Our quickstarts (for example, Java and Go) are great for getting a LittleHorse Workflow running quickly. However, due to their brevity they do not show everything you need to build a useful application on top of LittleHorse.
This repository will grow into a collection of reference applications that show you how to use LittleHorse in various different situations.
As a prerequisite to all of these examples, you should have access to a LH Server and lhctl. To install the proper version of lhctl, you can run:
go install github.com/littlehorse-enterprises/littlehorse/[email protected]
To run a LittleHorse Cluster (Kafka, LH Server, and soon to be LH Dashboard) in a single container on your laptop, you can run:
docker run --name littlehorse -d -p 2023:2023 public.ecr.aws/littlehorse/lh-standalone:0.7.0-alpha.1
The lh-standalone image takes about 30 seconds to start up because we first need to start a Kafka cluster. Once LH is up and running, you should get a [] response from:
lhctl search wfSpec
NOTE: if your computer falls asleep for a while, you might benefit from stopping, removing, and re-starting the lh-standalone container.
An application generally has the following components:
- Program that generates and registers a
WfSpec - A few Task Workers
- Backend-for-Frontend that translates LittleHorse Primitives into Business Domain Objects.
- A LittleHorse Cluster. In this case, the easiest way to get that is to use the
lh-standaloneimage.
In LittleHorse, one TaskDef can be used in different WfSpecs; which means that one Task Worker can be used in multiple applications.
The following is the repo structure:
The common-tasks folder contains a set of Task Worker implementations that can be independently used by various different applications.
Each task, regardless of what language it is written in, has the following two bash scripts:
.../register.sh, which registers theTaskDefto the LittleHorse Server..../run-worker.sh, which starts the Task Worker so that it can execute tasks.
Note that if a WfSpec for an application depends on a TaskDef, you must run the .../register.sh for the relevant TaskDefs before attempting to deploy the WfSpec; else, the rpc PutWfSpec will fail.
The applications folder contains a set of various demo applications that use LittleHorse. Each application will have some or all of the following components:
- A program that registers a
WfSpec(every application has this) - A backend-for-frontend or REST API
- A frontend which drives the workflow
Please consult the README of each individual application for instructions about how to run that app.