mirror of
https://github.com/vale981/ray
synced 2025-03-06 02:21:39 -05:00
b91246a093
We currently measure end-to-end training time in our benchmarks, which includes setup overhead. This is an unequal comparison, as setup overhead for vanilla training cannot be accurately expressed and was instead just disregarded. By comparing the raw training times in the actual training loop, we will get a more accurate expression of any potential overhead or benefit in using Ray vs. vanilla tensorflow/torch. Signed-off-by: Kai Fricke <kai@anyscale.com>
4652 lines
96 KiB
YAML
4652 lines
96 KiB
YAML
# Global release test configuration file.
|
|
# All your release test configuration should go here. Adding release tests here
|
|
# will automatically enable them in the Buildkite release testing schedules
|
|
# (except they have frequency: disabled).
|
|
# Here is an example configuration for reference:
|
|
#- name: example_test
|
|
# # Tests with the same group will be grouped in the Buildkite UI
|
|
# group: Example group
|
|
# # Provide the working directory which will be uploaded to the cluster
|
|
# working_dir: example_dir
|
|
#
|
|
# # For release test infra migration, we provide these fields that are populated
|
|
# # in the database
|
|
# legacy:
|
|
# test_name: example_test
|
|
# test_suite: examples
|
|
#
|
|
# # How often to run the tests.
|
|
# # One of [disabled, any, multi, nightly, weekly].
|
|
# frequency: weekly
|
|
# # Owning team. This field will be persisted to the database
|
|
# team: ml
|
|
#
|
|
# # Python version. This optional field determines which Python version to run tests
|
|
# # on. This must be a string!
|
|
# python: "3.7"
|
|
#
|
|
# # Optional location of a bash setup script to run on the driver
|
|
# # when setting up the local environment. Relative to working_dir
|
|
# driver_setup: setup_driver.sh
|
|
#
|
|
# # Cluster information
|
|
# cluster:
|
|
# # Location of cluster env, relative to working_dir
|
|
# cluster_env: cluster_env.yaml
|
|
# # Location of cluster compute, relative to working_dir
|
|
# cluster_compute: cluster_compute.yaml
|
|
# # Autosuspend parameter passed to the cluster.
|
|
# # The cluster will automatically terminate if inactive for this
|
|
# # many minutes. Defaults to 10 if not set.
|
|
# autosuspend_mins: 10
|
|
# # Optional cloud_id to use instead of the default cloud
|
|
# cloud_id: cld_12345678
|
|
# # Alternatively, you can specify a cloud name
|
|
# cloud_name: anyscale_default_cloud
|
|
#
|
|
# # Run configuration for the test
|
|
# run:
|
|
# # Type of test. Can be sdk_command or client (job to be implemented soon).
|
|
# # Uses either Anyscale SDK commands or the Ray client to run the actual
|
|
# # release test.
|
|
# type: sdk_command
|
|
#
|
|
# # File manager to use to transfer files to and from the cluster.
|
|
# # Can be any of [sdk, client, job].
|
|
# file_manager: sdk
|
|
#
|
|
# # If you want to wait for nodes to be ready, you can specify this here:
|
|
# wait_for_nodes:
|
|
# # Number of nodes
|
|
# num_nodes: 16
|
|
# # Timeout for waiting for nodes. If nodes are not up by then, the
|
|
# # test will fail.
|
|
# timeout: 600
|
|
#
|
|
# # Optional prepare script to be run on the cluster before the test script
|
|
# prepare: python prepare.py
|
|
# # The prepare command can have a separate timeout
|
|
# prepare_timeout: 300
|
|
#
|
|
# # Main script to run as the test script
|
|
# script: python workloads/train_small.py
|
|
# # Timeout in seconds. After this time the test is considered as failed.
|
|
# timeout: 600
|
|
#
|
|
# # You can specify smoke test definitions here. If a smoke test is triggered,
|
|
# # it will deep update the main test configuration with the values provided
|
|
# # here. Smoke tests will automatically run with IS_SMOKE_TEST=1 as en
|
|
# # environment variable and receive the --smoke-test flag as a parameter in the
|
|
# # run script.
|
|
# smoke_test:
|
|
# # Smoke tests can have different frequencies. A smoke test is only triggered
|
|
# # when the regular test is not matched.
|
|
# frequency: nightly
|
|
# # Here we adjust the run timeout down and run on less nodes. The test script
|
|
# # remains the same.
|
|
# run:
|
|
# timeout: 300
|
|
# wait_for_nodes:
|
|
# num_nodes: 4
|
|
# timeout: 600
|
|
#
|
|
# # After the test finished, this handler (in alerts/) will process the results.
|
|
# # It can then let the test fail, e.g. if a metric regression is observed.
|
|
# alert: default
|
|
|
|
#######################
|
|
# Cluster scaling tests
|
|
#######################
|
|
- name: cluster_tune_scale_up_down
|
|
group: Cluster tests
|
|
working_dir: cluster_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: cpt_autoscaling_1-3.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/tune_scale_up_down.py
|
|
wait_for_nodes:
|
|
num_nodes: 0
|
|
|
|
type: sdk_command
|
|
|
|
alert: default
|
|
|
|
|
|
#########################
|
|
# AIR release tests
|
|
#########################
|
|
- name: long_running_horovod_tune_test
|
|
group: AIR tests
|
|
working_dir: air_tests
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: horovod/app_config_master.yaml
|
|
cluster_compute: horovod/compute_tpl.yaml
|
|
|
|
run:
|
|
timeout: 36000
|
|
script: python horovod/workloads/horovod_tune_test.py
|
|
long_running: true
|
|
wait_for_nodes:
|
|
num_nodes: 3
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 1800
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_data_bulk_ingest
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: data_20_nodes.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/data_benchmark.py --dataset-size-gb=200 --num-workers=20
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
# AIR benchmarks for XGBoost CUJ
|
|
- name: air_benchmark_xgboost_cpu_10
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: xgboost_app_config.yaml
|
|
cluster_compute: xgboost_compute_tpl.yaml
|
|
|
|
run:
|
|
timeout: 36000
|
|
script: python workloads/xgboost_benchmark.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 10
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 1800
|
|
|
|
alert: default
|
|
|
|
# Ray AIR distributed Torch benchmarks
|
|
- name: air_benchmark_torch_mnist_cpu_4x1
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_4.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/torch_benchmark.py run --num-runs 3 --num-epochs 20 --num-workers 4 --cpus-per-worker 8
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: air_benchmark_torch_mnist_gpu_4x4
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_4x4.yaml
|
|
|
|
run:
|
|
timeout: 4800
|
|
script: python workloads/torch_benchmark.py run --num-runs 3 --num-epochs 120 --num-workers 16 --cpus-per-worker 4 --batch-size 1024 --use-gpu
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: nightly
|
|
|
|
cluster:
|
|
cluster_compute: compute_gpu_2x2.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/torch_benchmark.py run --num-runs 3 --num-epochs 60 --num-workers 4 --cpus-per-worker 4 --batch-size 512 --use-gpu
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_torch_mnist_cpu_1x4
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/torch_benchmark.py run --num-runs 3 --num-epochs 20 --num-workers 4 --cpus-per-worker 2
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_torch_batch_prediction_gpu_1x1_20gb
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_1.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/gpu_batch_prediction.py --data-size-gb 20
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
# TODO(jiaodong): This leads to CUDA OOM on release tool but succeeds on
|
|
# workspaces, re-enabled it once inline conversions are fixed
|
|
# - name: air_benchmark_torch_batch_prediction_gpu_4x4_100gb
|
|
# group: AIR tests
|
|
# working_dir: air_tests/air_benchmarks
|
|
|
|
# frequency: nightly
|
|
# team: ml
|
|
# env: staging
|
|
|
|
# stable: false
|
|
|
|
# cluster:
|
|
# cluster_env: app_config.yaml
|
|
# cluster_compute: compute_gpu_16.yaml
|
|
|
|
# run:
|
|
# timeout: 10800
|
|
# script: python workloads/gpu_batch_prediction.py --data-size-gb 100
|
|
|
|
# type: sdk_command
|
|
# file_manager: job
|
|
|
|
# alert: default
|
|
|
|
|
|
- name: air_benchmark_torch_mnist_cpu_4x4
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_4.yaml
|
|
|
|
run:
|
|
timeout: 5400
|
|
script: python workloads/torch_benchmark.py run --num-runs 3 --num-epochs 20 --num-workers 16 --cpus-per-worker 2
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_tune_torch_mnist
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_8.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/tune_torch_benchmark.py --num-runs 3 --num-trials 8 --num-workers 4
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 8
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: air_benchmark_tune_torch_mnist_gpu
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_4_g4_12xl.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/tune_torch_benchmark.py --num-runs 2 --num-trials 4 --num-workers 4 --use-gpu
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: air_benchmark_tune_torch_mnist_large_gpu
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_8_g4_12xl.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/tune_torch_benchmark.py --num-runs 2 --num-trials 4 --num-workers 8 --use-gpu
