Package AI Runtime code_source via a tgz artifact, not a sync overlay

acceptance/bundle/ai_runtime_task/empty_code_source/output.txt

@@ -1,4 +1,5 @@
 
 >>> [CLI] bundle deploy
-Error: code_source_path "./src" has no files to package (all excluded by .gitignore or sync.exclude, or the directory is empty)
+Building air_code_source_src...
+Error: artifact tgz: no files to pack under "[TEST_TMP_DIR]" (empty, gitignored, or no `include` match)
 

acceptance/bundle/ai_runtime_task/local_code_source/databricks.yml

@@ -1,20 +1,13 @@
 bundle:
   name: ai-runtime-test
 
-sync:
-  # *.log excluded; data/*.bin force-included despite .gitignore.
-  exclude:
-    - "**/*.log"
-  include:
-    - src/data/*.bin
-
 resources:
   jobs:
     train:
       name: "[${bundle.target}] AI Runtime training"
       tasks:
-        # Two AI Runtime tasks with distinct local code dirs: each is packaged into
-        # its own content-addressed tarball, both under the repo root's .air_snapshots.
+        # Two AI Runtime tasks with distinct local code dirs: each is turned into its
+        # own tgz artifact and uploaded through the standard artifact path.
         - task_key: train
           environment_key: default
           ai_runtime_task:

acceptance/bundle/ai_runtime_task/local_code_source/output.txt

@@ -1,129 +1,29 @@
 
-=== deploy packages and uploads the local code sources
+=== deploy packages the local code sources as tgz artifacts and uploads them
 
 >>> [CLI] bundle deploy
+Building air_code_source_src...
+Building air_code_source_src2...
+Uploading .databricks/air_code_source/air_code_source_src.tar.gz...
+Uploading .databricks/air_code_source/air_code_source_src2.tar.gz...
 Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files...
 Created jobs.train
-Files: 13 uploaded, 0 deleted
+Files: 9 uploaded, 0 deleted
 Resources: 1 created, 0 changed, 0 deleted, 0 unchanged
 
-=== each task's tarball holds only synced files (both under the repo root .air_snapshots)
+=== both code_source_paths rewritten to the uploaded artifact remote paths
 
->>> list_code_snapshot.py
-# .air_snapshots/[SNAPSHOT].tar.gz
-src2/command.sh
-src2/train.py
-# .air_snapshots/[SNAPSHOT].tar.gz
-src/.gitignore
-src/command.sh
-src/data/model.bin
-src/train.py
+>>> jq -s .[] | select(.path=="/api/2.2/jobs/create") | .body.tasks[].ai_runtime_task.code_source_path out.requests.txt
+"/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/artifacts/.internal/air_code_source_src.tar.gz"
+"/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/artifacts/.internal/air_code_source_src2.tar.gz"
 
-=== both code_source_paths point into the bundle .air_snapshots, command_paths rewritten, deps on environments spec
+=== both code tarballs uploaded
 
