Overview

A process is an execution of flows written in Concord DSL running in a project or standalone. A process can represent a single deployment, CI/CD job or any other, typically a “one-off”, type of workload.

Let’s take a look at an example:

configuration:
  arguments:
    todoId: "1"

flows:
  default:
    - task: http
      in:
        url: "https://jsonplaceholder.typicode.com/todos/${todoId}"
        response: json
        out: myTodo

    - if: "${myTodo.content.completed}"
      then:
        - log: "All done!"
      else:
        - log: "You got a todo item: ${myTodo.content.title}"

When executed this flow performs a number of steps:

• fetches a JSON object from the specified URL;
• saves the response as a flow variable;
• checks if the retrieved “todo” is completed or not;
• prints out a message depending whether the condition is true or not.

The example demonstrates a few concepts:

• flow definitions use Concord’s YAML-based DSL;
• flows can call tasks. And tasks can perform useful actions;
• flows can use conditional expressions;
• tasks can save their results as flow variables
• an expression language can be used to work with data inside flows;

There are multiple ways how to execute a Concord process: using a Git repository, sending the necessary files in the API request, using a trigger, etc.

No matter how the process was started it goes through the same execution steps:

• project repository data is cloned or updated;
• binary payload from the process invocation is added to the workspace;
• configuration parameters from different sources are merged together;
• imports and templates are downloaded and applied;
• the process is added to the queue;
• one of the agents picks up the process from the queue;
• the agent downloads the process state, dependencies and imports;
• the agent starts the runtime in the process’ working directory;
• the flow configured as entry point is invoked.

During its life, a process can go though various statuses:

• NEW - the process start request is received, passed the initial validation and saved for execution;
• PREPARING - the start request is being processed. During this status, the Server prepares the initial process state;
• ENQUEUED - the process is ready to be picked up by one of the Agents;
• STARTING - the process was dispatched to an Agent and is being prepared to start on the Agent’s side;
• RUNNING - the process is running;
• SUSPENDED - the process is waiting for an external event (e.g. a form);
• RESUMING - the Server received the event the process was waiting for and now prepares the process’ resume state;
• FINISHED - final status, the process was completed successfully. Or, at least, all process-level errors were handled in the process itself;
• FAILED - the process failed with an unhandled error;
• CANCELLED - the process was cancelled by a user;
• TIMED_OUT - the process exceeded its execution time limit.

Runtime

The runtime is what actually executes the process. It is an interpreter written in Java that executes flows written in Concord DSL. Typically this is executed in a separate JVM process.

Currently there are two versions of the runtime:

• concord-v1 - the current stable version, used by default;
• concord-v2 - new and improved version introduced in 1.42.0. Not recommended for production usage yet.

The runtime can be specified using configuration.runtime parameter in the concord.yml file:

configuration:
  runtime: "concord-v2"

or in the request parameters:

$ curl -F runtime=concord-v2 ... https://concord.example.com/api/v1/process

Process Events

During process execution, Concord records various events: process status changes, task calls, internal plugin events, etc. The data is stored in the database and used later in the Concord Console and other components.

Events can be retrieved using the API. Currently, those event types are:

• PROCESS_STATUS - process status changes;
• ELEMENT - flow element events (such as task calls).

In addition, plugins can use their own specific event types. For example, the Ansible plugin uses custom events to record playbook execution details. This data is extensively used by the Concord Console to provide visibility into the playbook execution - hosts, playbook steps, etc.

Event recording can be configured in the Runner section of the process’ configuration object.