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 8
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
# Ray AIR distributed Tensorflow benchmarks
|
|
- name: air_benchmark_tensorflow_mnist_cpu_4x1
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_4.yaml
|
|
|
|
run:
|
|
timeout: 5400
|
|
script: python workloads/tensorflow_benchmark.py run --num-runs 3 --num-epochs 20 --num-workers 4 --cpus-per-worker 8
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_tensorflow_mnist_cpu_1x4
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 5400
|
|
script: python workloads/tensorflow_benchmark.py run --num-runs 3 --num-epochs 20 --num-workers 4 --cpus-per-worker 2
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_tensorflow_mnist_cpu_4x4
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
stable: false
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_cpu_4.yaml
|
|
|
|
run:
|
|
timeout: 5400
|
|
script: python workloads/tensorflow_benchmark.py run --num-runs 3 --num-epochs 20 --num-workers 16 --cpus-per-worker 2
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_tensorflow_mnist_gpu_4x4
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_4x4.yaml
|
|
|
|
run:
|
|
timeout: 5400
|
|
script: python workloads/tensorflow_benchmark.py run --num-runs 3 --num-epochs 200 --num-workers 16 --cpus-per-worker 4 --batch-size 64 --use-gpu
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: nightly
|
|
|
|
cluster:
|
|
cluster_compute: compute_gpu_2x2.yaml
|
|
|
|
run:
|
|
script: python workloads/tensorflow_benchmark.py run --num-runs 3 --num-epochs 60 --num-workers 4 --cpus-per-worker 4 --batch-size 512 --use-gpu
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_pytorch_training_e2e_gpu_1x1_20gb
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_1.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/pytorch_training_e2e.py --data-size-gb 20
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: air_benchmark_pytorch_training_e2e_gpu_4x4_100gb
|
|
group: AIR tests
|
|
working_dir: air_tests/air_benchmarks
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
stable: false
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_gpu_16.yaml
|
|
|
|
run:
|
|
timeout: 10800
|
|
script: python workloads/pytorch_training_e2e.py --data-size-gb=100 --num-workers=16
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
#######################
|
|
# XGBoost release tests
|
|
#######################
|
|
- name: xgboost_train_small
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: train_small
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/train_small.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: client
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_train_moderate
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: train_moderate
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_moderate.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/train_moderate.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_train_gpu
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: train_gpu
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config_gpu.yaml
|
|
cluster_compute: tpl_gpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/train_gpu.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 5
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_distributed_api_test
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: distributed_api_test
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/distributed_api_test.py
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_ft_small_elastic
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: ft_small_elastic
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/ft_small_elastic.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_ft_small_non_elastic
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: ft_small_non_elastic
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/ft_small_non_elastic.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_tune_small
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: tune_small
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/tune_small.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_tune_32x4
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: tune_32x4
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_moderate.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/tune_32x4.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
- name: xgboost_tune_4x32
|
|
group: XGBoost
|
|
working_dir: xgboost_tests
|
|
|
|
legacy:
|
|
test_name: tune_4x32
|
|
test_suite: xgboost_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_moderate.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/tune_4x32.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: xgboost_tests
|
|
|
|
#######################
|
|
# LightGBM tests
|
|
#######################
|
|
- name: lightgbm_train_small
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: train_small
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/train_small.py
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: lightgbm_train_moderate
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: train_moderate
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_moderate.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/train_moderate.py
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: lightgbm_distributed_api_test
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: distributed_api_test
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/distributed_api_test.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: lightgbm_ft_small_non_elastic
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: ft_small_non_elastic
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/ft_small_non_elastic.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: lightgbm_tune_small
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: tune_small
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/tune_small.py
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: lightgbm_tune_16x4
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: tune_16x4
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_moderate.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/tune_16x4.py
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: lightgbm_tune_4x16
|
|
group: LightGBM tests
|
|
working_dir: lightgbm_tests
|
|
|
|
legacy:
|
|
test_name: tune_4x16
|
|
test_suite: lightgbm_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_moderate.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/tune_4x16.py
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
#######################
|
|
# ML user tests
|
|
#######################
|
|
- name: ml_user_horovod_user_test_latest
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: horovod_user_test_latest
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: horovod/app_config.yaml
|
|
cluster_compute: horovod/compute_tpl.yaml
|
|
|
|
driver_setup: horovod/driver_setup_latest.sh
|
|
run:
|
|
timeout: 1200
|
|
script: python horovod/horovod_user_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_horovod_user_test_master
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: horovod_user_test_master
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: horovod/app_config_master.yaml
|
|
cluster_compute: horovod/compute_tpl.yaml
|
|
|
|
driver_setup: horovod/driver_setup_master.sh
|
|
run:
|
|
timeout: 1200
|
|
script: python horovod/horovod_user_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_train_tensorflow_mnist_test
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: train_tensorflow_mnist_test
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: train/app_config.yaml
|
|
cluster_compute: train/compute_tpl.yaml
|
|
|
|
driver_setup: train/driver_setup.sh
|
|
run:
|
|
timeout: 36000
|
|
script: python train/train_tensorflow_mnist_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_train_torch_linear_test
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: train_torch_linear_test
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: train/app_config.yaml
|
|
cluster_compute: train/compute_tpl.yaml
|
|
|
|
driver_setup: train/driver_setup.sh
|
|
run:
|
|
timeout: 36000
|
|
script: python train/train_torch_linear_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_xgboost_gpu_connect_latest
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: xgboost_gpu_connect_latest
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: xgboost/app_config_gpu.yaml
|
|
cluster_compute: xgboost/tpl_gpu_small_scaling.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python xgboost/train_gpu_connect.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_xgboost_gpu_connect_master
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: xgboost_gpu_connect_master
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: xgboost/app_config_gpu_master.yaml
|
|
cluster_compute: xgboost/tpl_gpu_small_scaling.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python xgboost/train_gpu_connect.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_ray_lightning_user_test_latest
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: ray_lightning_user_test_latest
