pipelines/backend
Eterna2 d482698e41
[api-server] Object store folder path is configurable and can work with AWS (secure and region flag, and IAM credentials) (#2080)
* Make pipelineFolder a configurable param

* Can config the secure flag and region config for api-server object store

* Updated minio client to use a chained credential provider so that IAM can be used.

* fix bug in passing secure flag to minio

* accesskey and secretkey for minio client is now optional.

* Expose GetPipelineKey instead of giving objectStore a default folder path because there are eother functions that need to access the store directly.

* fix fake bad object store.

* region shld be provided when attempting to create a bucket.

* Rename pipelineFolder to pipelinePath.
2020-02-11 10:52:08 -08:00
..
api Swagger generated code for UploadPipelineVersion method (#3010) 2020-02-07 20:23:53 -08:00
metadata_writer Service - Metadata writer (#2674) 2020-01-14 23:17:32 -08:00
src [api-server] Object store folder path is configurable and can work with AWS (secure and region flag, and IAM credentials) (#2080) 2020-02-11 10:52:08 -08:00
test Upload local file as new pipeline version Step 1 (#3001) 2020-02-07 16:13:36 +08:00
Dockerfile [Sample] Update notebook sample link (#3002) 2020-02-07 11:05:46 -08:00
Dockerfile.bazel Add arm64 support for ml-pipeline (#2507) 2019-11-07 17:34:19 -08:00
Dockerfile.persistenceagent Garbage collect the completed workflow after persisted to database (#1802) 2019-08-12 23:53:18 -07:00
Dockerfile.scheduledworkflow skip storing log to files (#1788) 2019-08-10 11:20:26 -07:00
Dockerfile.viewercontroller skip storing log to files (#1788) 2019-08-10 11:20:26 -07:00
Dockerfile.visualization Visualizations - Updated the TFMA and TFDV versions (#2946) 2020-01-31 18:59:20 -08:00
OWNERS Update backend OWNERS (#2435) 2019-10-21 11:54:07 -07:00
README.md Use Remote Build Execution for Bazel builds. (#1031) 2019-09-30 10:41:38 -07:00
build_api_server.sh [Backend] Improve parameter patching (#3016) 2020-02-10 00:15:53 -08:00

README.md

This directory contains code for the components that comprise the Kubeflow Pipelines backend.

Building & Testing

All components can be built using Bazel. To build everything under backend, run: bazel build --action_env=PATH --define=grpc_no_ares=true //backend/...

To run all tests: bazel test --action_env=PATH --define=grpc_no_ares=true //backend/...

The API server itself can only be built/tested using Bazel. The following commands target building and testing just the API server. bazel build --action_env=PATH --define=grpc_no_ares=true backend/src/apiserver/... bazel test --action_env=PATH --define=grpc_no_ares=true backend/src/apiserver/...

Building APIServer Image using Remote Build Execution

If you are a dev in the Kubeflow Pipelines team, you can use Remote Build Execution Service to build the API Server image using Bazel with use of a shared cache for speeding up the build. To do so, execute the following command:

./build_api_server.sh -i gcr.io/cloud-ml-pipelines-test/api-server:dev

Building Go client library and swagger files

After making changes to proto files, the Go client libraries and swagger files need to be regenerated and checked-in. The backend/api/generate_api.sh script takes care of this.

Updating BUILD files

As the backend is written in Go, the BUILD files can be updated automatically using Gazelle. Whenever a Go file is added or updated, run the following to ensure the BUILD files are updated as well: bazel run //:gazelle

If a new external Go dependency is added, or an existing one has its version bumped in the go.mod file, ensure the BUILD files pick this up by updating the WORKSPACE go_repository rules using the following command: bazel run //:gazelle -- update-repos --from_file=go.mod