[pull] master from ray-project:master

.buildkite/pipeline.build.yml

@@ -220,6 +220,9 @@
     # Todo (krfricke): Move mosaicml to train-test-requirements.txt
     - pip install "mosaicml==0.12.1"
     - DOC_TESTING=1 ./ci/env/install-dependencies.sh
+    # TODO(scottjlee): Move datasets to train/data-test-requirements.txt 
+    # (see https://github.com/ray-project/ray/pull/38432/)
+    - pip install "datasets==2.14.0"
     - ./ci/env/install-horovod.sh
     - ./ci/env/env_info.sh
     - bazel test --config=ci $(./scripts/bazel_export_options)
@@ -383,6 +386,57 @@
     - ray job submit --address http://localhost:8265 --runtime-env python/ray/tests/chaos/runtime_env.yaml --working-dir python/ray/tests/chaos -- python potato_passer.py --num-actors=3 --pass-times=1000 --sleep-secs=0.01
 
 
+- label: ":kubernetes: :mending_heart: :ray-serve: serve chaos network delay test"
+  conditions: ["RAY_CI_LINUX_WHEELS_AFFECTED"]
+  instance_size: medium
+  commands:
+    - |
+      cleanup() {
+        if [ "${BUILDKITE_PULL_REQUEST}" = "false" ]; then ./ci/build/upload_build_info.sh; fi
+        kind delete cluster
+      }
+      trap cleanup EXIT
+    - ./ci/env/install-minimal.sh 3.8
+    - PYTHON=3.8 ./ci/env/install-dependencies.sh
+    # Specifying above somehow messes up the Ray install.
+    # Uninstall and re-install Ray so that we can use Ray Client.
+    # (Remove thirdparty_files to sidestep an issue with psutil.)
+    - pip uninstall -y ray && rm -rf /ray/python/ray/thirdparty_files
+    - pip install -e /ray/python
+    - echo "--- Setting up local kind cluster."
+    - ./ci/k8s/prep-k8s-environment.sh
+    - ./ci/k8s/prep-helm.sh
+    - echo "--- Building py38-cpu Ray image for the test."
+    - LINUX_WHEELS=1 ./ci/ci.sh build
+    - pip install -q docker
+    - python ci/build/build-docker-images.py --py-versions py38 --device-types cpu --build-type LOCAL --build-base
+    # Tag the image built in the last step. We want to be sure to distinguish the image from the real Ray nightly.
+    - docker tag rayproject/ray:nightly-py38-cpu ray-ci:kuberay-test
+    # Load the image into the kind node
+    - kind load docker-image ray-ci:kuberay-test
+    # Helm install KubeRay
+    - echo "--- Installing KubeRay operator and cluser."
+    - helm repo add kuberay https://ray-project.github.io/kuberay-helm/
+    - helm install kuberay-operator kuberay/kuberay-operator
+    - kubectl wait pod  -l app.kubernetes.io/name=kuberay-operator --for=condition=Ready=True  --timeout=5m
+    # We are in m4i.xlarge and have 4 cpus. Can't have too many nodes.
+    - helm install raycluster kuberay/ray-cluster --set image.repository=ray-ci --set image.tag=kuberay-test --set worker.replicas=2 --set worker.resources.limits.cpu=500m --set worker.resources.requests.cpu=500m --set head.resources.limits.cpu=500m --set head.resources.requests.cpu=500m --set head.containerEnv[0].name=RAY_SERVE_ENABLE_EXPERIMENTAL_STREAMING --set head.containerEnv[0].value=\"1\"
+    - kubectl wait pod -l ray.io/cluster=raycluster-kuberay --for=condition=Ready=True --timeout=5m
+    - kubectl port-forward --address 0.0.0.0 service/raycluster-kuberay-head-svc 8265:8265 &
+    # Helm install chaos-mesh
+    - echo "--- Installing chaos-mesh operator and CR."
+    - helm repo add chaos-mesh https://charts.chaos-mesh.org
+    - kubectl create ns chaos-mesh
+    - helm install chaos-mesh chaos-mesh/chaos-mesh -n=chaos-mesh --set chaosDaemon.runtime=containerd --set chaosDaemon.socketPath=/run/containerd/containerd.sock --version 2.6.1
+    - kubectl wait pod  --namespace chaos-mesh  -l app.kubernetes.io/instance=chaos-mesh --for=condition=Ready=True
+    - echo "--- Running the script without faults"
+    - ray job submit --address http://localhost:8265 --runtime-env python/ray/tests/chaos/streaming_llm.yaml --working-dir python/ray/tests/chaos -- python streaming_llm.py --num_queries_per_task=100 --num_tasks=2 --num_words_per_query=100
+    # Now add the delay, rerun the job
+    - kubectl apply -f python/ray/tests/chaos/chaos_network_delay.yaml
+    - echo "--- Running the script with fault of networking delay"
+    - ray job submit --address http://localhost:8265 --runtime-env python/ray/tests/chaos/streaming_llm.yaml --working-dir python/ray/tests/chaos -- python streaming_llm.py --num_queries_per_task=100 --num_tasks=2 --num_words_per_query=100
+
+
 - label: ":book: Documentation"
   commands:
     - export LINT=1
@@ -445,4 +499,6 @@
       - TRAIN_MINIMAL_INSTALL=1 ./ci/env/install-minimal.sh
       - ./ci/env/env_info.sh
       - python ./ci/env/check_minimal_install.py
-      - bazel test --config=ci $(./ci/run/bazel_export_options)  --build_tests_only --test_tag_filters=minimal python/ray/train/...
+      - bazel test --config=ci $(./ci/run/bazel_export_options)  --build_tests_only --test_tag_filters=minimal 
+        --test_env=RAY_AIR_NEW_PERSISTENCE_MODE=1
+        python/ray/train/...

