Asynchronous execution

Async mode

In many situation awaiting response when interacting with workflow instances might not be desired behaviour. It might be due to various reasons such as

  • long running service invocation

  • poor network that can cause dropped connections

  • there is no need to wait for the response at all .e.g fire and forget kind of use cases

With Automatiko you can take advantage of async execution very easily by deciding on the mode on request basis. That means workflow definition is not really aware of it and the responsibility how the invocation should happen is on consumer side.

It means that different consumers can chose to invoke the service either synchronously or asynchronously. Or even the same client can invoke it one time with async mode and another with sync mode.

The asynchronous mode is given as HTTP header X-ATK-Mode and currently supports just single value (case insensitive) async.

curl -i -X 'POST' \
  'http://localhost:8080/registrations' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -H 'X-ATK-Mode: async' \
  -d '{
    "lastName": "doe",
    "name": "john"

The important part is ` -H 'X-ATK-Mode: async'` which instructs the service to execute the logic in background but return directly a response with 202 Accepted response code and payload that includes identifier of the instance being processed so it can be easily fetched later on.

HTTP/1.1 202 Accepted
Content-Length: 98
Content-Type: application/json


This applies to several operations of the service but the main rule here is whatever operation leads to workflow progression it can be executed in asynchronous way. This includes

  • start of new instance

  • update of variables (as it can trigger logic that listens to data changes e.g. conditional events)

  • completing of user tasks

  • canceling of user tasks

  • signals sent to workflow instance


Sometimes having identifier of the workflow instance might not be sufficient. For example when given operation leads to an end of the workflow instance and by that instance is being removed from the system. In such situation it won’t be possible to be polled to get the results. To over come such limitations, when executing in async mode clients can provide a callback URL that the service should invoke upon completion with complete payload of the instance as it would return when invoked synchronously.

To do that, additional header is expected to be given X-ATK-Callback and the value should be a valid HTTP URL.

curl -i -X 'POST' \
  'http://localhost:8080/registrations' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -H 'X-ATK-Mode: async' \
  -H 'X-ATK-Callback: https://my.custom.server.endpoint/registrations' \
  -d '{
    "lastName": "doe",
    "name": "john"

Upon completion Automatiko will POST a complete data model to the given URL. It will also set request header X-ATK-Status that will indicate what is the status of the workflow instance with one of the following values

  • Active - when instance is still active (meaning did not complete or was aborted)

  • Completed - when instance has reached an end state

  • Aborted - when instance was aborted due to business logic

  • Failed - when instance failed at execution and is now in error state

Keep in mind that security related exception that are also performed in background might cause that a callback is not invoked. For instance when user task is about to be completed but a user who has no access to given task then callback won’t be invoked as there is actually nothing to post back since instance was not modified at all. In general this should be an edge case as such user should not be able to even see the task or know its details.

Authentication on callbacks

There might be a need to authenticate when calling back after async execution. For that Automatiko allows to configure a single authentication info for callbacks that can be defined

  • via file

  • system properties

  • environment variables

Several authentication modes are provided

  • none

  • basic

  • OAuth2

  • Custom HTTP header

  • on behalf - taking the header from the incoming call

  • query parameter on callback url

Below is a table with all configuration options for authentication

Table 1. Authentication configuration parameters
Property name Environment variable Description BuildTime only



Defines what type of authentication is to be used on callback when using async execution (expected values basic, oauth, custom, on-behalf)




Specifies basic authorization string, expected user name and password encrypted with Base64 but without `Basic ` prefix




Specifies user name to be used for basic authentication




Specifies password to be used for basic authentication




Specifies complete access token to be used as bearer token on the callback call




Specifies client id to be used to obtain OAuth token




Specifies client secret to be used to obtain OAuth token




Specifies refresh token to be used to automatically refresh access token




Specifies refresh token endpoint




Specifies scopes to be set when obtaining token




Specifies name of HTTP header to be set on the callback call




Specifies custom value to be set on the callback call




Specifies name of the header to be taken from request headers that acts like the "on behalf" information