Added telemetry for artifact deployment properties (#2679)

andrewnester · web-flow · commit aa758d2cd54d · 2025-04-09T15:11:08.000+02:00
## Changes
Added telemetry for artifact deployment properties

## Why
This allows us to get insights into usage of 
- Wheel patching
- Python trampoline for an older DBRs
- Build command inferring vs. explicit command
- How often build files for artifacts are specified

## Tests
Added acceptance test

&lt;!-- If your PR needs to be included in the release notes for next
release,
add a separate entry in NEXT_CHANGELOG.md as part of your PR. --&gt;
diff --git a/acceptance/bundle/telemetry/deploy-artifact-path-type/output.txt b/acceptance/bundle/telemetry/deploy-artifact-path-type/output.txt
@@ -32,6 +32,12 @@ Deployment complete!
     "complex_variable_count": 0,
     "lookup_variable_count": 0,
     "target_count": 2,
+    "bool_values": [
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": false
+      }
+    ],
     "bundle_mode": "TYPE_UNSPECIFIED",
     "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
   }
@@ -58,6 +64,12 @@ Deployment complete!
     "complex_variable_count": 0,
     "lookup_variable_count": 0,
     "target_count": 2,
+    "bool_values": [
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": false
+      }
+    ],
     "bundle_mode": "TYPE_UNSPECIFIED",
     "workspace_artifact_path_type": "UC_VOLUME"
   }
diff --git a/acceptance/bundle/telemetry/deploy-artifact-path-type/script b/acceptance/bundle/telemetry/deploy-artifact-path-type/script
@@ -2,6 +2,6 @@ trace $CLI bundle deploy -t one
 
 trace $CLI bundle deploy -t two
 
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
 
 rm out.requests.txt
diff --git a/acceptance/bundle/telemetry/deploy-config-file-count/output.txt b/acceptance/bundle/telemetry/deploy-config-file-count/output.txt
@@ -39,6 +39,12 @@ Deployment complete!
           "complex_variable_count": 0,
           "lookup_variable_count": 0,
           "target_count": 1,
+          "bool_values": [
+            {
+              "key": "python_wheel_wrapper_is_set",
+              "value": false
+            }
+          ],
           "bundle_mode": "TYPE_UNSPECIFIED",
           "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
         }
diff --git a/acceptance/bundle/telemetry/deploy-config-file-count/script b/acceptance/bundle/telemetry/deploy-config-file-count/script
@@ -1,6 +1,6 @@
 trace $CLI bundle deploy
 
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
 
 cmd_exec_id=$(extract_command_exec_id.py)
 
diff --git a/acceptance/bundle/telemetry/deploy-mode/output.txt b/acceptance/bundle/telemetry/deploy-mode/output.txt
@@ -38,6 +38,12 @@ A common practice is to use a username or principal name in this path, i.e. use
     "complex_variable_count": 0,
     "lookup_variable_count": 0,
     "target_count": 2,
+    "bool_values": [
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": false
+      }
+    ],
     "bundle_mode": "DEVELOPMENT",
     "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
   }
@@ -64,6 +70,12 @@ A common practice is to use a username or principal name in this path, i.e. use
     "complex_variable_count": 0,
     "lookup_variable_count": 0,
     "target_count": 2,
+    "bool_values": [
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": false
+      }
+    ],
     "bundle_mode": "PRODUCTION",
     "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
   }
diff --git a/acceptance/bundle/telemetry/deploy-mode/script b/acceptance/bundle/telemetry/deploy-mode/script
@@ -2,6 +2,6 @@ trace $CLI bundle deploy -t dev
 
 trace $CLI bundle deploy -t prod
 
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
 
 rm out.requests.txt
diff --git a/acceptance/bundle/telemetry/deploy-no-uuid/output.txt b/acceptance/bundle/telemetry/deploy-no-uuid/output.txt
@@ -43,6 +43,12 @@ Deployment complete!
           "complex_variable_count": 0,
           "lookup_variable_count": 0,
           "target_count": 1,
+          "bool_values": [
+            {
+              "key": "python_wheel_wrapper_is_set",
+              "value": false
+            }
+          ],
           "bundle_mode": "TYPE_UNSPECIFIED",
           "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
         }
diff --git a/acceptance/bundle/telemetry/deploy-no-uuid/script b/acceptance/bundle/telemetry/deploy-no-uuid/script
@@ -1,6 +1,6 @@
 trace $CLI bundle deploy
 
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
 
 cmd_exec_id=$(extract_command_exec_id.py)
 
diff --git a/acceptance/bundle/telemetry/deploy-target-count/output.txt b/acceptance/bundle/telemetry/deploy-target-count/output.txt
@@ -27,6 +27,12 @@ Deployment complete!
     "complex_variable_count": 0,
     "lookup_variable_count": 0,
     "target_count": 3,
+    "bool_values": [
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": false
+      }
+    ],
     "bundle_mode": "TYPE_UNSPECIFIED",
     "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
   }
diff --git a/acceptance/bundle/telemetry/deploy-target-count/script b/acceptance/bundle/telemetry/deploy-target-count/script
@@ -1,5 +1,5 @@
 trace $CLI bundle deploy -t one
 
