Merge pull request #11697 from velconia/merge_fix_kill_bug_to_14

Fix kill fail bug (#11635)

Dockerfile

@@ -76,7 +76,8 @@ RUN easy_install -U pip && \
     pip install sphinx-rtd-theme==0.1.9 recommonmark
 
 RUN pip install pre-commit 'ipython==5.3.0' && \
-    pip install 'ipykernel==4.6.0' 'jupyter==1.0.0'
+    pip install 'ipykernel==4.6.0' 'jupyter==1.0.0' && \
+    pip install opencv-python
 
 #For docstring checker
 RUN pip install pylint pytest astroid isort

paddle/scripts/paddle_build.sh

@@ -22,7 +22,7 @@
 function print_usage() {
     echo -e "\n${RED}Usage${NONE}:
     ${BOLD}${SCRIPT_NAME}${NONE} [OPTION]"
-    
+
     echo -e "\n${RED}Options${NONE}:
     ${BLUE}build${NONE}: run build for x86 platform
     ${BLUE}build_android${NONE}: run build for android platform
@@ -198,7 +198,7 @@ function build_android() {
     fi
 
     ANDROID_STANDALONE_TOOLCHAIN=$ANDROID_TOOLCHAINS_DIR/$ANDROID_ARCH-android-$ANDROID_API
-    
+
     cat <<EOF
     ============================================
     Generating the standalone toolchain ...
@@ -212,13 +212,13 @@ EOF
           --arch=$ANDROID_ARCH \
           --platform=android-$ANDROID_API \
           --install-dir=$ANDROID_STANDALONE_TOOLCHAIN
-    
+
     BUILD_ROOT=${PADDLE_ROOT}/build_android
     DEST_ROOT=${PADDLE_ROOT}/install_android
-    
+
     mkdir -p $BUILD_ROOT
     cd $BUILD_ROOT
-    
+
     if [ $ANDROID_ABI == "armeabi-v7a" ]; then
       cmake -DCMAKE_SYSTEM_NAME=Android \
             -DANDROID_STANDALONE_TOOLCHAIN=$ANDROID_STANDALONE_TOOLCHAIN \
@@ -286,7 +286,7 @@ function build_ios() {
           -DWITH_TESTING=OFF \
           -DWITH_SWIG_PY=OFF \
           -DCMAKE_BUILD_TYPE=Release
-    
+
     make -j 2
 }
 
@@ -331,14 +331,14 @@ EOF
 function bind_test() {
     # the number of process to run tests
     NUM_PROC=6
-    
+
     # calculate and set the memory usage for each process
     MEM_USAGE=$(printf "%.2f" `echo "scale=5; 1.0 / $NUM_PROC" | bc`)
     export FLAGS_fraction_of_gpu_memory_to_use=$MEM_USAGE
-    
+
     # get the CUDA device count
     CUDA_DEVICE_COUNT=$(nvidia-smi -L | wc -l)
-    
+
     for (( i = 0; i < $NUM_PROC; i++ )); do
         cuda_list=()
         for (( j = 0; j < $CUDA_DEVICE_COUNT; j++ )); do

python/paddle/fluid/tests/unittests/CMakeLists.txt

@@ -15,7 +15,7 @@ if(NOT WITH_DISTRIBUTE)
 endif(NOT WITH_DISTRIBUTE)
 
 list(REMOVE_ITEM TEST_OPS test_seq_concat_op) # FIXME(helin): https://github.com/PaddlePaddle/Paddle/issues/8290
-list(REMOVE_ITEM TEST_OPS test_modified_huber_loss_op) # FIXME(qijun) https://github.com/PaddlePaddle/Paddle/issues/5184 
+list(REMOVE_ITEM TEST_OPS test_modified_huber_loss_op) # FIXME(qijun) https://github.com/PaddlePaddle/Paddle/issues/5184
 list(REMOVE_ITEM TEST_OPS test_lstm_unit_op) # # FIXME(qijun) https://github.com/PaddlePaddle/Paddle/issues/5185
 list(REMOVE_ITEM TEST_OPS test_nce) # FIXME(qijun) https://github.com/PaddlePaddle/Paddle/issues/7778
 list(REMOVE_ITEM TEST_OPS test_recurrent_op) # FIXME(qijun) https://github.com/PaddlePaddle/Paddle/issues/6152
@@ -43,8 +43,6 @@ list(REMOVE_ITEM TEST_OPS test_warpctc_op)
 list(REMOVE_ITEM TEST_OPS test_dist_train)
 list(REMOVE_ITEM TEST_OPS test_parallel_executor_crf)
 list(REMOVE_ITEM TEST_OPS test_parallel_executor_fetch_feed)
-# TODO(wuyi): this test hungs on CI, will add it back later
-list(REMOVE_ITEM TEST_OPS test_listen_and_serv_op)
 foreach(TEST_OP ${TEST_OPS})
     py_test_modules(${TEST_OP} MODULES ${TEST_OP})
 endforeach(TEST_OP)
@@ -52,3 +50,4 @@ py_test_modules(test_warpctc_op MODULES test_warpctc_op ENVS FLAGS_warpctc_dir=$
 py_test_modules(test_dist_train MODULES test_dist_train SERIAL)
 py_test_modules(test_parallel_executor_crf MODULES test_parallel_executor_crf SERIAL)
 py_test_modules(test_parallel_executor_fetch_feed MODULES test_parallel_executor_fetch_feed SERIAL)
+set_tests_properties(test_listen_and_serv_op PROPERTIES TIMEOUT 20)

python/paddle/fluid/tests/unittests/test_listen_and_serv_op.py

@@ -94,7 +94,7 @@ class TestListenAndServOp(OpTest):
         self._wait_ps_ready(p1.pid)
 
         # raise SIGTERM to pserver
-        os.kill(p1.pid, signal.SIGKILL)
+        os.kill(p1.pid, signal.SIGINT)
         p1.join()
 
         # run pserver on CPU in async mode
@@ -102,7 +102,7 @@ class TestListenAndServOp(OpTest):
         self._wait_ps_ready(p2.pid)
 
         # raise SIGTERM to pserver
-        os.kill(p2.pid, signal.SIGKILL)
+        os.kill(p2.pid, signal.SIGTERM)
         p2.join()