Taurus Configuration Linter

Due to the non-strict nature of YAML, syntactic and semantic errors in the configuration files are not a rare thing. Linting subsystem is an attempt to automatically warn user about errors, misprints and straight-out errors in the configuration files.

Linting is done right after all configuration files are loaded and just before any modules are prepared.

Linting output

Here's the example of linter output:

WARNING: at path 'execution.0.ramp': unfamiliar name 'ramp'. Did you mean 'ramp-up'?
WARNING: at path 'execution.0.throuhhput': unfamiliar name 'throuhhput'. Did you mean 'throughput'?
WARNING: at path 'execution.0.scenario': scenario 'foobar' is used but isn't defined

For each issue found, linter prints one warning. Each warning contains a path to the problematic code (e.g. execution.0.scenario) and a warning message. The paths are the same as used in CLI overrides and can be read pretty easily. For example, path execution.0.ramp means the key ramp that is defined inside the first element of the execution list.

Running the linter

Linter is always enabled by default and is invoked right after configuration is loaded. By default, linter prints a warning for each issue found, but it doesn't prevent test execution.

If you need to lint the config without actually running the test, you can use the -lint command-lint flag.


$ bzt -lint my-config.yml

Using -lint results in the following exit codes. The exit code is equal 0 if there were no issues found in the configuration and 1 if issues were found.

If for some reason you want to disable the linting, you can pass a -nolint flag, which will disable linting completely.

Configuring Linter

Just like everything in Taurus, linting can be configured within a configuration file.

    disable: false  # set to true to disable linting
    lint-and-exit: true  # stop Taurus after linting, thus preventing test execution
    ignored-warnings:  # list of warning ids to ignore. empty by default
    - single-execution
    - possible-typo

Warnings Produced by Linter

Taurus implements the following checks:

  • possible-typo: when Taurus detects a possible typo in the name of some field.
  • single-execution: when execution is a dict instead of the list. This is a supported case by Taurus, but it's use is discouraged.
  • execution-non-list: when execution is not a list, which is an error.
  • no-scenario: when execution item doesn't define the scenario to use for the test.
  • scenario-non-dict: when scenario (defined either inside execution, or in scenarios) is not a dict.
  • undefined-scenario: when execution item specifies scenario that isn't defined by the configuration.
  • no-script-or-requests: when scenario doesn't define neither script nor requests.
  • script-and-requests: when scenario defines both script and requests, making requests useless.