]> User's Manual: Generating Dashboard Report

JMeter supports dashboard report generation to get graphs and statistics from a test plan.
This chapter describes how to configure and use the generator.

The dashboard generator is a modular extension of JMeter. Its default behavior is to read and process samples from CSV files to generate HTML files containing graph views. It can generate the report at end of a load test or on demand.

This report provides the following metrics:

  • APDEX (Application Performance Index) table that computes for every transaction the APDEX based on configurable values for tolerated and satisfied thresholds
  • A request summary graph showing the Success and failed requests (Transaction Controller Sample Results are not taken into account) percentage:
  • A Statistics table providing in one table a summary of all metrics per transaction including 3 configurable percentiles :
  • An error table providing a summary of all errors and their proportion in the total requests :
  • A Top 5 Errors by Sampler table providing for every Sampler (excluding Transaction Controller by default) the top 5 Errors :
  • Zoomable chart where you can check/uncheck every transaction to show/hide it for:
    • Response times Over Time (Includes Transaction Controller Sample Results) :
    • Response times Percentiles Over Time (successful responses only) :
    • Active Threads Over Time :
    • Bytes throughput Over Time (Ignores Transaction Controller Sample Results) :
    • Latencies Over Time (Includes Transaction Controller Sample Results) :
    • Connect Time Over Time (Includes Transaction Controller Sample Results) :
    • Hits per second (Ignores Transaction Controller Sample Results):
    • Response codes per second (Ignores Transaction Controller Sample Results):
    • Transactions per second (Includes Transaction Controller Sample Results):
    • Response Time vs Request per second (Ignores Transaction Controller Sample Results):
    • Latency vs Request per second (Ignores Transaction Controller Sample Results):
    • Response time Overview (Excludes Transaction Controller Sample Results) :
    • Response times percentiles (Includes Transaction Controller Sample Results):
    • Times vs Threads (Includes Transaction Controller Sample Results):
      In distributed mode, this graph shows a horizontal axis the number of threads for 1 server. It's a current limitation
    • Response Time Distribution (Includes Transaction Controller Sample Results):

Dashboard generation uses JMeter properties to customize the report. Some properties are used for general settings and others are used for a particular graph configuration or exporter configuration. All report generator properties can be found in file reportgenerator.properties. To customize these properties, you should copy them in user.properties file and modify them.

Ensure you set property jmeter.reportgenerator.exporter.html.series_filter to keep only the transactions you want in the report if you don't want everything.
In the example below you must only modify Search|Order, keep the rest: jmeter.reportgenerator.exporter.html.series_filter=^(Search|Order)(-success|-failure)?$

To enable the generator to operate, the CSV file generated by JMeter must include certain required data which are correct by default in the last live version of JMeter.
If you modified those settings, check that your JMeter configuration follows these settings (these are the defaults): jmeter.save.saveservice.bytes = true # Only available with HttpClient4 #jmeter.save.saveservice.sent_bytes=true jmeter.save.saveservice.label = true jmeter.save.saveservice.latency = true jmeter.save.saveservice.response_code = true jmeter.save.saveservice.response_message = true jmeter.save.saveservice.successful = true jmeter.save.saveservice.thread_counts = true jmeter.save.saveservice.thread_name = true jmeter.save.saveservice.time = true jmeter.save.saveservice.connect_time = true jmeter.save.saveservice.assertion_results_failure_message = true # the timestamp format must include the time and should include the date. # For example the default, which is milliseconds since the epoch: jmeter.save.saveservice.timestamp_format = ms # Or the following would also be suitable # jmeter.save.saveservice.timestamp_format = yyyy/MM/dd HH:mm:ss

If you use Transaction Controllers, to ensure most accurate results:

  • uncheck the box (this is the default configuration): Generate parent sample
  • If Transaction Controller is used as a Container to represent a request for an HTML Page that will trigger Ajax calls and you only want in your report the Transaction Controller, then Right click on the node and Apply Naming Policy
    You will obtain this:

All properties must be prefixed with jmeter.reportgenerator. Title used in the generated report.
Default: "Apache JMeter Dashboard" Default date format from SimpleDateFormat Java API with Locale.ENGLISH.
Default Date format is yyyyMMddHHmmss
Useful when you would like to generate a report after the load test, and the results file contains timestamp in another time zone. In this case the date format must include the time zone (zzz). If jmeter.save.saveservice.timestamp_format does not contain year then use 1970 as year
Example: dd/MM/yyyy HH:mm:ss zzz Start date of the range of data to use for report.
Date format is defined by date_format property.
Default: not filled which means data range will be used from the beginning End date of the range of data to use for report.
Date format is defined by date_format property.
Default: not filled which means data range will be used until the end Granularity of over time graphs. Data is aggregated to have 1 minute ticks.
Granularity must be higher than 1 second (1000) otherwise throughput graphs will be incorrect Default: "60000" (1 minute) Sets the satisfaction threshold for the APDEX calculation (in ms).
Default: 500 Sets the tolerance threshold for the APDEX calculation (in ms).
Default: 1500 Sets satisfaction and tolerance threshold to specific samples.
Use sample names or regular expression.
Format is sample_name:satisfaction|tolerance[;]
Values are in milliseconds. Notice the colon between sample name and values, the pipe between thresholds and the semicolon at the end to separate different samples. Don't forget to escape after semicolon to span multiple lines. Example: jmeter.reportgenerator.apdex_per_transaction=sample(\\d+):1000|2000;\ samples12:3000|4000;\ scenar01-12:5000|6000 Sets the filter of samples to keep for generating graphs and statistics. An empty value deactivates the filtering.
Format: Regular expression.
Default: "" Sets the temporary directory used by the generation process if it needs file I/O operations.
Default: temp Sets the size of the sliding window used by percentile evaluation. Caution: higher value provides a better accuracy but needs more memory.
Default: 20000 Percentiles used by Summary table and Percentile graphs can be adjusted to different values by using the 3 properties:
  • aggregate_rpt_pct1 : Defaults to 90
  • aggregate_rpt_pct2 : Defaults to 95
  • aggregate_rpt_pct3 : Defaults to 99
 Relative paths are built from the JMeter working directory (default: bin).

