Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

fix(v2,sdk.v2): Remove need for base image root access to write artifacts #6530

Closed
wants to merge 8 commits into from
Closed

fix(v2,sdk.v2): Remove need for base image root access to write artifacts #6530

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
ed8f635
Add `emptyDir` volume mounted at `/gcs`
judahrand Sep 10, 2021
1378dec
Move `launch` binary to `/var/run/kfp/bin`
judahrand Sep 10, 2021
caa31ab
Move S3 and Minio artifacts to `/var/run/kfp/artifact`
judahrand Sep 10, 2021
b03cbee
Move output parameters to `/var/run/kfp/parameter`
judahrand Sep 12, 2021
203f8bf
Move `outputMetadata` to `/var/run/kfp`
judahrand Sep 12, 2021
13e1af7
Merge branch 'master' into feat/non-root-artifact
judahrand Oct 24, 2021
165f839
Ensure directory for binary exists
judahrand Oct 28, 2021
729c7ce
Merge branch 'master' into feat/non-root-artifact
judahrand Oct 29, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 10 additions & 6 deletions sdk/python/kfp/compiler/testdata/v2_compatible_two_step_pipeline.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,7 @@ spec:
- '{{$}}'
- --function_to_execute
- preprocess
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, preprocess, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -111,7 +111,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: python:3.7
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -134,11 +135,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.8.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
- name: train
container:
Expand Down Expand Up @@ -181,7 +183,7 @@ spec:
- '{{$}}'
- --function_to_execute
- train
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, train, --pipeline_name, '{{inputs.parameters.pipeline-name}}',
--run_id, $(KFP_RUN_ID), --run_resource, workflows.argoproj.io/$(WORKFLOW_ID),
Expand Down Expand Up @@ -219,7 +221,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: python:3.7
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -241,11 +244,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.8.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
arguments:
parameters:
Expand Down
16 changes: 10 additions & 6 deletions sdk/python/kfp/compiler/testdata/v2_compatible_two_step_pipeline_with_custom_launcher.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,7 @@ spec:
- '{{$}}'
- --function_to_execute
- preprocess
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, preprocess, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -111,7 +111,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: python:3.7
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -134,11 +135,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: my-custom-image
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
- name: train
container:
Expand Down Expand Up @@ -181,7 +183,7 @@ spec:
- '{{$}}'
- --function_to_execute
- train
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, train, --pipeline_name, '{{inputs.parameters.pipeline-name}}',
--run_id, $(KFP_RUN_ID), --run_resource, workflows.argoproj.io/$(WORKFLOW_ID),
Expand Down Expand Up @@ -219,7 +221,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: python:3.7
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -241,11 +244,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: my-custom-image
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
arguments:
parameters:
Expand Down
12 changes: 8 additions & 4 deletions sdk/python/kfp/compiler/v2_compat.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,14 +44,18 @@ def update_op(op: dsl.ContainerOp,
launcher_container = dsl.UserContainer(
name="kfp-launcher",
image=image_name,
command=["launcher", "--copy", "/kfp-launcher/launch"],
command=["launcher", "--copy", "/var/run/kfp/bin/launch"],
mirror_volume_mounts=True)

op.add_init_container(launcher_container)
op.add_volume(k8s_client.V1Volume(name='kfp-launcher'))
op.add_volume_mount(
k8s_client.V1VolumeMount(
name='kfp-launcher', mount_path='/kfp-launcher'))
name='kfp-launcher', mount_path='/var/run/kfp'))
op.add_volume(k8s_client.V1Volume(name='gcs'))
op.add_volume_mount(
k8s_client.V1VolumeMount(
name='gcs', mount_path='/gcs'))

# op.command + op.args will have the following sections:
# 1. args passed to kfp-launcher
Expand All @@ -62,7 +66,7 @@ def update_op(op: dsl.ContainerOp,
#
# example:
# - command:
# - /kfp-launcher/launch
# - /var/run/kfp/bin/launch
# - '--mlmd_server_address'
# - $(METADATA_GRPC_SERVICE_HOST)
# - '--mlmd_server_port'
Expand All @@ -88,7 +92,7 @@ def update_op(op: dsl.ContainerOp,
# import xxx
# ...
op.command = [
"/kfp-launcher/launch",
"/var/run/kfp/bin/launch",
"--mlmd_server_address",
"$(METADATA_GRPC_SERVICE_HOST)",
"--mlmd_server_port",
Expand Down
4 changes: 2 additions & 2 deletions sdk/python/kfp/v2/components/types/artifact_types.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,8 +20,8 @@
from typing import Dict, Generic, List, Optional, Type, TypeVar, Union

