274 lines
9.9 KiB
YAML
274 lines
9.9 KiB
YAML
group: rllib tests
|
|
depends_on:
|
|
- forge
|
|
- ray-core-build(*)
|
|
- ray-dashboard-build
|
|
steps:
|
|
# builds
|
|
- name: rllibbuild-multipy
|
|
label: "wanda: rllibbuild-py{{array.python}}"
|
|
wanda: ci/docker/rllib.build.wanda.yaml
|
|
tags: cibase
|
|
array:
|
|
python:
|
|
- "3.10"
|
|
- "3.12"
|
|
env:
|
|
PYTHON: "{{array.python}}"
|
|
BASE_TYPE: "ml"
|
|
BUILD_VARIANT: "build"
|
|
RAYCI_IS_GPU_BUILD: "false"
|
|
depends_on: oss-ci-base_ml-multipy($)
|
|
|
|
- name: rllibgpubuild
|
|
wanda: ci/docker/rllib.build.wanda.yaml
|
|
depends_on: oss-ci-base_gpu-multipy(python=3.12)
|
|
env:
|
|
PYTHON: "3.12"
|
|
BASE_TYPE: "gpu"
|
|
BUILD_VARIANT: "gpubuild"
|
|
RAYCI_IS_GPU_BUILD: "true"
|
|
tags: cibase
|
|
|
|
# tests
|
|
- label: ":brain: rllib: component testing"
|
|
tags: rllib_directly
|
|
parallelism: 4
|
|
instance_type: large
|
|
commands:
|
|
# All tests to with tags for components without gpu or multi_gpu tags
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}" --parallelism-per-worker 3
|
|
--only-tags env,evaluation,models,offline,policy,utils,algorithms,callbacks,core
|
|
--except-tags gpu,multi_gpu,manual
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
depends_on: rllibbuild-multipy(python=3.12)
|
|
|
|
- label: ":brain: rllib: examples"
|
|
tags: rllib
|
|
parallelism: 6
|
|
instance_type: large
|
|
commands:
|
|
# Tests all examples without gpu, multi_gpu or examples_use_all_cores tag
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}" --parallelism-per-worker 2
|
|
--only-tags examples
|
|
--except-tags gpu,multi_gpu,manual,examples_use_all_core
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
# Tests all examples without gpu or multi_gpu
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}"
|
|
--only-tags examples_use_all_core
|
|
--except-tags gpu,multi_gpu,manual
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
--skip-ray-installation # reuse the same docker image as the previous run
|
|
depends_on: rllibbuild-multipy(python=3.12)
|
|
|
|
- label: ":brain: rllib: learning tests"
|
|
tags: rllib
|
|
parallelism: 5
|
|
instance_type: large
|
|
commands:
|
|
# learning tests without a gpu
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}" --parallelism-per-worker 3
|
|
--only-tags learning_tests
|
|
--except-tags gpu,multi_gpu,learning_tests_use_all_core,manual
|
|
--test-arg --framework=torch
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
# learning tests without a gpu but use all cores
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}"
|
|
--only-tags learning_tests_use_all_core
|
|
--except-tags gpu,multi_gpu,manual
|
|
--test-arg --framework=torch
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
--skip-ray-installation # reuse the same docker image as the previous run
|
|
depends_on: rllibbuild-multipy(python=3.12)
|
|
|
|
- label: ":brain: rllib: gpu tests"
|
|
tags:
|
|
- rllib_gpu
|
|
- gpu
|
|
- skip-on-microcheck
|
|
parallelism: 5
|
|
instance_type: gpu
|
|
commands:
|
|
# All gpu tagged tests
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}"
|
|
--only-tags gpu
|
|
--except-tags multi_gpu,manual
|
|
--test-env=RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--test-env=RLLIB_NUM_GPUS=1
|
|
--build-name rllibgpubuild-py3.12
|
|
--python-version 3.12
|
|
depends_on: rllibgpubuild
|
|
|
|
- label: ":brain: rllib: multi-gpu tests"
|
|
tags:
|
|
- rllib_gpu
|
|
- gpu
|
|
- skip-on-microcheck
|
|
parallelism: 5
|
|
instance_type: gpu-large
|
|
commands:
|
|
# All multi-gpu tagged tests (including all learning, examples and component testings)
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers "$${BUILDKITE_PARALLEL_JOB_COUNT}" --worker-id "$${BUILDKITE_PARALLEL_JOB}"
|
|
--parallelism-per-worker 2
|
|
--gpus 4
|
|
--only-tags multi_gpu
|
|
--except-tags manual
|
|
--build-name rllibgpubuild-py3.12
|
|
--python-version 3.12
|
|
depends_on: rllibgpubuild
|
|
|
|
- label: ":brain: rllib: doc tests"
|
|
if: '!(build.pull_request.labels includes "docs-go")'
|
|
tags:
|
|
- rllib_directly
|
|
- doc
|
|
- skip-on-microcheck
|
|
instance_type: medium
|
|
commands:
|
|
# doc tests
|
|
- bazel run //ci/ray_ci:test_in_docker -- python/ray/... //doc/... rllib
|
|
--only-tags doctest
|
|
--except-tags gpu,manual
