Pipelines API User Guide¶
Note: Ploomber Cloud is in beta
Community Plan¶
The free Community Plan has some limitations:
Limited to 50 pipeline tasks daily
No custom resources (all tasks executed with 2 vCPUs and 4 GiB of RAM)
If you wish to remove the daily limit and request custom resources, you may switch to the Teams Plan.
Setup¶
Note: We’re constantly improving Ploomber Cloud; ensure you’re running the latest version for the best experience:
pip install ploomber --upgrade
Then set your key:
ploomber cloud set-key {your-key}
Dependencies¶
To add dependencies required by your pipeline, create a
requirements.lock.txt
file. For example:
pandas
scikit-learn
The first time you submit a pipeline to Ploomber Cloud it’ll take a few minutes since it has to build a Docker image from scratch, subsequent Docker builds will be a lot faster since we’ll cache your image.
Custom task resources¶
To request custom resources for a task, add a cloud.yaml
file, with
the following format:
task_resources:
{task-name}:
vcpus: {number}
memory: {number} # MiB
gpu: {number}
For example, if we want the task fit
to request 8 vcpus
and 32GB of RAM:
task_resources:
fit:
vcpus: 8
memory: 32768 # 32 * 1024 = 32768
Note: Custom resources are not available in the community plan.