->>> print_requests.py --nostamp --sort --del-field raw_body //.air_snapshots/ //jobs/create
-{
-  "method": "POST",
-  "path": "/api/2.0/workspace-files/import-file/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/.air_snapshots/[SNAPSHOT].tar.gz",
-  "q": {
-    "overwrite": "true"
-  }
-}
-{
-  "method": "POST",
-  "path": "/api/2.0/workspace-files/import-file/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/.air_snapshots/[SNAPSHOT].tar.gz",
-  "q": {
-    "overwrite": "true"
-  }
-}
-{
-  "method": "POST",
-  "path": "/api/2.2/jobs/create",
-  "body": {
-    "deployment": {
-      "kind": "BUNDLE",
-      "metadata_file_path": "/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/state/metadata.json"
-    },
-    "edit_mode": "UI_LOCKED",
-    "environments": [
-      {
-        "environment_key": "default",
-        "spec": {
-          "dependencies": [
-            "torch>=2.0.0"
-          ],
-          "environment_version": "5"
-        }
-      }
-    ],
-    "format": "MULTI_TASK",
-    "max_concurrent_runs": 1,
-    "name": "[default] AI Runtime training",
-    "queue": {
-      "enabled": true
-    },
-    "tasks": [
-      {
-        "ai_runtime_task": {
-          "code_source_path": "/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/.air_snapshots/[SNAPSHOT].tar.gz",
-          "deployments": [
-            {
-              "command_path": "/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/src/command.sh",
-              "compute": {
-                "accelerator_count": 8,
-                "accelerator_type": "GPU_8xH100"
-              }
-            }
-          ],
-          "experiment": "my-training"
-        },
-        "environment_key": "default",
-        "task_key": "train"
-      },
-      {
-        "ai_runtime_task": {
-          "code_source_path": "/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/.air_snapshots/[SNAPSHOT].tar.gz",
-          "deployments": [
-            {
-              "command_path": "/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/src2/command.sh",
-              "compute": {
-                "accelerator_count": 8,
-                "accelerator_type": "GPU_8xH100"
-              }
-            }
-          ],
-          "experiment": "my-training-2"
-        },
-        "environment_key": "default",
-        "task_key": "train2"
-      }
-    ]
-  }
-}
+>>> jq -r .path
+/api/2.0/workspace-files/import-file/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/artifacts/.internal/air_code_source_src.tar.gz
+/api/2.0/workspace-files/import-file/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/artifacts/.internal/air_code_source_src2.tar.gz
 
-=== re-planning unchanged code is a no-op (no changes)
-
->>> [CLI] bundle plan
-Plan: 0 to add, 0 to change, 0 to delete, 1 unchanged
-
-=== editing a file changes the snapshot hash (content-addressed name changes)
-
->>> [CLI] bundle deploy
-Uploading bundle files to /Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files...
-Updated jobs.train
-Files: 4 uploaded, 1 deleted
-Resources: 0 created, 1 changed, 0 deleted, 0 unchanged
-
->>> print_requests.py --nostamp --sort --del-field raw_body //.air_snapshots/
-{
-  "method": "POST",
-  "path": "/api/2.0/workspace-files/import-file/Workspace/Users/[USERNAME]/.bundle/ai-runtime-test/default/files/.air_snapshots/[SNAPSHOT].tar.gz",
-  "q": {
-    "overwrite": "true"
-  }
-}
-
-=== destroy removes the deployed bundle (including the synced snapshots)
+=== destroy removes the deployed bundle
 
 >>> [CLI] bundle destroy --auto-approve
 The following resources will be deleted:

acceptance/bundle/ai_runtime_task/local_code_source/script

@@ -1,29 +1,18 @@
-# Two AI Runtime tasks each with a local directory code_source_path. Each is
-# packaged into a content-addressed tarball written into the bundle
-# (.air_snapshots/, at the repo root) and uploaded by normal bundle file sync;
-# code_source_path/command_path are rewritten. Pip deps ride on the job's
-# environments[].spec.dependencies (no requirements.yaml is synthesized).
+# Two AI Runtime tasks with distinct local directory code sources. Each directory is
+# turned into a tgz artifact (see bundle/config/mutator/aicode), built and uploaded
+# through the standard artifact path, and its code_source_path is rewritten to the
+# uploaded remote path. Pip deps ride on the job's environments[].spec.dependencies.
 
-title "deploy packages and uploads the local code sources\n"
+title "deploy packages the local code sources as tgz artifacts and uploads them\n"
 trace $CLI bundle deploy
 
-title "each task's tarball holds only synced files (both under the repo root .air_snapshots)\n"
-trace list_code_snapshot.py
+title "both code_source_paths rewritten to the uploaded artifact remote paths\n"
+# Filters use a leading // so Git Bash on Windows does not path-convert them.
+trace jq -s '.[] | select(.path=="/api/2.2/jobs/create") | .body.tasks[].ai_runtime_task.code_source_path' out.requests.txt
 
-title "both code_source_paths point into the bundle .air_snapshots, command_paths rewritten, deps on environments spec\n"
-# --del-field raw_body drops the binary tarball upload payload (kept readable). Filters
-# use a leading // so Git Bash on Windows does not path-convert them. --keep is not
-# passed, so print_requests.py --nostamp consumes out.requests.txt.
-trace print_requests.py --nostamp --sort --del-field raw_body '//.air_snapshots/' '//jobs/create'
+title "both code tarballs uploaded\n"
+trace jq -r .path < out.requests.txt | grep import | grep '.tar.gz' | sort
 