.buildkite/pipeline.gpu_large.yml

@@ -67,11 +67,13 @@
     - DOC_TESTING=1 TRAIN_TESTING=1 TUNE_TESTING=1 ./ci/env/install-dependencies.sh
     - pip install -Ur ./python/requirements/ml/dl-gpu-requirements.txt
     - ./ci/env/install-horovod.sh
-    - ./ci/env/env_info.sh
     # Test examples with newer version of `transformers`
     # TODO(amogkam): Remove when https://github.com/ray-project/ray/issues/36011 
     # is resolved.
-    - pip install transformers==4.30.2
+    # TODO(scottjlee): Move datasets to train/data-test-requirements.txt 
+    # (see https://github.com/ray-project/ray/pull/38432/)
+    - pip install transformers==4.30.2 datasets==2.14.0
+    - ./ci/env/env_info.sh
     - bazel test --config=ci $(./scripts/bazel_export_options)
       --test_tag_filters=doctest,-cpu python/ray/... doc/...
 

.buildkite/pipeline.ml.yml

@@ -32,7 +32,7 @@
 - label: ":steam_locomotive: Train tests and examples"
   conditions: ["NO_WHEELS_REQUIRED", "RAY_CI_TRAIN_AFFECTED"]
   instance_size: large
-  parallelism: 4
+  parallelism: 3
   commands:
     - cleanup() { if [ "${BUILDKITE_PULL_REQUEST}" = "false" ]; then ./ci/build/upload_build_info.sh; fi }; trap cleanup EXIT
     # Todo (krfricke): Move mosaicml to train-test-requirements.txt
@@ -343,7 +343,7 @@
 - label: ":steam_locomotive: :floppy_disk: New persistence mode: Train tests and examples"
   conditions: ["NO_WHEELS_REQUIRED", "RAY_CI_TRAIN_AFFECTED"]
   instance_size: large
-  parallelism: 4
+  parallelism: 3
   commands:
     - cleanup() { if [ "${BUILDKITE_PULL_REQUEST}" = "false" ]; then ./ci/build/upload_build_info.sh; fi }; trap cleanup EXIT
     # Todo (krfricke): Move mosaicml to train-test-requirements.txt
@@ -454,6 +454,9 @@
   commands:
     - cleanup() { if [ "${BUILDKITE_PULL_REQUEST}" = "false" ]; then ./ci/build/upload_build_info.sh; fi }; trap cleanup EXIT
     - DATA_PROCESSING_TESTING=1 ARROW_VERSION=12.* ./ci/env/install-dependencies.sh
+    # TODO(scottjlee): Move datasets to train/data-test-requirements.txt 
+    # (see https://github.com/ray-project/ray/pull/38432/)
+    - pip install "datasets==2.14.0"
     - ./ci/env/env_info.sh
     - ./ci/run/run_bazel_test_with_sharding.sh --config=ci $(./ci/run/bazel_export_options) --action_env=RAY_DATA_USE_STREAMING_EXECUTOR=1 --build_tests_only --test_tag_filters=-data_integration,-doctest python/ray/data/...
