An execution represent the result of a specific run of a Plan, containing all the information (output values, status, …) about the keywords and controls of this run.
Executing a Plan
You can run a plan, and create a new corresponding execution by click on the “play” button, available in the Plan list, Keywords list and the Plan editor:
This will display the following generic popup, specifying the execution parameters (in green) and the execution mode
The following execution mode are possible:
The Simulate execution mode (in yellow in the screenshot). When choosing this mode, only the Plan logic will be validated and the keywords will not be executed
The Execution mode (in red), running the Plan and the keywords
The Schedule mode (in blue), allowing to schedule repetitive executions of the plan (see below for more detail)
The execution parameters can be defined in the Admin tab, under the Settings:
Under the Executions tab, you can view the whole list of your test plans executions.
You can filter the executions table using any of the header fields available :
Description : usually the test plan name you executed
Start time : the beginning of the execution
End time : the end of the execution
User : the user who executed the test plan
Environment : the target environment
Status : the current execution status
Result : the execution end result
From the execution table view, clicking on an execution will redirect you to the execution details, which are composed of the below views.
This table display the test plan execution steps in details, by default in reverse chronological order (last executed step first).
Again, you can use the table headers to filter the execution steps table and as well change the chronological display order :
If you are using TestSet and TestCase object like the above example, you can see that the execution steps are grouped by test case : you can open a test case details by clicking on it :
In addition, you can re-execute only selected test cases instead of the whole test plan by ticking its checkbox and click on the execute button :
You can toggle the display of the step execution details as follow. For Keywords it will show you for instance the input and output from the keyword call as well as the details of the keyword measurements:
Finally, you can open directly an execution step in the Execution tree (see below) by clicking on its associated button :
The table display your test plan execution steps as a tree in chronological order. Any node can be expanded to display its content :
The performance tab of the execution view will provide you with the most frequently required metrics (keyword throughput, average response time, etc) out-of-the-box. These metrics provide both statistic over time as well as in a summary form, covering the entire execution. Furthermore, the different charts and tables can be refreshed in real time or browsed as in a stationary version of the view. Lastly, for more advanced queries, drill-downs based on custom dimensions or raw measurement browsing, an interactive session can be opened as well as persisted.
When you first start a new execution, you will be taken to a monitoring view covering the last 30 seconds of the test and refreshed every 5 seconds, as illustrated on the screenshot below.
You are presented, from top to bottom and left to right with the following information:
- a line chart representing the average response time of the measurements (avg), aggregated over 30 intervals covering the selected time period
- a stacked area chart representing the number of measurements per second (tps), aggregated over 30 intervals covering the selected time period
- a stacked area chart representing the number of errors (either FAILED or TECHNICAL_ERROR), aggregated over 30 intervals covering the selected time period
- a summary table of all metrics applicable to the measurements (including percentiles, median, counts, etc), aggregated over a single interval covering the entire time period
- a scattered chart of the raw values of the last 100 measurements found in the selected period
- a table listing the last 100 measurement values found in the selected period
You can change the refresh rate and selected timeframe by clicking the top-level refresh button, and switching to the desired refresh rate (5, 10, 30, 60, 300 seconds or “OFF” for no automatic refresh):
You can also change the timeframe toggle (covering either the last 5 seconds, 30 seconds, 5 minutes, 30 minutes or all-time data originating from the execution). Another important toggle is the Keywords vs Custom toggle, which allows you, in the event that you’ve used the (measurement api) to record custom measurements, to switch between keywords and custom measurements.
When an execution completes, is aborted or when the refresh rate is set to “OFF”, an additional tool for time-based selection and visualization is added to the top of the performance tab: the timeline widget.
The rest of the view remains very similar to the real-time version, the only other notable change being that the two “last 100 measurement” dashlets are not displayed, because we believe they are much less relevant at the end of an execution.
Custom time range selection
The timeline widget displayed at the top of the view represents an aggregation over time of the the different keyword status found during the execution (usually PASSED, FAILED and TECHNICAL_ERROR). Implicitely the stack of all threw values is also an indication of the throughput of the execution in keywords per second.
By dragging and dropping alongside the timeline, the rest of the charts and tables will automatically update to the newly selected time frame:
You can then decide to slide your selection across in order to browse through the test:
Or reset the selection to the original “all-time” state by clicking the white area of the brush:
While we believe the real-time and summary views along with time-based selection will cover most analysis needs, users also have the option to further investigate and mine their data by starting an interactive session. From this view, you will also be able to export raw measurements to a file in case you’d like to use a 3rd party tool to process the data.