_GCS_LOCAL_MOUNT_PREFIX = '/gcs/'
_MINIO_LOCAL_MOUNT_PREFIX = '/minio/'
_S3_LOCAL_MOUNT_PREFIX = '/s3/'
_MINIO_LOCAL_MOUNT_PREFIX = '/var/run/kfp/artifact/minio/'
_S3_LOCAL_MOUNT_PREFIX = '/var/run/kfp/artifact/s3/'


class Artifact(object):
Expand Down
40 changes: 25 additions & 15 deletions sdk/python/tests/compiler/testdata/uri_artifacts.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@ spec:
args: [sh, -c, 'mkdir -p "$(dirname $0)" && python -c "import random; result
= ''heads'' if random.randint(0,1) == 0 else ''tails''; print(result, end='''')"
| tee $0', '{{$.outputs.parameters[''output''].output_file}}']
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, flip-coin, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -72,7 +72,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: python:alpine3.6
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -93,11 +94,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.6.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
- name: for-loop-2
inputs:
Expand Down Expand Up @@ -125,7 +127,7 @@ spec:
set -e -x
gsutil cat "$0"
- '{{$.inputs.artifacts[''input_gcs_path''].uri}}'
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, read-from-gcs, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -160,7 +162,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: google/cloud-sdk:slim
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -177,11 +180,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.6.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
- name: read-from-gcs-2
container:
Expand All @@ -192,7 +196,7 @@ spec:
set -e -x
gsutil cat "$0"
- '{{$.inputs.artifacts[''input_gcs_path''].uri}}'
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, read-from-gcs-2, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -227,7 +231,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: google/cloud-sdk:slim
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -244,11 +249,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.6.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
- name: read-from-gcs-3
container:
Expand All @@ -259,7 +265,7 @@ spec:
set -e -x
gsutil cat "$0"
- '{{$.inputs.artifacts[''input_gcs_path''].uri}}'
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, read-from-gcs-3, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -294,7 +300,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: google/cloud-sdk:slim
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -311,11 +318,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.6.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
- name: uri-artifact-pipeline
inputs:
Expand Down Expand Up @@ -377,7 +385,7 @@ spec:
echo "$0" | gsutil cp - "$1"
- '{{$.inputs.parameters[''text'']}}'
- '{{$.outputs.artifacts[''output_gcs_path''].uri}}'
command: [/kfp-launcher/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
command: [/var/run/kfp/bin/launch, --mlmd_server_address, $(METADATA_GRPC_SERVICE_HOST),
--mlmd_server_port, $(METADATA_GRPC_SERVICE_PORT), --runtime_info_json, $(KFP_V2_RUNTIME_INFO),
--container_image, $(KFP_V2_IMAGE), --task_name, write-to-gcs, --pipeline_name,
'{{inputs.parameters.pipeline-name}}', --run_id, $(KFP_RUN_ID), --run_resource,
Expand Down Expand Up @@ -413,7 +421,8 @@ spec:
- configMapRef: {name: metadata-grpc-configmap, optional: true}
image: google/cloud-sdk:slim
volumeMounts:
- {mountPath: /kfp-launcher, name: kfp-launcher}
- {mountPath: /var/run/kfp, name: kfp-launcher}
- {mountPath: /gcs, name: gcs}
inputs:
parameters:
- {name: pipeline-name}
Expand All @@ -433,11 +442,12 @@ spec:
pipelines.kubeflow.org/v2_component: "true"
pipelines.kubeflow.org/enable_caching: "true"
initContainers:
- command: [launcher, --copy, /kfp-launcher/launch]
- command: [launcher, --copy, /var/run/kfp/bin/launch]
image: gcr.io/ml-pipeline/kfp-launcher:1.6.6
name: kfp-launcher
mirrorVolumeMounts: true
volumes:
- {name: gcs}
- {name: kfp-launcher}
arguments:
parameters:
Expand Down
Loading