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: ray-lightning/app_config.yaml
|
|
cluster_compute: ray-lightning/compute_tpl.yaml
|
|
|
|
driver_setup: ray-lightning/driver_setup.sh
|
|
run:
|
|
timeout: 1200
|
|
script: python ray-lightning/ray_lightning_user_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_ray_lightning_user_test_master
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: ray_lightning_user_test_master
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: ray-lightning/app_config_master.yaml
|
|
cluster_compute: ray-lightning/compute_tpl.yaml
|
|
|
|
driver_setup: ray-lightning/driver_setup.sh
|
|
run:
|
|
timeout: 1200
|
|
script: python ray-lightning/ray_lightning_user_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
- name: ml_user_tune_rllib_connect_test
|
|
group: ML user tests
|
|
working_dir: ml_user_tests
|
|
|
|
legacy:
|
|
test_name: tune_rllib_connect_test
|
|
test_suite: ml_user_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: ../rllib_tests/app_config.yaml
|
|
cluster_compute: tune_rllib/compute_tpl.yaml
|
|
|
|
driver_setup: tune_rllib/driver_setup.sh
|
|
run:
|
|
timeout: 2000
|
|
script: python tune_rllib/run_connect_tests.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
#######################
|
|
# Tune cloud tests
|
|
#######################
|
|
- name: tune_cloud_aws_no_sync_down
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: true
|
|
|
|
legacy:
|
|
test_name: aws_no_sync_down
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_aws_4x2.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py no_sync_down
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_aws_ssh_sync
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: true
|
|
|
|
legacy:
|
|
test_name: aws_ssh_sync
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_aws_4x2.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py ssh_sync
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_aws_durable_upload
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: true
|
|
|
|
legacy:
|
|
test_name: aws_durable_upload
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_aws_4x2.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py durable_upload --bucket s3://tune-cloud-tests/durable_upload
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_aws_durable_upload_rllib_str
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: false
|
|
|
|
legacy:
|
|
test_name: aws_durable_upload_rllib_str
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config_ml.yaml
|
|
cluster_compute: tpl_aws_4x2.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
|
|
script: python workloads/run_cloud_test.py durable_upload --trainable rllib_str
|
|
--bucket s3://tune-cloud-tests/durable_upload_rllib_str
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_aws_durable_upload_rllib_trainer
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: false
|
|
|
|
legacy:
|
|
test_name: aws_durable_upload_rllib_trainer
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config_ml.yaml
|
|
cluster_compute: tpl_aws_4x2.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py durable_upload --trainable rllib_trainer
|
|
--bucket s3://tune-cloud-tests/durable_upload_rllib_trainer
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_gcp_k8s_no_sync_down
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: true
|
|
|
|
legacy:
|
|
test_name: gcp_k8s_no_sync_down
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_gcp_k8s_4x8.yaml
|
|
cloud_id: cld_k8WcxPgjUtSE8RVmfZpTLuKM # anyscale_k8s_gcp_cloud
|
|
autosuspend_mins: 60
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py no_sync_down --cpus-per-trial 8
|
|
type: client
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_gcp_k8s_ssh_sync
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: true
|
|
|
|
legacy:
|
|
test_name: gcp_k8s_ssh_sync
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_gcp_k8s_4x8.yaml
|
|
cloud_id: cld_k8WcxPgjUtSE8RVmfZpTLuKM # anyscale_k8s_gcp_cloud
|
|
autosuspend_mins: 60
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py ssh_sync --cpus-per-trial 8
|
|
type: client
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_cloud_gcp_k8s_durable_upload
|
|
group: Tune cloud tests
|
|
working_dir: tune_tests/cloud_tests
|
|
|
|
stable: true
|
|
|
|
legacy:
|
|
test_name: gcp_k8s_durable_upload
|
|
test_suite: tune_cloud_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_gcp_k8s_4x8.yaml
|
|
cloud_id: cld_k8WcxPgjUtSE8RVmfZpTLuKM # anyscale_k8s_gcp_cloud
|
|
autosuspend_mins: 60
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/run_cloud_test.py durable_upload --cpus-per-trial 8 --bucket gs://tune-cloud-tests/durable_upload
|
|
type: client
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
alert: tune_tests
|
|
|
|
|
|
########################
|
|
# Tune scalability tests
|
|
########################
|
|
|
|
- name: tune_scalability_bookkeeping_overhead
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
|
|
legacy:
|
|
test_name: bookkeeping_overhead
|
|
test_suite: tune_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_1x16.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python workloads/test_bookkeeping_overhead.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_scalability_durable_trainable
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
|
|
legacy:
|
|
test_name: durable_trainable
|
|
test_suite: tune_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_16x2.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
script: python workloads/test_durable_trainable.py --bucket tune-cloud-tests
|
|
wait_for_nodes:
|
|
num_nodes: 16
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_scalability_long_running_large_checkpoints
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
|
|
legacy:
|
|
test_name: long_running_large_checkpoints
|
|
test_suite: tune_tests
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_1x32_hd.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
script: python workloads/test_long_running_large_checkpoints.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: nightly
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_scalability_network_overhead
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
legacy:
|
|
test_name: network_overhead
|
|
test_suite: tune_tests
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_100x2.yaml
|
|
|
|
run:
|
|
timeout: 900
|
|
prepare_timeout: 1200
|
|
script: python workloads/test_network_overhead.py
|
|
wait_for_nodes:
|
|
num_nodes: 100
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: nightly
|
|
|
|
cluster:
|
|
cluster_compute: tpl_20x2.yaml
|
|
|
|
run:
|
|
timeout: 400
|
|
prepare_timeout: 600
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_scalability_result_throughput_cluster
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
|
|
legacy:
|
|
test_name: result_throughput_cluster
|
|
test_suite: tune_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_16x64.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/test_result_throughput_cluster.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 16
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_scalability_result_throughput_single_node
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
|
|
legacy:
|
|
test_name: result_throughput_single_node
|
|
test_suite: tune_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_1x96.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/test_result_throughput_single_node.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
- name: tune_scalability_xgboost_sweep
|
|
group: Tune scalability tests
|
|
working_dir: tune_tests/scalability_tests
|
|
|
|
legacy:
|
|
test_name: xgboost_sweep
|
|
test_suite: tune_tests
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config_data.yaml
|
|
cluster_compute: tpl_16x64.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python workloads/test_xgboost_sweep.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 16
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
alert: tune_tests
|
|
|
|
|
|
########################
|
|
# Golden Notebook tests
|
|
########################
|
|
- name: golden_notebook_torch_tune_serve_test
|
|
group: Golden Notebook tests
|
|
working_dir: golden_notebook_tests
|
|
|
|
legacy:
|
|
test_name: torch_tune_serve_test
|
|
test_suite: golden_notebook_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: torch_tune_serve_app_config.yaml
|
|
cluster_compute: gpu_tpl.yaml
|
|
|
|
run:
|
|
timeout: 1800
|
|
script: python workloads/torch_tune_serve_test.py
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
|
|
#######################
|
|
# Long running tests
|
|
#######################
|
|
|
|
- name: long_running_actor_deaths
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: actor_deaths
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/actor_deaths.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_apex
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: apex
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
cluster:
|
|
cluster_env: ../rllib_tests/app_config.yaml
|
|
cluster_compute: tpl_cpu_3.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
script: python workloads/apex.py
|
|
long_running: true
|
|
wait_for_nodes:
|
|
num_nodes: 3