The “Interactive session” button in the top left corner of the view will take you to the following view which an instance of the “RTM dashboard” (originally for Response Time Manager).
The response time manager offers data selection, aggregation and visualization capabality, allowing users to filter and drill down into custom measurements in order to diagnose complex issues.
By default, two text filters will automatically be set in the default Measurement selector in order to only display data relating to the execution you had opened prior to clicking the interactive session button (eId=a_certain_execution_id) and focusing by keywords by default (type=keyword). However, various other attributes such as rnStatus (providing information on the keyword’s execution status) can be used to filter data.
For instance, by removing the eId filter and instead adding a date filter using the begin attribute (which points to the timestamp of each measurement), users can perform cross-execution queries and visualize data over multiple executions. In this example, we will query the raw measurement values of all executions over the last hour:
While the list of selectors depends on the custom measurements that you choose to create, below is a list of standard selectors and some of the possible values:
begin: start time of the measurement
eId: execution id of your test execution
planId: id of the executed plan
taskId: id of the related scheduled tasks if any
project: id of the related project if any
name: name of the measurements (either the keyword name or the custom name given while creating the measurement)
origin: name of the parent keyword measurement for custom measurements (same as name for keyword’s measurement)
type: type of the measurements (by default ‘keyword’ for automatic keywords measurements and custom for all others)
rnStatus: status of the transactions (PASSED, FAILED, TECHNICAL_ERROR…)
value: the value of the measurements (long number)
Different types of queries can be sent to different types of services in order to browse measurement data. Currently and by default, two main services are made available:
Raw values: this service allows users to browse individual measurements, which is great for investigating fine grained anomalies or just making sure that custom measurements are working as intended
Aggregates: this service performs overtime or summary-style aggregations based on dynamic criteria such as the name of a measurement, its status or any additional metadata attribute that was provided through the measurement API.
These services can be toggled by using the following button:
In either scenario, you can fire a new query (and subsequent aggregation) by clicking the Search button on either side of the dashboard:
The top part of the configuration dashlet contains service-specific configuration whereas the lower part contains configuration which is global to all services.
As mentionned above, by default the raw measurement values are displayed upon loading the dashboard. These values are however paged and presented in groups of 100 values. In order to browse the values, you can use the small arrows available on both sides of the Chart:
The table will update automatically.
Another important point to note is that the scattered value chart will always display the value of the measurement, while the table can also display metadata attribute values. These can be selected using the Table metrics dropdown selector:
In order to perform over-time or summary aggregation, two key inputs must be kept in mind and are exposed under the tab “Grouping” of the Service configuration pane:
interval size: tells RTM how large the aggregation intervals need to be and indirectly, how many datapoints will be returned and displayed. Defaults to “auto” which will automatically compute an aggregation with a target of 30 datapoints. However, arbitrary intervals can be provided in milliseconds, and the value “max” will compute an aggregation with a target of a single datapoint (also called summary mode).
split attribute: tells RTM how to split series (or group measurements for aggregation). By default, a distinct series is created for each keyword name, however, any attribute can be used for grouping & splitting (such as rnStatus for error analysis or an empty or non-existing attribute for merging all keywords into a single series). Mutliple attributes can even be combined with a “;” for drill-down oriented queries. For instance: entering name;rnStatus will lead the computation of a distinct series for each keyword name and status combination.
Additional advanced parameters are made available under the Perf tab. These allow for the partitioning and parallel computation of complex queries. The default values should only be increased if sufficient hardware is available on the backend (please contact our support for more specific information).
Much like in Raw values mode, various metrics can be selected for display in the table below. These, however are not related to metadata attributes, but are numeric results of computations based on the measurement’s values (for instance, mean/average, counts, sums, percentiles, etc).
Note: the actual computation of metrics is not yet configurable as of step 3.13 but the underlying framework allows for extending metrics and we plan on exposing metric computation to our users in the future. New metrics such as standard deviation or concurrent transaction count, and new services such as a service allowing the visualization of the entire response time distribution or all percentile,s are also on our roadmap.
Table sort: allows to organize datapoints based on time series or grouping series (keywords) first
Table export: performs a CSV export of the table content and downloads the file (see Exporting measurements for raw measurements export)
Global parameters > Fields: the name of the key fields used for aggregation and browsing, should not need to be modified as part of step’s use cases
In the RTM dashboard, the idea is to keep both a chart and a table visualization in sync at all times, regardless of whether the user is browsing raw values or aggregating data.