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
 
 rm out.requests.txt
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/.gitignore b/acceptance/bundle/telemetry/deploy-whl-artifacts/.gitignore
@@ -0,0 +1,3 @@
+build/
+*.egg-info
+.databricks
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/databricks.yml b/acceptance/bundle/telemetry/deploy-whl-artifacts/databricks.yml
@@ -0,0 +1,28 @@
+bundle:
+  name: deploy-artifacts
+
+variables:
+  wrapper:
+    default: false
+
+experimental:
+  python_wheel_wrapper: ${var.wrapper}
+
+targets:
+  one:
+    artifacts:
+      test:
+        type: whl
+        path: ./my_test_code
+
+  two:
+    variables:
+      wrapper: true
+    artifacts:
+      test:
+        type: whl
+        path: ./my_test_code
+        dynamic_version: true
+        build: "python3 setup.py bdist_wheel"
+        files:
+          - source: "./my_test_code/dist/*.whl"
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/my_test_code/setup.py b/acceptance/bundle/telemetry/deploy-whl-artifacts/my_test_code/setup.py
@@ -0,0 +1,15 @@
+from setuptools import setup, find_packages
+
+import src
+
+setup(
+    name="my_test_code",
+    version=src.__version__,
+    author=src.__author__,
+    url="https://databricks.com",
+    author_email="john.doe@databricks.com",
+    description="my test wheel",
+    packages=find_packages(include=["src"]),
+    entry_points={"group_1": "run=src.__main__:main"},
+    install_requires=["setuptools"],
+)
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/my_test_code/src/__init__.py b/acceptance/bundle/telemetry/deploy-whl-artifacts/my_test_code/src/__init__.py
@@ -0,0 +1,2 @@
+__version__ = "0.0.1"
+__author__ = "Databricks"
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/my_test_code/src/__main__.py b/acceptance/bundle/telemetry/deploy-whl-artifacts/my_test_code/src/__main__.py
@@ -0,0 +1,16 @@
+"""
+The entry point of the Python Wheel
+"""
+
+import sys
+
+
+def main():
+    # This method will print the provided arguments
+    print("Hello from my func")
+    print("Got arguments:")
+    print(sys.argv)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/output.txt b/acceptance/bundle/telemetry/deploy-whl-artifacts/output.txt
@@ -0,0 +1,100 @@
+
+>>> [CLI] bundle deploy -t one
+Building test...
+Uploading my_test_code/dist/my_test_code-0.0.1-py3-none-any.whl...
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/deploy-artifacts/one/files...
+Deploying resources...
+Deployment complete!
+
+>>> [CLI] bundle deploy -t two
+Building test...
+Uploading .databricks/bundle/two/patched_wheels/test_my_test_code/my_test_code-0.0.1+[TIMESTAMP_NS]-py3-none-any.whl...
+Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/deploy-artifacts/two/files...
+Deploying resources...
+Deployment complete!
+
+>>> cat out.requests.txt
+{
+  "bundle_uuid": "[UUID]",
+  "deployment_id": "[UUID]",
+  "resource_count": 0,
+  "resource_job_count": 0,
+  "resource_pipeline_count": 0,
+  "resource_model_count": 0,
+  "resource_experiment_count": 0,
+  "resource_model_serving_endpoint_count": 0,
+  "resource_registered_model_count": 0,
+  "resource_quality_monitor_count": 0,
+  "resource_schema_count": 0,
+  "resource_volume_count": 0,
+  "resource_cluster_count": 0,
+  "resource_dashboard_count": 0,
+  "resource_app_count": 0,
+  "experimental": {
+    "configuration_file_count": 1,
+    "variable_count": 0,
+    "complex_variable_count": 0,
+    "lookup_variable_count": 0,
+    "target_count": 2,
+    "bool_values": [
+      {
+        "key": "artifact_build_command_is_set",
+        "value": false
+      },
+      {
+        "key": "artifact_files_is_set",
+        "value": false
+      },
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": false
+      }
+    ],
+    "bundle_mode": "TYPE_UNSPECIFIED",
+    "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
+  }
+}
+{
+  "bundle_uuid": "[UUID]",
+  "deployment_id": "[UUID]",
+  "resource_count": 0,
+  "resource_job_count": 0,
+  "resource_pipeline_count": 0,
+  "resource_model_count": 0,
+  "resource_experiment_count": 0,
+  "resource_model_serving_endpoint_count": 0,
+  "resource_registered_model_count": 0,
+  "resource_quality_monitor_count": 0,
+  "resource_schema_count": 0,
+  "resource_volume_count": 0,
+  "resource_cluster_count": 0,
+  "resource_dashboard_count": 0,
+  "resource_app_count": 0,
+  "experimental": {
+    "configuration_file_count": 1,
+    "variable_count": 0,
+    "complex_variable_count": 0,
+    "lookup_variable_count": 0,
+    "target_count": 2,
+    "bool_values": [
+      {
+        "key": "artifact_build_command_is_set",
+        "value": true
+      },
+      {
+        "key": "artifact_files_is_set",
+        "value": true
+      },
+      {
+        "key": "artifact_dynamic_version_is_set",
+        "value": true
+      },
+      {
+        "key": "python_wheel_wrapper_is_set",
+        "value": true
+      }
+    ],
+    "bundle_mode": "TYPE_UNSPECIFIED",
+    "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
+  }
+}
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/script b/acceptance/bundle/telemetry/deploy-whl-artifacts/script
@@ -0,0 +1,7 @@
+trace $CLI bundle deploy -t one
+
+trace $CLI bundle deploy -t two
+
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson | .entry.databricks_cli_log.bundle_deploy_event'
+
+rm out.requests.txt
diff --git a/acceptance/bundle/telemetry/deploy-whl-artifacts/test.toml b/acceptance/bundle/telemetry/deploy-whl-artifacts/test.toml
@@ -0,0 +1,8 @@
+Ignore = [
+    '.venv',
+    'dist',
+    'build',
+    '*egg-info',
+    '.databricks',
+    "__pycache__",
+]
diff --git a/acceptance/bundle/telemetry/deploy/output.txt b/acceptance/bundle/telemetry/deploy/output.txt
@@ -49,6 +49,12 @@ Deployment complete!
           "complex_variable_count": 0,
           "lookup_variable_count": 0,
           "target_count": 1,
+          "bool_values": [
+            {
+              "key": "python_wheel_wrapper_is_set",
+              "value": false
+            }
+          ],
           "bundle_mode": "TYPE_UNSPECIFIED",
           "workspace_artifact_path_type": "WORKSPACE_FILE_SYSTEM"
         }
diff --git a/acceptance/bundle/telemetry/deploy/script b/acceptance/bundle/telemetry/deploy/script
@@ -1,6 +1,6 @@
 trace $CLI bundle deploy
 
