Management of Pipelines
Pipelines in LDIO can be created in YAML or JSON configuration (although all example configurations are made in YAML, these can also be formatted in JSON).
A default pipeline looks as follows:
name: my-first-pipeline
input:
name: name of LDI Input
config:
foo: bar
adapter:
name: name of LDI Adapter
config:
foo: bar
transformers:
- name: name of LDI Transformer
config:
foo: bar
outputs:
- name: name of LDI Transformer
config:
foo: bar
- Note that one orchestrator can have multiple pipelines
- Note that one pipeline can have multiple LDI Transformers and LDI Outputs
Anatomy of a pipeline
Each pipeline is built up of the following components:
- LDIO Input: A component that will receive data (not necessarily LD) to then feed the LDIO pipeline.
- LDIO Adapter: To be used in conjunction with the LDIO Input, the LDIO Adapter will transform the provided content into and internal Linked Data model and sends it down the pipeline.
- LDIO Transformer: A component that takes in a Linked Data model, transforms/modifies it and then puts it back on the pipeline.
- LDIO Output: A component that will take in Linked Data and will export it to external sources.
stateDiagram-v2
direction LR
LDI_Input --> LDI_Transformer : LD
LDI_Transformer --> LDI_Output : LD
state LDI_Input {
direction LR
[*] --> LDI_Adapter : Non LD
state LDI_Adapter {
direction LR
[*] --> adapt
adapt --> [*]
}
LDI_Adapter --> [*] : LD
}
state LDI_Transformer {
direction LR
[*] --> transform
transform --> [*]
}
state LDI_Output {
direction LR
[*] --> [*]
}
Persistence of Pipelines
By default, all pipelines defined after startup (via management API) will be lost on restart.
To prevent this behaviour, add the orchestrator.directory
property as follows:
orchestrator:
directory: "{directory in application folder}"
If this directory does not exist, it will be created.
NOTE: An application config can be defined by creating an application YAML file in the LDIO directory (in docker, this correlates to
/ldio/application.yml
).
Pausing & Resuming LDIO
Sometimes it might be preferred to pause an LDIO pipeline instead of deleting and recreating it. The endpoints to manage pipelines can be found here
The exact behaviour of a paused pipeline depends on its input component and can be found in the documentation of these components. However, it will always complete its current run through the pipeline and then seize sending any output.