Scenarios

• Defined using simple XML files
  • Allows hierarchical structure and schema validation
  • Has reasonably balanced horizontal aestetics - compared to JSON

• Definition files are located in /tools/test/scenarios

• Scenario can import other scenarios

  • import can be qualified with a namespace 

• Comprised of steps or groups with dependencies

• Dependencies can be inherent or injected later
• Defines process flow as an execution DAG

Steps and Groups

• Step is an individual node in the scenario process flow
  
  • associated with some form of executable via exec attribute 
• Group is a collection of other groups and steps
  
  • used for grouping related sets of steps
  • used for encapsulation and definition of shared dependencies
• Has a name unique within the scope of a scenario
  
  • namespace can be used to preserve uniqueness for scenario import
  • Specifies other steps or groups it inherently requires
  • Can be marked as optional using if or unless attributes

Properties and Environment

• Most attributes support property expansion
  
  • specified as ${property}
  • evaluated using values of environment variables or Java properties
• Pass-through properties for downstream steps
  
  • @key=value included in output one property per line
  • only supported for exec attribute
• Steps and groups can specify env attribute
  
  • value can be a path to a file to be sourced-in
  • value can be “~“ to indicate command exit status should be ignored
• Steps and groups can specify cwd attribute
  • value is the directory to be set before execution of the step command

Parallel Chains

• Dynamically clones process flow subgraph region
• Used for repeating steps for independent portions of test environment
  • e.g. install, uninstall, collect logs on all ONOS instances concurrently
• Has an iteration variable specified as ${var#} property
• Clones subgraphs for each of var1, var2, … varN
• Stops cloning when varX becomes undefined or empty
• Value N can be abstracted using ${#} special property
• The <parallel> tag itself does not support the name or requires attributes. However, you can wrap the parallel chain in a group tag to get the same effect.

Sequential Chains

• Dynamically clones process flow subgraph region
• Used for sequencing steps for portions of test environment
  • e.g. process a task for all ONOS instances sequentially
• Has an iteration variable specified as ${var#} property
• Clones subgraphs for each of var1, var2, … varN
• Stops cloning when varX becomes undefined or empty
• Value N can be abstracted using ${#} special property
• Value N-1 can be abstracted using ${#-1} property

Dependencies

• Inherent dependencies specified as requires attribute of each step and group element
  
  • specified as a comma-separated list of step or group names
• Injected dependencies specified as requires attribute of a dependency element
  
  • used for chaining steps or groups imported from other scenarios
• Hard dependencies
  
  • forces wait and the failure of the predecessor results in the step being skipped
• Soft dependencies - designated by “~” prefix
  
  • forces wait but the failure of the  predecessor has no impact
• The "^" requirement is used to indicate the previous step. Note that this only works between a step and a step. If one of the linked steps is within a different group, parallel, or sequential chain then you must specify the requirement by name.