     - ./ci/run/run_bazel_test_with_sharding.sh --config=ci $(./ci/run/bazel_export_options) --action_env=RAY_DATA_USE_STREAMING_EXECUTOR=1 --build_tests_only --test_tag_filters=ray_data,-doctest python/ray/air/...
@@ -465,6 +468,9 @@
   commands:
     - cleanup() { if [ "${BUILDKITE_PULL_REQUEST}" = "false" ]; then ./ci/build/upload_build_info.sh; fi }; trap cleanup EXIT
     - DATA_PROCESSING_TESTING=1 ARROW_VERSION=nightly ./ci/env/install-dependencies.sh
+    # TODO(scottjlee): Move datasets to train/data-test-requirements.txt 
+    # (see https://github.com/ray-project/ray/pull/38432/)
+    - pip install "datasets==2.14.0"
     - ./ci/env/env_info.sh
     - ./ci/run/run_bazel_test_with_sharding.sh --config=ci $(./ci/run/bazel_export_options) --build_tests_only --test_tag_filters=-data_integration,-doctest python/ray/data/...
     - ./ci/run/run_bazel_test_with_sharding.sh --config=ci $(./ci/run/bazel_export_options) --build_tests_only --test_tag_filters=ray_data,-doctest python/ray/air/...
@@ -476,6 +482,9 @@
   commands:
     - cleanup() { if [ "${BUILDKITE_PULL_REQUEST}" = "false" ]; then ./ci/build/upload_build_info.sh; fi }; trap cleanup EXIT
     - DATA_PROCESSING_TESTING=1 ARROW_VERSION=12.* ./ci/env/install-dependencies.sh
+    # TODO(scottjlee): Move datasets to train/data-test-requirements.txt 
+    # (see https://github.com/ray-project/ray/pull/38432/)
+    - pip install "datasets==2.14.0"
     - ./ci/env/env_info.sh
     - ./ci/run/run_bazel_test_with_sharding.sh --config=ci $(./ci/run/bazel_export_options) --build_tests_only --test_tag_filters=-data_integration,-doctest python/ray/data/...
     - ./ci/run/run_bazel_test_with_sharding.sh --config=ci $(./ci/run/bazel_export_options) --build_tests_only --test_tag_filters=ray_data,-doctest python/ray/air/...
@@ -510,7 +519,9 @@
     - DOC_TESTING=1 INSTALL_HOROVOD=1 ./ci/env/install-dependencies.sh
     # TODO (shrekris-anyscale): Remove transformers after core transformer
     # requirement is upgraded
-    - pip install "transformers==4.30.2"
+    # TODO(scottjlee): Move datasets to train/data-test-requirements.txt 
+    # (see https://github.com/ray-project/ray/pull/38432/)
+    - pip install "transformers==4.30.2" "datasets==2.14.0"
     - ./ci/env/env_info.sh
     - bazel test --config=ci $(./ci/run/bazel_export_options) --build_tests_only --test_tag_filters=-timeseries_libs,-external,-ray_air,-gpu,-post_wheel_build,-doctest,-datasets_train,-highly_parallel doc/...
 

