Releasing dependencies
This describes the steps to release changes over multiple repos with dependencies. We assume there are changes in the service package, used by the engine package, in turn a dependency of multiple repos.
Changes on the main branch of a repo are automatically deployed to staging. Tags are deployed to production. See CI/CD for more information.
info
The automatic building and releasing of Kubeflow Pipelines needs further refinement, and still work in progress. See Kubeflow CI/CD.
Service Package
Repo: deepdesk-service
- Test/commit/build fix
- Update
setup.pyversion<v-service> - New GitHub release
<v-service>
Engine Package
Repo: deepdesk-engine
- Update service dependency to
<v-service> - Update
setup.pyengine version to<v-engine-lib> - New GitHub release
<v-engine-lib>
Engine API
Repo: deepdesk-engine-api
- Update engine dependency to
<v-engine-lib> - New GitHub release
<v-engine-api>
Train Component
Repo: deepdesk-train
- Update engine dependency to
<v-engine-lib> - New GitHub release
<v-train> - deepdesk-kubeflow-text-pipeline: Automatically updated and released
- deepdesk-kubeflow-url-pipeline: Automatically updated and released
Importer Component
Repo: deepdesk-kubeflow-db-importer
- Update engine dependency to
<v-engine-lib> - Update
daily_exports_to_bigquery.yamlversion to<v-importer> - Update
load_from_bigquery.yamlversion to<v-importer> - Update
file_dump_to_bigquery_v2.yamlversion to<v-importer> - New GitHub release
<v-importer> - Automatically updated and released
- Automatically updated and released