Runs
Run is the main unit of execution and monitoring in the Trickest platform.
Run represents specific workflow execution where nodes are executed in a sequence to produce and forward results.
Creating runs
Run, as a workflow execution, can be created in several ways:
- Trigger single manual execution of a workflow in platform UI
- Define scheduler to execute a workflow at fixed intervals in the platform UI
- Use Trickest CLI and execute a workflow right from your terminal
Each run has a mode property that indicates its creation method: MANUAL, SCHEDULE and CLI.
Run statuses
Workflow Execution
Each Run, as a workflow execution, goes through several statuses, starting from the point of its creation until execution is finished:
| Status | Description |
|---|---|
| Pending | The Platform is preparing machines assigned to the workflow execution. This is the time it takes to get machines up and ready to execute the workflow. |
| Running | The Platform has started executing a workflow. |
| Completed | Workflow execution is completed (all workflow nodes are executed). |
| Stopping | The Platform is stopping the workflow execution after user has explicitly requested the abortion of the execution. This is the time it takes to get machines released. |
| Stopped | The workflow execution is stopped at the explicit request of the user, before all workflow nodes are executed. |
Node Execution
Each node, as a part of a workflow, is executed individually within a Run. Each node execution goes through several statuses:
| Status | Description |
|---|---|
| Initializing | The machine, assigned to the node, is preparing requirements for node's execution (pulling Docker images, input files). |
| Running | The machine, assigned to the node, began with performing node's execution. Performing assigned computational task is in progress. |
| Succeeded | Node's execution is completed successfully. Result is ready to be viewed and downloaded. |
| Failed | Something went wrong during node execution. Check error log for more information. |
Execution machines
Each node, as a part of a workflow, executes individually within a Run and requires a machine for its execution.
Trickest provides several types of machines available to be used in Run. Before creating a Run, appropriate type of machine should be defined for each node. One virtual machine of each used type will be dedicated to a Run, and each node will be executed on a virtual machine of its defined type.
Learn to assign appropriate machine type to a node.
Execution parallelism
Trickest enables you to instantly create multiple branches within a workflow and handle/execute each concurrently before seamlessly merging back into a single flow.
Splitter
Trickest provides a special util node, splitter, that enables you to accomplish the execution of specific workflow branches in batches, in parallel.
More precisely, splitter enables you to loop over a file and execute a series of steps in a workflow (specific set of connected nodes) for each line in parallel.
Learn to execute in batches using splitter.
Control your parallelism
For each workflow, containing a minimum of 2 nodes arranged in parallel or splitter node, on its run creation, Trickest enables you to optionally increase the number of machines dedicated to a workflow execution and have full control over the number of concurrent executions.
Learn to assign number of machines to a run and achieve parallelism.
Results
Each node, as a part of a workflow, executes individually within a Run, producing its own output. Each node output, as a part of workflow results, is available to be viewed and downloaded and used in further analysis.