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_impala
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: impala
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
cluster:
|
|
cluster_env: app_config_np.yaml
|
|
cluster_compute: tpl_cpu_1_large.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
script: python workloads/impala.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_many_actor_tasks
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: many_actor_tasks
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/many_actor_tasks.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_many_drivers
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: many_drivers
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/many_drivers.py --iteration-num=4000
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_many_ppo
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: many_ppo
|
|
test_suite: long_running_tests
|
|
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
cluster:
|
|
cluster_env: ../rllib_tests/app_config.yaml
|
|
cluster_compute: many_ppo.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
script: python workloads/many_ppo.py
|
|
long_running: true
|
|
wait_for_nodes:
|
|
num_nodes: 1
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_many_tasks
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: many_tasks
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/many_tasks.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_many_tasks_serialized_ids
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: many_tasks_serialized_ids
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/many_tasks_serialized_ids.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_node_failures
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: node_failures
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/node_failures.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_pbt
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: pbt
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
cluster:
|
|
cluster_env: ../rllib_tests/app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/pbt.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_serve
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
legacy:
|
|
test_name: serve
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/serve.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_serve_failure
|
|
group: Long running tests
|
|
working_dir: long_running_tests
|
|
|
|
stable: false
|
|
|
|
legacy:
|
|
test_name: serve_failure
|
|
test_suite: long_running_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_cpu_1.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
prepare: ray stop
|
|
script: python workloads/serve_failure.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
|
|
run:
|
|
timeout: 600
|
|
|
|
alert: long_running_tests
|
|
|
|
- name: long_running_distributed_pytorch_pbt_failure
|
|
group: Long running tests
|
|
working_dir: long_running_distributed_tests
|
|
|
|
legacy:
|
|
test_name: pytorch_pbt_failure
|
|
test_suite: long_running_distributed
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl.yaml
|
|
|
|
run:
|
|
timeout: 86400
|
|
script: python workloads/pytorch_pbt_failure.py
|
|
long_running: true
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: disabled
|
|
run:
|
|
timeout: 3600
|
|
|
|
alert: long_running_tests
|
|
|
|
########################
|
|
# Jobs tests
|
|
########################
|
|
|
|
- name: jobs_basic_local_working_dir
|
|
group: Jobs tests
|
|
working_dir: jobs_tests
|
|
|
|
legacy:
|
|
test_name: jobs_basic_local_working_dir
|
|
test_suite: jobs_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_4_xlarge.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/jobs_basic.py --working-dir "workloads"
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: jobs_basic_remote_working_dir
|
|
group: Jobs tests
|
|
working_dir: jobs_tests
|
|
|
|
legacy:
|
|
test_name: jobs_basic_remote_working_dir
|
|
test_suite: jobs_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_4_xlarge.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/jobs_basic.py --working-dir "https://github.com/anyscale/job-services-cuj-examples/archive/refs/heads/main.zip"
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
########################
|
|
# Runtime env tests
|
|
########################
|
|
- name: runtime_env_rte_many_tasks_actors
|
|
group: Runtime env tests
|
|
working_dir: runtime_env_tests
|
|
|
|
legacy:
|
|
test_name: rte_many_tasks_actors
|
|
test_suite: runtime_env_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: rte_small.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/rte_many_tasks_actors.py
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: runtime_env_wheel_urls
|
|
group: Runtime env tests
|
|
working_dir: runtime_env_tests
|
|
|
|
legacy:
|
|
test_name: wheel_urls
|
|
test_suite: runtime_env_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: rte_minimal.yaml
|
|
|
|
run:
|
|
timeout: 9000
|
|
script: python workloads/wheel_urls.py
|
|
wait_for_nodes:
|
|
num_nodes: 1
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: runtime_env_rte_ray_client
|
|
group: Runtime env tests
|
|
working_dir: runtime_env_tests
|
|
|
|
legacy:
|
|
test_name: rte_ray_client
|
|
test_suite: runtime_env_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: rte_minimal.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python workloads/rte_ray_client.py
|
|
wait_for_nodes:
|
|
num_nodes: 1
|
|
|
|
type: client
|
|
|
|
alert: default
|
|
|
|
|
|
########################
|
|
# Serve tests
|
|
########################
|
|
|
|
- name: serve_single_deployment_1k_noop_replica
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: single_deployment_1k_noop_replica
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_32_cpu.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
long_running: false
|
|
script: python workloads/single_deployment_1k_noop_replica.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: serve_multi_deployment_1k_noop_replica
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: multi_deployment_1k_noop_replica
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_32_cpu.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
long_running: false
|
|
script: python workloads/multi_deployment_1k_noop_replica.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: serve_autoscaling_single_deployment
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: autoscaling_single_deployment
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_8_cpu_autoscaling.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
long_running: false
|
|
script: python workloads/autoscaling_single_deployment.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: serve_autoscaling_multi_deployment
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: autoscaling_multi_deployment
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_32_cpu_autoscaling.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
long_running: false
|
|
script: python workloads/autoscaling_multi_deployment.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: serve_serve_micro_benchmark
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: serve_micro_benchmark
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_single_node.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
long_running: false
|
|
script: python workloads/serve_micro_benchmark.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: serve_serve_micro_benchmark_k8s
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: serve_micro_benchmark_k8s
|
|
test_suite: serve_tests
|
|
|
|
# TODO(architkulkarni) Reenable after K8s migration. Currently failing
|
|
frequency: disabled
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_single_node_k8s.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
long_running: false
|
|
script: python workloads/serve_micro_benchmark.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: deployment_graph_long_chain
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: deployment_graph_long_chain
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_single_node_32_cpu.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
long_running: false
|
|
script: python workloads/deployment_graph_long_chain.py --chain-length=10 --num-clients=4 --local-test=False
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
stable: False
|
|
|
|
- name: deployment_graph_wide_ensemble
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: deployment_graph_wide_ensemble
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_single_node_32_cpu.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
long_running: false
|
|
script: python workloads/deployment_graph_wide_ensemble.py --fanout-degree=10 --num-clients=4 --local-test=False
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
stable: False
|
|
|
|
- name: serve_handle_long_chain
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: serve_handle_long_chain
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_single_node_32_cpu.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
long_running: false
|
|
script: python workloads/serve_handle_long_chain.py --chain-length=10 --num-clients=4 --local-test=False