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
 
 cmd_exec_id=$(extract_command_exec_id.py)
 deployment_id=$(cat .databricks/bundle/default/deployment.json | jq -r .id)
diff --git a/acceptance/bundle/templates/telemetry/custom-template/script b/acceptance/bundle/templates/telemetry/custom-template/script
@@ -13,4 +13,4 @@ update_file.py out.requests.txt $bundle_uuid '[BUNDLE-UUID]'
 update_file.py out.databricks.yml $bundle_uuid '[BUNDLE-UUID]'
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/bundle/templates/telemetry/dbt-sql/script b/acceptance/bundle/templates/telemetry/dbt-sql/script
@@ -13,4 +13,4 @@ update_file.py out.requests.txt $bundle_uuid '[BUNDLE-UUID]'
 update_file.py out.databricks.yml $bundle_uuid '[BUNDLE-UUID]'
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/bundle/templates/telemetry/default-python/script b/acceptance/bundle/templates/telemetry/default-python/script
@@ -13,4 +13,4 @@ update_file.py out.requests.txt $bundle_uuid '[BUNDLE-UUID]'
 update_file.py out.databricks.yml $bundle_uuid '[BUNDLE-UUID]'
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/bundle/templates/telemetry/default-sql/script b/acceptance/bundle/templates/telemetry/default-sql/script
@@ -13,4 +13,4 @@ update_file.py out.requests.txt $bundle_uuid '[BUNDLE-UUID]'
 update_file.py out.databricks.yml $bundle_uuid '[BUNDLE-UUID]'
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/telemetry/failure/script b/acceptance/telemetry/failure/script
@@ -1,4 +1,4 @@
 trace $CLI selftest send-telemetry --debug
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/telemetry/partial-success/script b/acceptance/telemetry/partial-success/script
@@ -1,4 +1,4 @@
 trace $CLI selftest send-telemetry --debug
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/telemetry/success/script b/acceptance/telemetry/success/script
@@ -5,4 +5,4 @@ update_file.py out.requests.txt \
  '[CMD-EXEC-ID]'
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/acceptance/telemetry/timeout/script b/acceptance/telemetry/timeout/script
@@ -4,4 +4,4 @@ export DATABRICKS_CLI_TELEMETRY_TIMEOUT=0.1
 trace $CLI selftest send-telemetry --debug
 
 # pretty print the telemetry payload.
-trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs.[] | fromjson'
+trace cat out.requests.txt | jq 'select(has("path") and .path == "/telemetry-ext") | .body.protoLogs[] | fromjson'
diff --git a/bundle/artifacts/build.go b/bundle/artifacts/build.go
diff --git a/bundle/artifacts/prepare.go b/bundle/artifacts/prepare.go
diff --git a/bundle/bundle.go b/bundle/bundle.go
diff --git a/bundle/metrics/metrics.go b/bundle/metrics/metrics.go
diff --git a/bundle/phases/deploy.go b/bundle/phases/deploy.go
diff --git a/bundle/trampoline/python_wheel.go b/bundle/trampoline/python_wheel.go
diff --git a/libs/telemetry/protos/bundle_deploy.go b/libs/telemetry/protos/bundle_deploy.go
