[YOLO] clean logic for image repos/tags used to run functional tests (#1429)

Author: datarobot-harness[bot]

harness_scripts/functional_by_framework/check_by_framework_run1_step_entrypoint.sh

@@ -25,6 +25,9 @@ set +e
 id
 set -e
 
+title "Upgrade pip"
+pip install -U pip
+
 title "Installing pytest"
 pip install pytest pytest-xdist
 
@@ -33,6 +36,7 @@ pip install pytest pytest-xdist
 # with the datarobot-drum version specified in the environment's requirements.txt file.
 if [ "${FRAMEWORK}" != "java_codegen" ]; then
     title "Uninstalling datarobot-drum"
+    # I think we don't need to uninstall datarobot-mlops, but removing it broke drum re-installation.
     pip uninstall datarobot-drum datarobot-mlops -y
 
     title "Installing dependencies, with datarobot-drum installed from source-code"
@@ -64,8 +68,13 @@ if [ "${FRAMEWORK}" != "java_codegen" ]; then
       DRUM_SOURCE_DIR_TMP="/tmp/custom_model_runner"
       cp -r ${DRUM_SOURCE_DIR} ${DRUM_SOURCE_DIR_TMP}
     fi
-    # Install datarobot-drum from source code, but keep dependencies that were installed by the environment
-    pip install --force-reinstall ${DRUM_SOURCE_DIR_TMP}${EXTRA} ${INST_ENV_REQ_CMD}
+    # Install datarobot-drum from source code.
+    # Testing image either was just built (if env changed), or the latest release image is used for testing.
+    # I think we should not reinstall all the deps.
+    # We only install DRUM from source, if some deps were changed they are upgraded.
+    # This saves time by avoiding heavy AI/nvidia packages reinstallation.
+    # pip install --force-reinstall ${DRUM_SOURCE_DIR_TMP}${EXTRA} ${INST_ENV_REQ_CMD}
+    pip install --upgrade ${DRUM_SOURCE_DIR_TMP}${EXTRA}
 fi
 
 cd "${ROOT_DIR}"

harness_scripts/functional_by_framework/check_env_changed_check_diff.sh

@@ -20,35 +20,25 @@ echo "--- changed paths ---"
 echo "${changed_paths}"
 echo "--- --- --- --- ---"
 
-# by default define namespace, repo, and tag for the existing flow without changes
-# e.g. datarobotdev/datarobot-user-models:public_dropin_envs_python3_sklearn_latest
+# by default set namespace, repo, and tag as
+# datarobot/<image_repository>:<env_version_id>
+# e.g. datarobot/env-python-sklearn:12355123abc918234
 env_info="${ENV_FOLDER}/${FRAMEWORK}/env_info.json"
 ENV_VERSION_ID=$(jq -r '.environmentVersionId' ${env_info})
 
+# once we implement image promotion, change it to datarobot.
 test_image_namespace=datarobotdev
-test_image_repository=datarobot-user-models
-
-test_image_tag_base=${ENV_FOLDER}_${FRAMEWORK}_${ENV_VERSION_ID}
-test_image_tag=${test_image_tag_base}
+test_image_tag=${ENV_VERSION_ID}
 
 changed_deps=false;
 
-
 IMAGE_REPOSITORY=$(jq -r '.imageRepository' ${env_info})
 if [ "${IMAGE_REPOSITORY}" = "null" ]; then
-  echo "Image repository is not defined in env_info.json"
+  echo "Image repository must be defined in 'imageRepository' in env_info.json"
 else
-  # if env_info has imageRepository
-  # point test_image_namespace to datarobot
-  # point test_image_repository to defined repo
-  # point tag to ENV_VERSION_ID
   # e.g. datarobot/env-python-sklearn:12355123abc918234
-  echo "read ${IMAGE_REPOSITORY}"
-  # after promotion work, change to datarobot
-  test_image_namespace=datarobotdev
+  echo "Image repo read from env_info.json: ${IMAGE_REPOSITORY}"
   test_image_repository=${IMAGE_REPOSITORY}
-  test_image_tag_base=${ENV_VERSION_ID}
-  test_image_tag=${ENV_VERSION_ID}
 fi
 
 
@@ -58,11 +48,11 @@ if echo "${changed_paths}" | grep "${ENV_FOLDER}/${FRAMEWORK}" > /dev/null; then
     test_image_namespace=datarobotdev
     if [ -n $TRIGGER_PR_NUMBER ] && [ "$TRIGGER_PR_NUMBER" != "null" ]; then
         # datarobotdev/env-python-sklearn:12355123abc918234_PR_NUM
-        # or
-        # datarobotdev/datarobot-user-models:public_dropin_envs_python3_sklearn_PR_NUM
-        test_image_tag=${test_image_tag_base}
+        # placeholder in case we want to add PR number back,
+        # but then it will be difficult to promote
+        test_image_tag=${test_image_tag}
     else
-        test_image_tag=${test_image_tag_base}_${CODEBASE_BRANCH}
+        test_image_tag=${test_image_tag}_${CODEBASE_BRANCH}
         # If the test_image_tag may contain a slash, replace it with an underscore (POSIX compliant)
         while case $test_image_tag in */*) true;; *) false;; esac; do
           test_image_tag=${test_image_tag%%/*}_${test_image_tag#*/}

tests/functional/test_fit_per_framework.py

@@ -1043,5 +1043,5 @@ def test_fit_with_fit_metadata(
             assert os.path.exists(output / FIT_METADATA_FILENAME)
             data = json.load(open(output / FIT_METADATA_FILENAME))
             assert "fit_memory_usage" in data.keys()
-            assert 250 > data["fit_memory_usage"] > 100
-            assert 250 > data["prediction_memory_usage"] > 100
+            assert 300 > data["fit_memory_usage"] > 100
+            assert 300 > data["prediction_memory_usage"] > 100