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
stable: False
|
|
|
|
- name: serve_handle_wide_ensemble
|
|
group: Serve tests
|
|
working_dir: serve_tests
|
|
|
|
legacy:
|
|
test_name: serve_handle_wide_ensemble
|
|
test_suite: serve_tests
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl_single_node_32_cpu.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
long_running: false
|
|
script: python workloads/serve_handle_wide_ensemble.py --fanout-degree=10 --num-clients=4 --local-test=False
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
stable: False
|
|
|
|
########################
|
|
# Train tests
|
|
########################
|
|
|
|
- name: train_horovod_multi_node_test
|
|
group: Train tests
|
|
working_dir: train_tests/horovod
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python train_horovod_multi_node_test.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
########################
|
|
# RLlib tests
|
|
########################
|
|
|
|
- name: rllib_learning_tests_a2c
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=a2c
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_a3c
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=a3c
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_apex
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 1gpu_24cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=apex
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_appo
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=appo
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_ppo
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=ppo
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_bc
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 1gpu_16cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=bc
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_cql
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 1gpu_16cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=cql
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_ddpg
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=ddpg
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_dqn
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=dqn
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_es
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 4gpus_64cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=es
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_impala
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=impala
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_marwil
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 1gpu_16cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=marwil
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: rllib_learning_tests_sac
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=sac
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_learning_tests_slateq
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=slateq
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: rllib_learning_tests_td3
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 2gpus_32cpus.yaml
|
|
|
|
run:
|
|
timeout: 18000
|
|
script: python learning_tests/run.py --yaml-sub-dir=td3
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
|
|
- name: rllib_multi_gpu_learning_tests
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: multi_gpu_learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 8gpus_96cpus.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python multi_gpu_learning_tests/run.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_multi_gpu_with_lstm_learning_tests
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: multi_gpu_with_lstm_learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 8gpus_96cpus.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python multi_gpu_with_lstm_learning_tests/run.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_multi_gpu_with_attention_learning_tests
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: multi_gpu_with_attention_learning_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: nightly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 8gpus_96cpus.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python multi_gpu_with_attention_learning_tests/run.py
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
alert: default
|
|
|
|
- name: rllib_stress_tests
|
|
group: RLlib tests
|
|
working_dir: rllib_tests
|
|
|
|
legacy:
|
|
test_name: stress_tests
|
|
test_suite: rllib_tests
|
|
|
|
frequency: weekly
|
|
team: ml
|
|
env: staging
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: 4gpus_544_cpus.yaml
|
|
|
|
run:
|
|
timeout: 5400
|
|
script: python stress_tests/run_stress_tests.py
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 6
|
|
|
|
type: sdk_command
|
|
file_manager: job
|
|
|
|
smoke_test:
|
|
frequency: nightly
|
|
|
|
run:
|
|
timeout: 2000
|
|
|
|
alert: default
|
|
|
|
########################
|
|
# Core Nightly Tests
|
|
########################
|
|
|
|
- name: shuffle_10gb
|
|
group: core-multi-test
|
|
|
|
team: core
|
|
frequency: multi
|
|
working_dir: nightly_tests
|
|
|
|
legacy:
|
|
test_name: shuffle_10gb
|
|
test_suite: nightly_tests
|
|
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_single.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=50 --partition-size=200e6
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: shuffle_50gb
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: shuffle_50gb
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_single.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=50 --partition-size=1e9
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: shuffle_50gb_large_partition
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: shuffle_50gb_large_partition
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_single.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=500 --partition-size=100e6
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: shuffle_100gb
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: shuffle_100gb
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_multi.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=200 --partition-size=500e6
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: non_streaming_shuffle_100gb
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: non_streaming_shuffle_100gb
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_multi.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=200 --partition-size=500e6
|
|
--no-streaming
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 4
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: non_streaming_shuffle_50gb_large_partition
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: non_streaming_shuffle_50gb_large_partition
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_single.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=500 --partition-size=100e6
|
|
--no-streaming
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: non_streaming_shuffle_50gb
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: non_streaming_shuffle_50gb
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_single.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=50 --partition-size=1e9
|
|
--no-streaming
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: stress_test_placement_group
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: stress_test_placement_group
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/placement_group_tests_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python stress_tests/test_placement_group.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: shuffle_1tb_1000_partition
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: shuffle_1tb_1000_partition
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_large_scale.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=1000 --partition-size=1e9
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: non_streaming_shuffle_1tb_1000_partition
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: non_streaming_shuffle_1tb_1000_partition
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_large_scale.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python shuffle/shuffle_test.py --num-partitions=1000 --partition-size=1e9
|
|
--no-streaming
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: shuffle_1tb_5000_partitions
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: shuffle_1tb_5000_partitions
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_large_scale.yaml
|
|
|
|
run:
|
|
timeout: 9000
|
|
script: python shuffle/shuffle_test.py --num-partitions=5000 --partition-size=200e6
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: decision_tree_autoscaling
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: decision_tree_autoscaling
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: decision_tree/decision_tree_app_config.yaml
|
|
cluster_compute: decision_tree/autoscaling_compute.yaml
|
|
|
|
run:
|
|
timeout: 3000
|
|
script: python decision_tree/cart_with_tree.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: decision_tree_autoscaling_20_runs
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: decision_tree_autoscaling_20_runs
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: decision_tree/decision_tree_app_config.yaml