You can define some overall properties which are used by the generator configuration. These properties are freely named but you should use the prefix jmeter.reportgenerator. in order to avoid property overlap.

E.g.:

Property definition:
jmeter.reportgenerator.overall_granularity=60000
Property reference:
${jmeter.reportgenerator.overall_granularity}

Each property describing a graph configuration must be prefixed with jmeter.reportgenerator.graph. followed by the graph identifier.

All graphs support these properties:

The fully qualified class name of the graph
The class of the graph must extend org.apache.jmeter.report.processor.graph.AbstractGraphConsumer.
See Default graph section for more details. Defines whether the graph discards controller samples. Default: false Sets the title of the graph. Default: ""

Specific graph properties must use the prefix: jmeter.reportgenerator.graph.<graph_id>.property The name of the property will be mapped using camel case transformation and the matching method of the class will be called with the property value as argument.

E.g.: jmeter.reportgenerator.graph.<graph_id>.property.set_granularity=150 induces the call of the method setGranularity(150) on the instance of the graph.

Each property describing an exporter configuration must be prefixed with jmeter.reportgenerator.exporter followed by the exporter identifier.

All exporters support these properties:

The fully qualified class name of the exporter
The class of the exporter must implement org.apache.jmeter.report.dashboard.DataExporter . Defines whether series_filter (see below) apply only on sample series. Default: true Sets the filter of series. An empty value deactivates the filtering. If not empty, regex should end with (-success|-failure)?$
Format: regular expression. Default: "" Defines whether only controller series are shown. Default: false

Specific exporter properties must use the prefix jmeter.reportgenerator.exporter.<exporter_id>.property

Sets the destination directory for generated html pages. Default: report-output Sets the source directory of template files from which the html pages are generated. Default: report-template

Graph properties allow exporters to overwrite some graph data.
They must use the prefix: jmeter.reportgenerator.exporter.<exporter_id>.graph_options.<graph_id>

Sets the minimum abscissa for the graph. Sets the maximum abscissa for the graph. Sets the minimum ordinate for the graph. Sets the maximum ordinate for the graph.

Unlike the filtering in the section General properties which discards data before calculations, here the filtering is performed after the calculations and serves to simplify the final report.

The property series_filter allows to filter which series of a graph (resp. rows of a summary table) using regular expression that matches the name of the series (resp. of the row). However, even if the name of the series (resp. row) matches the filter, the setting of the other filtering properties can lead to its discarding. Conversely if there is no matching, the other properties can allow to keep it.

The following tables show how the setting of filtering properties works.

Cases of discarding when there is pattern matching
filter_only_sample_series Graph/Summary supports controllers discrimination The current series is a controller series show_controllers_only Discarded
False False - False False
True
- False
True
True False False
True
True False
True
True False - False
True
- False
True
True False False
True True
True False False
True

Cases of retention when there is no pattern matching
filter_only_sample_series Graph/Summary supports controllers discrimination Kept
False False False
True
True False True
True False

Incorrect filter configuration can lead to generate empty graphs/summary tables:
  • If you set the property show_controllers_only and the graph is configured to exclude controllers.
  • If the property series_filter matches none series.

You can copy the following configuration to your user.properties file in order to test the report generator.

# Configure this property to change the report title #jmeter.reportgenerator.report_title=Apache JMeter Dashboard # Change this parameter if you want to change the granularity of over time graphs. # Granularity must be higher than 1000 (1second) otherwise Throughput graphs will be incorrect # see Bug 60149 #jmeter.reportgenerator.overall_granularity=60000 Change this parameter if you want to change the granularity of Response time distribution # Set to 100 ms by default #jmeter.reportgenerator.graph.responseTimeDistribution.property.set_granularity=100 # Change this parameter if you want to override the APDEX satisfaction threshold. jmeter.reportgenerator.apdex_satisfied_threshold=1500 # Change this parameter if you want to override the APDEX tolerance threshold. jmeter.reportgenerator.apdex_tolerated_threshold=3000 # Sets the destination directory for generated html pages, it is better to change it for every generation # This will override the value set through -o command line option # jmeter.reportgenerator.exporter.html.property.output_dir=/tmp/test-report # Indicates which graph series are filtered (regular expression) # In the below example we filter on Search and Order samples # Note that the end of the pattern should always include (-success|-failure)?$ # Transactions per second suffixes Transactions with "-success" or "-failure" depending # on the result #jmeter.reportgenerator.exporter.html.series_filter=^(Search|Order)(-success|-failure)?$ # Indicates whether series filter apply only on sample series jmeter.reportgenerator.exporter.html.filters_only_sample_series=true Adapt the parameter output_dir to your environment.