Lastly, a CSV export of the table can be ordered by clicking the export button of the service configuration pane.
In case you’d like to use your own toolset to process the raw data, you can use the export functionality. The exported data still depends on the provided selectors.
A new performance dashboard has been introduced with step 3.17 in a beta version. It is based on an integration with grafana and is currently using a prometheus server as its data source.
In order to integrate the grafana dashboards in the step web UI, following configuration changes are required in your grafana server configuration file (defaults.ini).
# set to true if you want to allow browsers to render Grafana in a <frame>, <iframe>, <embed> or <object>. default is false.
allow_embedding = true
# The full public facing url
root_url = %(protocol)s://%(domain)s:%(http_port)s/grafana
# Serve Grafana from subpath specified in `root_url` setting. By default it is set to `false` for compatibility reasons.
serve_from_sub_path = true
In the step.properties configuration file, following section corresponds to the plugin configuration. While most properties default values should be used, you will most likely needneed to set the url of your grafana server and the authentication parameters (either user/password or api key).
# Grafana plugin
# Note: on recent grafana version, iframe support must be enabled in grafana ini file: allow_embedding = true
# Base url to grafana server (default: http://localhost:3000)
# An embedded proxy can be used for authentication and avoid CORS issue, but in this case a sub_path must be used.
# Modify grafana ini file to set serve_from_sub_path to true and add the below sub_path to the root_url, and uncomment:
# Except is grafana is running with anonymous support, one of the following authentication methods shall be used
# for the grafana api (retrieving dashboards URL) and for the embedded proxy
# API key
# Basic auth
### Dashboard configuration
The error view display an overview of the errors (grouped by error type) which occurred during the execution :
You can jump directly to the nodes producing the error by click on the error type :
step offers you the possibility to schedule executions of Test Plans based on JAVA CRON expressions. From the executions view ,click on the “Schedule” button on the top right :
Enter the CRON expression which will define when the plan will be executed then press the “OK” button:
You can then switch to the “Scheduler” tab in order to view, edit, control the scheduling entries you set up as well as jump to the performance interactive analysis view related to this specific task:
Here are some basics examples of CRON expressions. If you are not familiar with, we encourage you to take a look at the CRON official documentation.
0 0 12 * * ?
Fire at 12:00 PM (noon) every day
0 15 10 * * ?
Fire at 10:15 AM every day
0 * 14 * * ?
Fire every minute starting at 2:00 PM and ending at 2:59 PM, every day
0 0/5 14 * * ?
Fire every 5 minutes starting at 2:00 PM and ending at 2:55 PM, every day
If Automatic Housekeeping, you may want to flag executions to be retained by the housekeeping job. You can also save information relating to your execution in the Description box (for instance a reference run as part of a test release).
Advanced executions options
In case of load test executions with extreme high load (or depending on your system capacity), creating and persisting a report for each and every node execution will impact the maximum load your step instance can generate. In such cases, it is possible to disable the persistence of the nodes by setting one or more of the following predefined variables (for instance using a set control):
tec.execution.reportnodes.persistbefore=false: this will only persist the node state after it execution, thus you won’t see the nodes in RUNNING state but still have all information for your analysis and reporting. Note: this is the default behavior for nodes executed within a threadgroup.
tec.execution.reportnodes.persistonlynonpassed=true: by setting this options, only nodes which failed (or with technical errors) will be persisted, reducing drastically the pressure but keeping information to analyse error cases.
tec.execution.reportnodes.persistafter=false: finally this last option combined with the “persistbefore” one will disable the persistence of all report nodes (this is usually not required and not recommended).
The changes on the persistence of report nodes doesn’t impact the creation of the performance measurements, you will always be able to perform the performance analysis and reporting of your execution
Controlling the number of “Multiworkers”
In some contexts, mainly for functional testing, you may want to control the overall number of workers (or threads) to be used when starting a test execution disregarding the values actually set in your plans.
This can ben enabled by adding the following execution parameter in step. Go to the admin screens settings page and create an execution parameter with:
Once done, you will see an additional field when starting a test execution.
How it words: When stating a test by providing a value to this new field, you will set (override) the number of threads/workers that are used by the first nodes implementing a thread pool. This includes among others: nodes of type Testset, For, ThreadGroup. Each child nodes implementing a thread pool will see their number of threads reduced to 1.