|
|
cluster_compute: decision_tree/autoscaling_compute.yaml
|
|
|
|
run:
|
|
timeout: 9600
|
|
script: python decision_tree/cart_with_tree.py --concurrency=20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: autoscaling_shuffle_1tb_1000_partitions
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: autoscaling_shuffle_1tb_1000_partitions
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_autoscaling.yaml
|
|
|
|
run:
|
|
timeout: 4000
|
|
script: python shuffle/shuffle_test.py --num-partitions=1000 --partition-size=1e9
|
|
--no-streaming
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: pg_long_running_performance_test
|
|
group: core-multi-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: pg_long_running_performance_test
|
|
test_suite: nightly_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: placement_group_tests/app_config.yaml
|
|
cluster_compute: placement_group_tests/long_running_test_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python placement_group_tests/long_running_performance_test.py --num-stages
|
|
2000
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: microbenchmark
|
|
group: core-daily-test
|
|
team: core
|
|
frequency: nightly
|
|
working_dir: microbenchmark
|
|
env: prod
|
|
|
|
python: "3.7"
|
|
|
|
legacy:
|
|
test_name: microbenchmark
|
|
test_suite: microbenchmark
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_64.yaml
|
|
|
|
run:
|
|
timeout: 1800
|
|
script: OMP_NUM_THREADS=64 RAY_ADDRESS=local python run_microbenchmark.py
|
|
|
|
|
|
- name: microbenchmark_staging
|
|
group: core-daily-test
|
|
team: core
|
|
frequency: nightly
|
|
working_dir: microbenchmark
|
|
env: staging
|
|
|
|
python: "3.7"
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_64.yaml
|
|
|
|
run:
|
|
timeout: 1800
|
|
script: OMP_NUM_THREADS=64 RAY_ADDRESS=local python run_microbenchmark.py
|
|
|
|
|
|
- name: microbenchmark_38
|
|
group: core-daily-test
|
|
team: core
|
|
frequency: nightly
|
|
working_dir: microbenchmark
|
|
|
|
python: "3.8"
|
|
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: tpl_64.yaml
|
|
|
|
run:
|
|
timeout: 1800
|
|
script: OMP_NUM_THREADS=64 RAY_ADDRESS=local python run_microbenchmark.py
|
|
|
|
- name: dask_on_ray_10gb_sort
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dask_on_ray_10gb_sort
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: dask_on_ray/dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/dask_on_ray_sort_compute_template.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/dask_on_ray_sort.py --nbytes 10_000_000_000 --npartitions
|
|
50 --num-nodes 1 --ray --data-dir /tmp/ray --file-path /tmp/ray
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dask_on_ray_100gb_sort
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dask_on_ray_100gb_sort
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: dask_on_ray/dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/dask_on_ray_sort_compute_template.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/dask_on_ray_sort.py --nbytes 100_000_000_000 --npartitions
|
|
200 --num-nodes 1 --ray --data-dir /tmp/ray --file-path /tmp/ray
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dask_on_ray_large_scale_test_no_spilling
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dask_on_ray_large_scale_test_no_spilling
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: dask_on_ray/large_scale_dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/dask_on_ray_stress_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/large_scale_test.py --num_workers 20 --worker_obj_store_size_in_gb
|
|
20 --error_rate 0 --data_save_path /tmp/ray
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
app_config: dask_on_ray/large_scale_dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/large_scale_dask_on_ray_compute_template.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/large_scale_test.py --num_workers 4 --worker_obj_store_size_in_gb
|
|
20 --error_rate 0 --data_save_path /tmp/ray
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 5
|
|
|
|
- name: dask_on_ray_large_scale_test_spilling
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dask_on_ray_large_scale_test_spilling
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: dask_on_ray/large_scale_dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/dask_on_ray_stress_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/large_scale_test.py --num_workers 150 --worker_obj_store_size_in_gb
|
|
70 --error_rate 0 --data_save_path /tmp/ray
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
app_config: dask_on_ray/large_scale_dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/large_scale_dask_on_ray_compute_template.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/large_scale_test.py --num_workers 32 --worker_obj_store_size_in_gb
|
|
70 --error_rate 0 --data_save_path /tmp/ray
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 5
|
|
|
|
- name: stress_test_state_api_scale
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: stress_test_state_api_scale
|
|
test_suite: nightly_tests
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/stress_tests_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python stress_tests/test_state_api_scale.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
app_config: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/smoke_test_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python stress_tests/test_state_api_scale.py --smoke-test
|
|
|
|
|
|
- name: shuffle_20gb_with_state_api
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: shuffle_20gb_with_state_api
|
|
test_suite: nightly_tests
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/shuffle_compute_single.yaml
|
|
|
|
run:
|
|
timeout: 1000
|
|
script: python stress_tests/test_state_api_with_other_tests.py
|
|
nightly_tests/shuffle/shuffle_test.py --test-args="--num-partitions=100 --partition-size=200e6"
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: stress_test_many_tasks
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: stress_test_many_tasks
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/stress_tests_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python stress_tests/test_many_tasks.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
app_config: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/smoke_test_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python stress_tests/test_many_tasks.py --num-nodes=4 --smoke-test
|
|
|
|
- name: stress_test_dead_actors
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: stress_test_dead_actors
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/stress_tests_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python stress_tests/test_dead_actors.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
app_config: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/smoke_test_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python stress_tests/test_dead_actors.py --num-nodes=4 --num-parents=3
|
|
--num-children=3
|
|
|
|
# The full test is not stable, so run the smoke test only.
|
|
# See https://github.com/ray-project/ray/issues/23244.
|
|
- name: threaded_actors_stress_test
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: threaded_actors_stress_test
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: stress_tests/stress_tests_app_config.yaml
|
|
cluster_compute: stress_tests/smoke_test_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python stress_tests/test_threaded_actors.py --test-runtime 1800 --kill-interval_s
|
|
30
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 5
|
|
|
|
# - name: threaded_actors_stress_test
|
|
# group: core-daily-test
|
|
# working_dir: nightly_tests
|
|
# legacy:
|
|
# test_name: threaded_actors_stress_test
|
|
# test_suite: nightly_tests
|
|
#
|
|
# frequency: nightly
|
|
# team: core
|
|
# cluster:
|
|
# cluster_env: stress_tests/stress_tests_app_config.yaml
|
|
# cluster_compute: stress_tests/stress_test_threaded_actor_compute.yaml
|
|
#
|
|
# run:
|
|
# timeout: 7200
|
|
# script: python stress_tests/test_threaded_actors.py --test-runtime 3600 --kill-interval_s
|
|
# 60
|
|
#
|
|
# wait_for_nodes:
|
|
# num_nodes: 201
|
|
# timeout: 600
|
|
#
|
|
# type: sdk_command
|
|
# file_manager: sdk
|
|
#
|
|
# smoke_test:
|
|
# frequency: nightly
|
|
# cluster:
|
|
# app_config: stress_tests/stress_tests_app_config.yaml
|
|
# cluster_compute: stress_tests/smoke_test_compute.yaml
|
|
#
|
|
# run:
|
|
# timeout: 3600
|
|
# script: python stress_tests/test_threaded_actors.py --test-runtime 1800 --kill-interval_s
|
|
# 30
|
|
#
|
|
# wait_for_nodes:
|
|
# num_nodes: 5
|
|
# timeout: 600
|
|
|
|
- name: dask_on_ray_1tb_sort
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dask_on_ray_1tb_sort
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: dask_on_ray/dask_on_ray_app_config.yaml
|
|
cluster_compute: dask_on_ray/1tb_sort_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dask_on_ray/dask_on_ray_sort.py --nbytes 1_000_000_000_000 --npartitions
|
|
1000 --num-nodes 31 --ray --data-dir /tmp/ray --s3-bucket core-nightly-test
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: many_nodes_actor_test
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: many_nodes_actor_test