This configuration allows to generate a report where:

  • Over time graphs have a time granularity equal to 1 minute.
  • The satisfaction threshold for ADPEX calculation is 1 second and half.
  • The tolerance threshold for ADPEX calculation is 3 seconds.
  • The HTML files are generated in the directory /tmp/test-report.
  • Only series which the name begins with "s0" or "s1" are shown.
  • The previous filter only applies to graphs (resp. summary tables) where the series (resp. rows) match samples

The report generation can be done as a stand alone process from a sample log file or automatically after running load test.

Use the following command: jmeter -g <log file> -o <Path to output folder>

Use the following command: jmeter -n -t <test JMX file> -l <test log file> -e -o <Path to output folder>

You can generate the HTML report using menu item Tools → Generate HTML report:

Figure §-num;.1. HTML Report Dialog Menu

For each parameters see the following table :

The CSV output of a tes run The user.properties file used to run the load test The directory where you want the report to be created(must be empty)

If no output directory is defined, the controller will use ${JMETER_HOME}/bin/report-output.

You then only have to click on the Generate report button and wait for an information dialog to appear If report generation takes more than two minutes, adjust the property generate_report_ui.generation_timeout

 Due to limitations of this early version, each default graph must be declared in JMeter properties. Otherwise, the graph views will be empty.

All graphs provided by this report engine are located in the package org.apache.jmeter.report.processor.graph.impl

The dashboard generator provides the following graph classes:

Graph Description Supports controller discrimination
ActiveThreadsGraphConsumer This graph represents the number of active threads over time. False
BytesThroughputGraphConsumer This graph represents the throughput of received and sent data over time. False
CodesPerSecondGraphConsumer This graph represents the rate of response codes over time. False
HitsPerSecondGraphConsumer This graph represents the rate of finished requests over time. False
LatencyOverTimeGraphConsumer This graph represents the average latency time over time. True
ConnectTimeOverTimeGraphConsumer This graph represents the connection time over time. True
LatencyVSRequestGraphConsumer This graph represents the median and average latency time depending on the number of current requests. False
ResponseTimeDistributionGraphConsumer This graph represents the distribution of the samples depending on their elapsed time and name. True
ResponseTimeOverTimeGraphConsumer This graph represents the average response time over time. True
ResponseTimePercentilesGraphConsumer This graph represents the percentiles of the elapsed time over time. True
ResponseTimePercentilesOverTimeGraphConsumer This graph shows Min/Max and 3 percentiles response time over time. True
ResponseTimeVSRequestGraphConsumer This graph represents the median and average response time depending on the number of current requests. False
TimeVSThreadGraphConsumer This graph represents the average response time depending on the number of current active threads.

The *-aggregated series represent the average response time regardless of the number of current active threads. These series are represented by a sole point because the number of current active threads is aggregated to an average. So for these points:

  • The abscissa is the average of the number of current active threads when samples of the series finish.
  • The ordinate is the average of the response time for the samples of the series regardless of the number of current active threads.

 True
TransactionsPerSecondGraphConsumer This graph represents the rate of transaction by sample name over time. True

You can graph any sample_variable in CSV over time, you can customize your graphs by settings their properties in the user.properties file.
They must use the id prefix custom_: jmeter.reportgenerator.graph.custom_<your_graph_name_id>.property.<your_option_name> To specify that this graph is a customized one : jmeter.reportgenerator.graph.custom_<your_graph_name_id>.classname=org.apache.jmeter.report.processor.graph.impl.CustomGraphConsumer

Sets the X axis name of the graph. Sets the Y axis name of the graph. Sets the displayed message when the cursor is on a point of the graph. Name of the column you want to graph in the csv.

Here is an example of a custom graph configuration that graphs the variable ts-hit:

jmeter.reportgenerator.graph.custom_testGraph.classname=org.apache.jmeter.report.processor.graph.impl.CustomGraphConsumer jmeter.reportgenerator.graph.custom_testGraph.title=Chunk Hit jmeter.reportgenerator.graph.custom_testGraph.property.set_Y_Axis=Number of Hits jmeter.reportgenerator.graph.custom_testGraph.set_X_Axis=Over Time jmeter.reportgenerator.graph.custom_testGraph.property.set_granularity=60000 jmeter.reportgenerator.graph.custom_testGraph.property.set_Sample_Variable_Name=ts-hit jmeter.reportgenerator.graph.custom_testGraph.property.set_Content_Message=Number of Hits : If you want to contribute new graphs or improve current ones, you can read this developer documentation.
Read this documentation on contributing.