[Release] Fix release tests (#9976)

* In Progress.

* Fix torch version.

* Fix asan run.sh

* Addressed code review.

ci/asan_tests/run_asan_tests.sh

@@ -5,16 +5,20 @@ set -euxo pipefail
 ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE:-$0}")"; pwd)"
 
 asan_install() {
-  (cd "${ROOT_DIR}"/../python && pip install -e . --verbose)
+  echo "installing"
+  (cd "${ROOT_DIR}"/../ray/python && pip install -e . --verbose)
 }
 
 asan_setup() {
   echo "Setting up the environment"
+  pip uninstall -y ray || true
+  conda uninstall -y wrapt || true
+  pip install wrapt || true
   pip install -r ray-project/requirements.txt
   pip install -U pytest==5.4.3
 
   echo "Installing cython example"
-  (cd "${ROOT_DIR}"/../doc/examples/cython && python setup.py install --user)
+  (cd "${ROOT_DIR}"/../ray/doc/examples/cython && python setup.py install --user)
 
   echo "Settting up the shell"
   echo "build --config=asan" >> ~/.bazelrc  # Setup cache
@@ -22,6 +26,7 @@ asan_setup() {
   echo "ASAN_OPTIONS=detect_leaks=0" >> ~/.bashrc
 
   echo "Compiling ray"
+  cd "${ROOT_DIR}"/../ray/
   git fetch
   git pull origin master
   asan_install || true
@@ -32,16 +37,13 @@ asan_run() {
     export LD_PRELOAD="/usr/lib/gcc/x86_64-linux-gnu/7/libasan.so"
     export ASAN_OPTIONS="detect_leaks=0"
 
-    cd "${ROOT_DIR}"/../..
-
-    # async plasma test
-    python -m pytest -v --durations=5 --timeout=300 python/ray/tests/test_async.py
+    cd "${ROOT_DIR}"/../ray
 
     # Ray tests
-    bazel test --test_tag_filters=-jenkins_only python/ray/serve/...
-    bazel test --test_tag_filters=-jenkins_only python/ray/dashboard/...
-    bazel test --test_tag_filters=-jenkins_only python/ray/tests/...
-    bazel test --test_tag_filters=-jenkins_only python/ray/tune/...
+    bazel test --test_tag_filters=-jenkins_only --test_output=streamed python/ray/serve/...
+    bazel test --test_tag_filters=-jenkins_only --test_output=streamed python/ray/dashboard/...
+    bazel test --test_tag_filters=-jenkins_only --test_output=streamed python/ray/tests/...
+    bazel test --test_tag_filters=-jenkins_only --test_output=streamed python/ray/tune/...
   )
 }
 

ci/long_running_tests/run.sh

@@ -51,6 +51,7 @@ echo "workload: $workload"
 wheel="https://s3-us-west-2.amazonaws.com/ray-wheels/$ray_branch/$commit/ray-$ray_version-cp36-cp36m-manylinux1_x86_64.whl"
 
 pip install -U pip
+unset RAY_ADDRESS
 source activate tensorflow_p36 && pip install -q -U "$wheel" Click
 source activate tensorflow_p36 && pip install -q "ray[all]" "gym[atari]"
 source activate tensorflow_p36 && python "workloads/$workload.py"

ci/microbenchmark/run.sh

@@ -48,4 +48,5 @@ wget "https://s3-us-west-2.amazonaws.com/ray-wheels/$ray_branch/$commit/ray-$ray
 pip uninstall -y -q ray
 pip install -U "ray-$ray_version-cp36-cp36m-manylinux1_x86_64.whl"
 
+unset RAY_ADDRESS
 OMP_NUM_THREADS=64 ray microbenchmark

ci/regression_test/rllib_regresssion_tests/run.sh

@@ -1,4 +1,4 @@
 #!/usr/bin/env bash
 
-source activate tensorflow_p36 && pip install torch torchvision
+source activate tensorflow_p36 && pip install torch==1.4 torchvision
 source activate tensorflow_p36 && rllib train -f compact-regression-test.yaml

ci/regression_test/stress_tests/ray-project/cluster.yaml

@@ -97,7 +97,7 @@ setup_commands:
     # - git clone https://github.com/ray-project/ray || true
     # - ray/ci/travis/install-bazel.sh
     - pip install -U pip 
-    - conda uninstall -y terminado
+    - conda uninstall -y terminado || true
     - pip install terminado
     - pip install boto3==1.4.8 cython==0.29.0
     # - cd ray/python; git checkout master; git pull; pip install -e . --verbose

ci/regression_test/stress_tests/test_dead_actors.py

@@ -10,7 +10,7 @@ import ray
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-ray.init(address="localhost:6379")
+ray.init(address="auto")
 
 # These numbers need to correspond with the autoscaler config file.
 # The number of remote nodes in the autoscaler should upper bound

ci/regression_test/stress_tests/test_many_tasks.py

@@ -9,7 +9,7 @@ import ray
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-ray.init(address="localhost:6379")
+ray.init(address="auto")
 
 # These numbers need to correspond with the autoscaler config file.
 # The number of remote nodes in the autoscaler should upper bound