'Biggest Top-Level Dominator Class Loaders' of section 'Top Consumers'
[
Subtask: Test 'Biggest Top-Level Dominator Packages' of section 'Top Consumers'
[
Subtask: Test 'Class Histogram' of section 'System Overview'
[
Subtask: Test 'Leaks' of section 'Leak Suspects'
[
Subtask: Finding problem suspects
[
Subtask: Searching suspicious single objects ...
[
Subtask: Searching suspicious groups of objects ...
[
Subtask: building histogram
[
Subtask: Preparing results
[
Subtask: building histogram
[
Subtask: building histogram
[
Subtask: Finding paths
[
Subtask: Test 'Overview' of section 'Leaks'
[
Subtask: Test 'Problem Suspect 1' of section 'Leaks'
[
Subtask: Test 'Description' of section 'Problem Suspect 1'
[
Subtask: Test 'Shortest Paths To the Accumulation Point' of section 'Problem Suspect 1'
[
Subtask: Test 'Shortest Paths To the Accumulation Point' of section 'Problem Suspect 1'
[
Subtask: Test 'Accumulated Objects in Dominator Tree' of section 'Problem Suspect 1'
[
Subtask: Test 'Accumulated Objects in Dominator Tree' of section 'Problem Suspect 1'
[
Subtask: Test 'Accumulated Objects by Class in Dominator Tree' of section 'Problem Suspect 1'
[
Subtask: Test 'Accumulated Objects by Class in Dominator Tree' of section 'Problem Suspect 1'
[
Subtask: Test 'All Accumulated Objects by Class' of section 'Problem Suspect 1'
[
Subtask: Test 'All Accumulated Objects by Class' of section 'Problem Suspect 1'
[]

Unzip the report myheapdump_Leak_Suspects.zip
View the index.html

The result is a HTML report.