|
|
--parallelism-per-worker 2
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
# doc examples
|
|
- bazel run //ci/ray_ci:test_in_docker -- //doc/... rllib
|
|
--except-tags gpu,post_wheel_build,timeseries_libs,doctest
|
|
--parallelism-per-worker 2
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
--skip-ray-installation # reuse the same docker image as the previous run
|
|
# documentation test
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--only-tags documentation
|
|
--parallelism-per-worker 2
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
--skip-ray-installation # reuse the same docker image as the previous run
|
|
depends_on: rllibbuild-multipy(python=3.12)
|
|
|
|
- label: ":brain: rllib: python 3.10 component testing ({{matrix.worker_id}})"
|
|
if: build.pull_request.labels includes "continuous-build" || pipeline.id == "0189e759-8c96-4302-b6b5-b4274406bf89" || pipeline.id == "018f4f1e-1b73-4906-9802-92422e3badaa"
|
|
tags: rllib_directly
|
|
instance_type: large
|
|
commands:
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib
|
|
--workers 4 --worker-id {{matrix.worker_id}} --parallelism-per-worker 3
|
|
--only-tags env,evaluation,models,offline,policy,utils,algorithms,callbacks,core
|
|
--except-tags gpu,multi_gpu,manual
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.10
|
|
--python-version 3.10
|
|
depends_on: rllibbuild-multipy(python=3.10)
|
|
matrix:
|
|
setup:
|
|
worker_id: ["0", "1", "2", "3"]
|
|
|
|
- label: ":brain: rllib: flaky component & examples tests"
|
|
key: rllib_flaky_tests_02
|
|
tags:
|
|
- rllib
|
|
- rllib_flaky
|
|
- skip-on-premerge
|
|
instance_type: large
|
|
commands:
|
|
# flaky components
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib --run-flaky-tests --parallelism-per-worker 3
|
|
--only-tags env,evaluation,models,offline,policy,utils,algorithms,callbacks,core
|
|
--except-tags learning_tests,examples,documentation,gpu,multi_gpu,manual
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
|
|
# flaky examples
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib --run-flaky-tests --parallelism-per-worker 3
|
|
--only-tags examples
|
|
--except-tags multi_gpu,gpu,manual,examples_use_all_core
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
--skip-ray-installation # reuse the same docker image as the previous run
|
|
|
|
# flaky examples use all core
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib --run-flaky-tests --parallelism-per-worker 3
|
|
--only-tags examples_use_all_core
|
|
--except-tags gpu,multi_gpu,manual
|
|
--test-env RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
--skip-ray-installation # reuse the same docker image as the previous run
|
|
depends_on: rllibbuild-multipy(python=3.12)
|
|
soft_fail: true
|
|
|
|
- label: ":brain: rllib: flaky learning tests"
|
|
key: rllib_flaky_tests_01
|
|
tags:
|
|
- rllib
|
|
- rllib_flaky
|
|
- skip-on-premerge
|
|
instance_type: large
|
|
commands:
|
|
# torch
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib --run-flaky-tests
|
|
--only-tags learning_tests
|
|
--except-tags gpu,multi_gpu,manual
|
|
--test-arg --framework=torch
|
|
--build-name rllibbuild-py3.12
|
|
--python-version 3.12
|
|
depends_on: rllibbuild-multipy(python=3.12)
|
|
soft_fail: true
|
|
|
|
- label: ":brain: rllib: flaky gpu tests"
|
|
key: rllib_flaky_gpu_tests
|
|
tags:
|
|
- rllib_gpu
|
|
- gpu
|
|
- rllib_flaky
|
|
- skip-on-premerge
|
|
instance_type: gpu
|
|
commands:
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib --run-flaky-tests
|
|
--only-tags gpu
|
|
--except-tags multi_gpu,manual
|
|
--test-env=RAY_USE_MULTIPROCESSING_CPU_COUNT=1
|
|
--test-env=RLLIB_NUM_GPUS=1
|
|
--build-name rllibgpubuild-py3.12
|
|
--python-version 3.12
|
|
depends_on: rllibgpubuild
|
|
soft_fail: true
|
|
|
|
- label: ":brain: rllib: flaky multi-gpu tests"
|
|
key: rllib_flaky_multi_gpu_tests
|
|
tags:
|
|
- rllib_gpu
|
|
- gpu
|
|
- rllib_flaky
|
|
- skip-on-premerge
|
|
instance_type: gpu-large
|
|
commands:
|
|
- bazel run //ci/ray_ci:test_in_docker -- //rllib/... rllib --run-flaky-tests
|
|
--parallelism-per-worker 2
|
|
--gpus 4
|
|
--only-tags multi_gpu
|
|
--except-tags manual
|
|
--build-name rllibgpubuild-py3.12
|
|
--python-version 3.12
|
|
depends_on: rllibgpubuild
|
|
soft_fail: true
|