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: many_nodes_tests/app_config.yaml
|
|
cluster_compute: many_nodes_tests/compute_config.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python many_nodes_tests/actor_test.py
|
|
wait_for_nodes:
|
|
num_nodes: 251
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
|
|
- name: pg_autoscaling_regression_test
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: pg_autoscaling_regression_test
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: placement_group_tests/app_config.yaml
|
|
cluster_compute: placement_group_tests/compute.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python placement_group_tests/pg_run.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: placement_group_performance_test
|
|
group: core-daily-test
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: placement_group_performance_test
|
|
test_suite: nightly_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: placement_group_tests/app_config.yaml
|
|
cluster_compute: placement_group_tests/pg_perf_test_compute.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python placement_group_tests/placement_group_performance_test.py
|
|
wait_for_nodes:
|
|
num_nodes: 5
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
|
|
#########################
|
|
# Core Scalability Tests
|
|
#########################
|
|
|
|
- name: single_node
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: single_node
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: single_node.yaml
|
|
|
|
run:
|
|
timeout: 12000
|
|
prepare: sleep 0
|
|
script: python single_node/test_single_node.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: object_store
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: object_store
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: object_store.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python object_store/test_object_store.py
|
|
wait_for_nodes:
|
|
num_nodes: 50
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: many_actors
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: many_actors
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: distributed.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_many_actors.py
|
|
wait_for_nodes:
|
|
num_nodes: 65
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: many_actors_smoke_test
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: many_actors_smoke_test
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: distributed_smoke_test.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: SMOKE_TEST=1 python distributed/test_many_actors.py
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: many_tasks
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: many_tasks
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: distributed.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_many_tasks.py --num-tasks=10000
|
|
wait_for_nodes:
|
|
num_nodes: 65
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: distributed_smoke_test.yaml
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_many_tasks.py --num-tasks=100
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
|
|
- name: many_pgs
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: many_pgs
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: distributed.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_many_pgs.py
|
|
wait_for_nodes:
|
|
num_nodes: 65
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: many_pgs_smoke_test
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: many_pgs_smoke_test
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: distributed_smoke_test.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: SMOKE_TEST=1 python distributed/test_many_pgs.py
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: many_nodes
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
env: staging
|
|
|
|
legacy:
|
|
test_name: many_nodes
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: many_nodes.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_many_tasks.py --num-tasks=1000
|
|
wait_for_nodes:
|
|
num_nodes: 250
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: scheduling_test_many_0s_tasks_single_node
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: scheduling_test_many_0s_tasks_single_node
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: scheduling.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_scheduling.py --total-num-task=1984000 --num-cpu-per-task=1
|
|
--task-duration-s=0 --total-num-actors=1 --num-actors-per-nodes=1
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: scheduling_test_many_0s_tasks_many_nodes
|
|
group: core-scalability-test
|
|
working_dir: benchmarks
|
|
legacy:
|
|
test_name: scheduling_test_many_0s_tasks_many_nodes
|
|
test_suite: benchmark_tests
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: scheduling.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python distributed/test_scheduling.py --total-num-task=1984000 --num-cpu-per-task=1
|
|
--task-duration-s=0 --total-num-actors=32 --num-actors-per-nodes=1
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 32
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
# - name: scheduling_test_many_5s_tasks_single_node
|
|
# group: core-scalability-test
|
|
# working_dir: benchmarks
|
|
# legacy:
|
|
# test_name: scheduling_test_many_5s_tasks_single_node
|
|
# test_suite: benchmark_tests
|
|
|
|
# frequency: nightly
|
|
# team: core
|
|
# cluster:
|
|
# cluster_env: app_config.yaml
|
|
# cluster_compute: scheduling.yaml
|
|
|
|
# run:
|
|
# timeout: 3600
|
|
# script: python distributed/test_scheduling.py --total-num-task=1984000 --num-cpu-per-task=1
|
|
# --task-duration-s=5 --total-num-actors=1 --num-actors-per-nodes=1
|
|
|
|
# wait_for_nodes:
|
|
# num_nodes: 32
|
|
# timeout: 600
|
|
|
|
# type: sdk_command
|
|
# file_manager: sdk
|
|
# stable: false
|
|
|
|
# - name: scheduling_test_many_5s_tasks_many_nodes
|
|
# group: core-scalability-test
|
|
# working_dir: benchmarks
|
|
# legacy:
|
|
# test_name: scheduling_test_many_5s_tasks_many_nodes
|
|
# test_suite: benchmark_tests
|
|
|
|
# frequency: nightly
|
|
# team: core
|
|
# cluster:
|
|
# cluster_env: app_config.yaml
|
|
# cluster_compute: scheduling.yaml
|
|
|
|
# run:
|
|
# timeout: 3600
|
|
# script: python distributed/test_scheduling.py --total-num-task=1984000 --num-cpu-per-task=1
|
|
# --task-duration-s=5 --total-num-actors=32 --num-actors-per-nodes=1
|
|
|
|
# wait_for_nodes:
|
|
# num_nodes: 32
|
|
# timeout: 600
|
|
|
|
# type: sdk_command
|
|
# file_manager: sdk
|
|
# stable: false
|
|
|
|
###############
|
|
# Dataset tests
|
|
###############
|
|
|
|
- name: inference
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: inference
|
|
test_suite: dataset_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: inference.yaml
|
|
|
|
run:
|
|
timeout: 600
|
|
script: python inference.py
|
|
wait_for_nodes:
|
|
num_nodes: 2
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: shuffle_data_loader
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: shuffle_data_loader
|
|
test_suite: dataset_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle_app_config.yaml
|
|
cluster_compute: shuffle_compute.yaml
|
|
|
|
run:
|
|
timeout: 1800
|
|
script: python dataset_shuffle_data_loader.py
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: parquet_metadata_resolution
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: parquet_metadata_resolution
|
|
test_suite: dataset_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: pipelined_training_app.yaml
|
|
cluster_compute: pipelined_training_compute.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python parquet_metadata_resolution.py --num-files 915
|
|
wait_for_nodes:
|
|
num_nodes: 15
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dataset_random_access
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
stable: false
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: pipelined_training_app.yaml
|
|
cluster_compute: pipelined_training_compute.yaml
|
|
|
|
run:
|
|
timeout: 1200
|
|
script: python dataset_random_access.py
|
|
wait_for_nodes:
|
|
num_nodes: 15
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: pipelined_data_ingest_benchmark
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: data_ingest_benchmark_compute.yaml
|
|
|
|
run:
|
|
timeout: 300
|
|
script: python data_ingest_benchmark.py --dataset-size-gb=200 --num-workers=20 --streaming
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: pipelined_training_50_gb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: pipelined_training_50_gb
|
|
test_suite: dataset_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: pipelined_training_app.yaml
|
|
cluster_compute: pipelined_training_compute.yaml
|
|
|
|
run:
|
|
timeout: 4800
|
|
script: python pipelined_training.py --epochs 1
|
|
wait_for_nodes:
|
|
num_nodes: 15
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: pipelined_ingestion_1500_gb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: pipelined_ingestion_1500_gb
|
|
test_suite: dataset_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: pipelined_ingestion_app.yaml
|
|
cluster_compute: pipelined_ingestion_compute.yaml
|
|
|
|
run:
|
|