LICENSE

@@ -186,7 +186,7 @@
       same "printed page" as the copyright notice for easier
       identification within third-party archives.
 
-   Copyright {yyyy} {name of copyright owner}
+   Copyright 2023 Ray Authors
 
    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.
@@ -447,4 +447,4 @@ Unless required by applicable law or agreed to in writing, software
 distributed under the License is distributed on an "AS IS" BASIS,
 WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 See the License for the specific language governing permissions and
-limitations under the License.
+limitations under the License.

dashboard/modules/metrics/dashboards/serve_dashboard_panels.py

@@ -139,8 +139,8 @@
         unit="replicas",
         targets=[
             Target(
-                expr="sum(ray_serve_deployment_replica_healthy{{{global_filters}}}) by (deployment)",
-                legend="{{deployment}}",
+                expr="sum(ray_serve_deployment_replica_healthy{{{global_filters}}}) by (application, deployment)",
+                legend="{{application, deployment}}",
             ),
         ],
         grid_pos=GridPos(0, 2, 8, 8),
@@ -152,8 +152,8 @@
         unit="qps",
         targets=[
             Target(
-                expr='sum(rate(ray_serve_deployment_request_counter{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (deployment)',
-                legend="{{deployment}}",
+                expr='sum(rate(ray_serve_deployment_request_counter{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (application, deployment)',
+                legend="{{application, deployment}}",
             ),
         ],
         grid_pos=GridPos(8, 2, 8, 8),
@@ -165,8 +165,8 @@
         unit="qps",
         targets=[
             Target(
-                expr='sum(rate(ray_serve_deployment_error_counter{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (deployment)',
-                legend="{{deployment}}",
+                expr='sum(rate(ray_serve_deployment_error_counter{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (application, deployment)',
+                legend="{{application, deployment}}",
             ),
         ],
         grid_pos=GridPos(16, 2, 8, 8),
@@ -178,8 +178,8 @@
         unit="ms",
         targets=[
             Target(
-                expr='histogram_quantile(0.5, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (deployment, le))',
-                legend="{{deployment}}",
+                expr='histogram_quantile(0.5, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (application, deployment, le))',
+                legend="{{application, deployment}}",
             ),
             Target(
                 expr='histogram_quantile(0.5, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (le))',
@@ -197,8 +197,8 @@
         unit="ms",
         targets=[
             Target(
-                expr='histogram_quantile(0.9, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (deployment, le))',
-                legend="{{deployment}}",
+                expr='histogram_quantile(0.9, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (application, deployment, le))',
+                legend="{{application, deployment}}",
             ),
             Target(
                 expr='histogram_quantile(0.9, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (le))',
@@ -216,8 +216,8 @@
         unit="ms",
         targets=[
             Target(
-                expr='histogram_quantile(0.99, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (deployment, le))',
-                legend="{{deployment}}",
+                expr='histogram_quantile(0.99, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (application, deployment, le))',
+                legend="{{application, deployment}}",
             ),
             Target(
                 expr='histogram_quantile(0.99, sum(rate(ray_serve_deployment_processing_latency_ms_bucket{{route=~"$Route",route!~"/-/.*",{global_filters}}}[5m])) by (le))',
@@ -235,8 +235,8 @@
         unit="requests",
         targets=[
             Target(
-                expr="sum(ray_serve_deployment_queued_queries{{{global_filters}}}) by (deployment)",
-                legend="{{deployment}}",
+                expr="sum(ray_serve_deployment_queued_queries{{{global_filters}}}) by (application, deployment)",
+                legend="{{application, deployment}}",
             ),
         ],
         fill=0,