Improve ci

.circleci/config.yml

@@ -79,12 +79,8 @@ jobs:
           command: |
 
             # pytest on cuda
-            export test1_cmd='CUDA_VISIBLE_DEVICES=0 py.test -vvv tests/ -k "on_cuda"'
-            docker exec -it pthd /bin/bash -c "$test1_cmd"
-
-            # pytest on cuda and distributed (one process)
-            export test2_cmd='export WORLD_SIZE=1 && py.test --dist=each --tx $WORLD_SIZE*popen//python=python3.7 tests -m distributed -vvv'
-            docker exec -it pthd /bin/bash -c "$test2_cmd"
+            export test_cmd='sh tests/run_gpu_tests.sh'
+            docker exec -it pthd /bin/bash -c "$test_cmd"
 
             # MNIST tests
 
@@ -132,8 +128,8 @@ jobs:
       - run:
           name: Run 1 Node 2 GPUs Unit Tests
           command: |
-            export test1_cmd='export WORLD_SIZE=2 && py.test --dist=each --tx $WORLD_SIZE*popen//python=python3.7 tests -m distributed -vvv'
-            docker exec -it pthd /bin/bash -c "$test1_cmd"
+            export test_cmd='sh tests/run_gpu_tests.sh 2'
+            docker exec -it pthd /bin/bash -c "$test_cmd"
 
 # -------------------------------------------------------------------------------------
 # Workflows

.github/workflows/tpu-tests.yml

@@ -49,4 +49,4 @@ jobs:
         export XRT_WORKERS="localservice:0;grpc://localhost:40934"
 
         python -c "import torch_xla; print('torch xla version:', torch_xla.__version__)"
-        py.test --cov ignite --cov-append --cov-report term-missing tests/ -vvv -m tpu
+        sh tests/run_tpu_tests.sh

.travis.yml

@@ -41,15 +41,7 @@ install:
   - python setup.py install
 
 script:
-  - CUDA_VISIBLE_DEVICES="" py.test --tx 4*popen//python=python$TRAVIS_PYTHON_VERSION --cov ignite --cov-report term-missing -vvv tests/
-  # Run test on cuda device
-  # As no GPUs on travis -> all tests will be skipped
-  - CUDA_VISIBLE_DEVICES=0 py.test --cov ignite --cov-append --cov-report term-missing -vvv tests/ -k "on_cuda"
-  # tests for distributed ops
-  # As no GPUs on travis -> all tests will be skipped
-  # 2 is the number of processes <-> number of available GPUs
-  - export WORLD_SIZE=2
-  - py.test --cov ignite --cov-append --cov-report term-missing --dist=each --tx $WORLD_SIZE*popen//python=python$TRAVIS_PYTHON_VERSION tests -m distributed -vvv
+  - sh tests/run_cpu_tests.sh
 
 after_success:
   - codecov

CONTRIBUTING.md

@@ -82,17 +82,19 @@ flake8 ignite/ tests/ examples/
 ```
 
 #### Run tests:
+
+To run a specific test, for example `test_terminate` from `test_engine.py`
 ```bash
-pytest tests/ -vvv
+pytest tests/ignite/engine/test_terminate -vvv -k test_terminate
 ```
-or tests with coverage (assuming installed `pytest-cov`):
+To run all tests with coverage (assuming installed `pytest-cov`):
 ```bash
-py.test --cov ignite --cov-report term-missing
+sh tests/run_cpu_tests.sh
 ```
 
 #### Send a PR
-If everything is OK, please send a Pull Request to https://github.com/pytorch/ignite
 
+If everything is OK, please send a Pull Request to https://github.com/pytorch/ignite
 
 If you are not familiar with creating a Pull Request, here are some guides:
 - http://stackoverflow.com/questions/14680711/how-to-do-a-github-pull-request

tests/ignite/conftest.py

@@ -37,7 +37,7 @@ def distributed_context_single_node_nccl(local_rank):
         "backend": "nccl",
         "world_size": int(os.environ["WORLD_SIZE"]),
         "rank": local_rank,
-        "init_method": "tcp://localhost:2222",
+        "init_method": "tcp://localhost:2223",
     }
 
     dist.init_process_group(**dist_info)
@@ -57,6 +57,7 @@ def distributed_context_single_node_nccl(local_rank):
 def distributed_context_single_node_gloo(local_rank):
 
     import os
+    from datetime import timedelta
 
     if "WORLD_SIZE" not in os.environ:
         os.environ["WORLD_SIZE"] = "1"
@@ -66,6 +67,7 @@ def distributed_context_single_node_gloo(local_rank):
         "world_size": int(os.environ["WORLD_SIZE"]),
         "rank": local_rank,
         "init_method": "tcp://localhost:2222",
+        "timeout": timedelta(seconds=60),
     }
 
     dist.init_process_group(**dist_info)

tests/run_cpu_tests.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+
+set -xeu
+
+CUDA_VISIBLE_DEVICES="" py.test --tx 4*popen//python=python$TRAVIS_PYTHON_VERSION --cov ignite --cov-report term-missing -vvv tests/
+
+export WORLD_SIZE=2
+
+CUDA_VISIBLE_DEVICES="" py.test --cov ignite --cov-append --cov-report term-missing --dist=each --tx $WORLD_SIZE*popen//python=python$TRAVIS_PYTHON_VERSION tests -m distributed -vvv

tests/run_gpu_tests.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+
+if [ -z "$1" ]; then
+    ws=1
+else
+    ws=$1
+fi
+
+set -xeu
+
+py.test -vvv tests/ -k 'on_cuda'
+
+export WORLD_SIZE=$ws
+
+py.test --dist=each --tx $WORLD_SIZE*popen//python=python3.7 tests -m distributed -vvv

tests/run_tpu_tests.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+set -xeu
+
+py.test --cov ignite --cov-append --cov-report term-missing tests/ -vvv -m tpu