timeout: 9600
|
|
script: python pipelined_training.py --epochs 2 --num-windows 5 --num-files 915
|
|
--debug
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: datasets_ingest_train_infer
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: datasets_ingest_train_infer
|
|
test_suite: dataset_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: ray_sgd_training_app.yaml
|
|
cluster_compute: ray_sgd_training_compute.yaml
|
|
|
|
run:
|
|
timeout: 14400
|
|
script: python ray_sgd_training.py --address auto --use-s3 --num-workers 16 --use-gpu
|
|
--large-dataset
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 66
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
smoke_test:
|
|
frequency: multi
|
|
cluster:
|
|
app_config: ray_sgd_training_app.yaml
|
|
cluster_compute: ray_sgd_training_smoke_compute.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
script: python ray_sgd_training.py --address auto --use-s3 --num-workers 8 --use-gpu
|
|
wait_for_nodes:
|
|
num_nodes: 8
|
|
|
|
- name: datasets_preprocess_ingest
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: datasets_preprocess_ingest
|
|
test_suite: dataset_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: ray_sgd_training_app.yaml
|
|
cluster_compute: ray_sgd_training_compute_no_gpu.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python ray_sgd_training.py --address auto --use-s3 --num-workers 16 --use-gpu
|
|
--large-dataset --debug
|
|
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: datasets_ingest_400G
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests/dataset
|
|
legacy:
|
|
test_name: datasets_ingest_400G
|
|
test_suite: dataset_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: ray_sgd_training_app.yaml
|
|
cluster_compute: dataset_ingest_400G_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python ray_sgd_runner.py --address auto --use-gpu --num-epochs 1
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dataset_shuffle_random_shuffle_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dataset_shuffle_random_shuffle_1tb
|
|
test_suite: dataset_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dataset/sort.py --num-partitions=1000 --partition-size=1e9 --shuffle
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dataset_shuffle_sort_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dataset_shuffle_sort_1tb
|
|
test_suite: dataset_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: python dataset/sort.py --num-partitions=1000 --partition-size=1e9
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dataset_shuffle_push_based_random_shuffle_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dataset_shuffle_push_based_random_shuffle_1tb
|
|
test_suite: dataset_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: RAY_DATASET_PUSH_BASED_SHUFFLE=1 python dataset/sort.py --num-partitions=1000 --partition-size=1e9 --shuffle
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dataset_shuffle_push_based_sort_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dataset_shuffle_push_based_sort_1tb
|
|
test_suite: dataset_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
script: RAY_DATASET_PUSH_BASED_SHUFFLE=1 python dataset/sort.py --num-partitions=1000 --partition-size=1e9
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: dataset_shuffle_push_based_random_shuffle_100tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: dataset_shuffle_push_based_random_shuffle_100tb
|
|
test_suite: dataset_test
|
|
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/100tb_shuffle_app_config.yaml
|
|
cluster_compute: shuffle/100tb_shuffle_compute.yaml
|
|
|
|
run:
|
|
timeout: 28800
|
|
script: RAY_DATASET_PUSH_BASED_SHUFFLE=1 python dataset/sort.py --num-partitions=100000 --partition-size=1e9 --shuffle
|
|
wait_for_nodes:
|
|
num_nodes: 100
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
##################
|
|
# Core Chaos tests
|
|
##################
|
|
|
|
- name: chaos_many_tasks_no_object_store
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_many_tasks_no_object_store
|
|
test_suite: chaos_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: chaos_test/app_config.yaml
|
|
cluster_compute: chaos_test/compute_template.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
wait_for_nodes:
|
|
num_nodes: 10
|
|
prepare: python setup_chaos.py --no-start
|
|
script: python chaos_test/test_chaos_basic.py --workload=tasks
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_many_actors
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_many_actors
|
|
test_suite: chaos_test
|
|
|
|
frequency: multi
|
|
team: core
|
|
cluster:
|
|
cluster_env: chaos_test/app_config.yaml
|
|
cluster_compute: chaos_test/compute_template.yaml
|
|
|
|
run:
|
|
timeout: 3600
|
|
wait_for_nodes:
|
|
num_nodes: 10
|
|
prepare: python setup_chaos.py --no-start
|
|
script: python chaos_test/test_chaos_basic.py --workload=actors
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_dask_on_ray_large_scale_test_no_spilling
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_dask_on_ray_large_scale_test_no_spilling
|
|
test_suite: chaos_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: chaos_test/dask_on_ray_app_config_reconstruction.yaml
|
|
cluster_compute: dask_on_ray/chaos_dask_on_ray_stress_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
prepare: python setup_chaos.py --node-kill-interval 100
|
|
script: python dask_on_ray/large_scale_test.py --num_workers 20 --worker_obj_store_size_in_gb
|
|
20 --error_rate 0 --data_save_path /tmp/ray
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_dask_on_ray_large_scale_test_spilling
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_dask_on_ray_large_scale_test_spilling
|
|
test_suite: chaos_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: chaos_test/dask_on_ray_app_config_reconstruction.yaml
|
|
cluster_compute: dask_on_ray/dask_on_ray_stress_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
prepare: python setup_chaos.py --node-kill-interval 100
|
|
script: python dask_on_ray/large_scale_test.py --num_workers 150 --worker_obj_store_size_in_gb
|
|
70 --error_rate 0 --data_save_path /tmp/ray
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_pipelined_ingestion_1500_gb_15_windows
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_pipelined_ingestion_1500_gb_15_windows
|
|
test_suite: chaos_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: dataset/pipelined_ingestion_app.yaml
|
|
cluster_compute: dataset/pipelined_ingestion_compute.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
wait_for_nodes:
|
|
num_nodes: 21
|
|
prepare: ' python setup_chaos.py --node-kill-interval 300'
|
|
script: python dataset/pipelined_training.py --epochs 1 --num-windows 15 --num-files
|
|
915 --debug
|
|
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_dataset_shuffle_push_based_sort_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_dataset_shuffle_push_based_sort_1tb
|
|
test_suite: chaos_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
prepare: ' python setup_chaos.py --node-kill-interval 1200 --max-nodes-to-kill 3'
|
|
script: RAY_DATASET_PUSH_BASED_SHUFFLE=1 python dataset/sort.py --num-partitions=1000 --partition-size=1e9
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_dataset_shuffle_sort_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_dataset_shuffle_sort_1tb
|
|
test_suite: chaos_test
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
prepare: ' python setup_chaos.py --node-kill-interval 900 --max-nodes-to-kill 3'
|
|
script: python dataset/sort.py --num-partitions=1000 --partition-size=1e9
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_dataset_shuffle_random_shuffle_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_dataset_shuffle_random_shuffle_1tb
|
|
test_suite: chaos_test
|
|
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
prepare: ' python setup_chaos.py --node-kill-interval 600 --max-nodes-to-kill 2'
|
|
script: python dataset/sort.py --num-partitions=1000 --partition-size=1e9 --shuffle
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: chaos_dataset_shuffle_push_based_random_shuffle_1tb
|
|
group: core-dataset-tests
|
|
working_dir: nightly_tests
|
|
legacy:
|
|
test_name: chaos_dataset_shuffle_push_based_random_shuffle_1tb
|
|
test_suite: chaos_test
|
|
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: core
|
|
cluster:
|
|
cluster_env: shuffle/shuffle_app_config.yaml
|
|
cluster_compute: shuffle/datasets_large_scale_compute_small_instances.yaml
|
|
|
|
run:
|
|
timeout: 7200
|
|
prepare: ' python setup_chaos.py --node-kill-interval 600 --max-nodes-to-kill 2'
|
|
script: RAY_DATASET_PUSH_BASED_SHUFFLE=1 python dataset/sort.py --num-partitions=1000 --partition-size=1e9 --shuffle
|
|
wait_for_nodes:
|
|
num_nodes: 20
|
|
type: sdk_command
|
|
file_manager: sdk
|
|
|
|
- name: k8s_serve_ha_test
|
|
group: k8s-test
|
|
working_dir: k8s_tests
|
|
legacy:
|
|
test_name: k8s_serve_ha_test
|
|
test_suite: k8s_tests
|
|
|
|
stable: false
|
|
|
|
frequency: nightly
|
|
team: serve
|
|
cluster:
|
|
cluster_env: app_config.yaml
|
|
cluster_compute: compute_tpl.yaml
|
|
|
|
run:
|
|
timeout: 1800
|
|
prepare: bash prepare.sh
|
|
script: python run_gcs_ft_on_k8s.py
|
|
type: sdk_command
|
|
file_manager: sdk
|