-title "re-planning unchanged code is a no-op (no changes)\n"
-trace $CLI bundle plan
-
-title "editing a file changes the snapshot hash (content-addressed name changes)\n"
-update_file.py src/train.py 'print("training")' 'print("training v2")'
-trace $CLI bundle deploy
-trace print_requests.py --nostamp --sort --del-field raw_body '//.air_snapshots/'
-
-title "destroy removes the deployed bundle (including the synced snapshots)\n"
+title "destroy removes the deployed bundle\n"
 trace $CLI bundle destroy --auto-approve
 rm out.requests.txt

acceptance/bundle/ai_runtime_task/local_code_source/test.toml

@@ -3,10 +3,3 @@ RecordRequests = true
 Ignore = [
     '.databricks',
 ]
-
-# The archive is content-addressed: <dir>_<sha256[:16]>.tar.gz. The hash is stable
-# given the committed inputs, but collapse it to a token so the test does not pin a
-# specific digest. Matches both src_ and src2_ archives.
-[[Repls]]
-Old = '(src2?)_[0-9a-f]{16}\.tar\.gz'
-New = '$1_[SNAPSHOT].tar.gz'

bundle/bundle.go

@@ -59,11 +59,6 @@ func (b *Bundle) SuppressProgress() bool {
 	return b.Quiet >= QuietAll
 }
 
-// AiCodeSnapshotDir is the sync-relative dir the aicode mutator writes AI Runtime
-// code snapshots into. Force-included in sync (see GetSyncIncludePatterns) so user
-// ignore rules can't filter the deployed job's code_source_path archives out.
-const AiCodeSnapshotDir = ".air_snapshots"
-
 // Filename where resources are stored for DATABRICKS_BUNDLE_ENGINE=direct
 const resourcesFilename = "resources.json"
 
@@ -224,12 +219,6 @@ type Bundle struct {
 	// comparison with remote state, but local file validation would incorrectly fail.
 	SkipLocalFileValidation bool
 
-	// HasAiRuntimeCodeSnapshot is set by the aicode.PackageCodeSource build-phase
-	// mutator when it packages a local AI Runtime code_source into the bundle's
-	// snapshot dir. GetSyncIncludePatterns reads it to force-sync that dir only for
-	// bundles that actually use the feature, rather than for every bundle.
-	HasAiRuntimeCodeSnapshot bool
-
 	// Tagging is used to normalize tag keys and values.
 	// The implementation depends on the cloud being targeted.
 	Tagging tags.Cloud
@@ -419,13 +408,6 @@ func (b *Bundle) GetSyncIncludePatterns(ctx context.Context) ([]string, error) {
 		return nil, err
 	}
 	includes := append(b.Config.Sync.Include, filepath.ToSlash(filepath.Join(internalDirRel, "*.*")))
-	// Force-sync generated AI Runtime code snapshots so a user ignore rule (e.g.
-	// "*.tar.gz" in .gitignore) can't filter them out — the deployed job's
-	// code_source_path points at these archives (see bundle/config/mutator/aicode).
-	// Scoped to bundles that actually package one, so it's not a global include.
-	if b.HasAiRuntimeCodeSnapshot {
-		includes = append(includes, AiCodeSnapshotDir+"/*")
-	}
 	return includes, nil
 }
 

bundle/bundle_test.go

@@ -71,29 +71,6 @@ func TestBundleLocalStateDir(t *testing.T) {
 	assert.Equal(t, filepath.Join(projectDir, ".databricks", "bundle", "default"), cacheDir)
 }
 
-func TestGetSyncIncludePatternsScopesSnapshotDir(t *testing.T) {
-	ctx := t.Context()
-	projectDir := t.TempDir()
-	f, err := os.Create(filepath.Join(projectDir, "databricks.yml"))
-	require.NoError(t, err)
-	f.Close()
-
-	b, err := Load(ctx, projectDir)
-	require.NoError(t, err)
-	b.Config.Bundle.Target = "default"
-
-	// Without an AI Runtime code snapshot, the dir is not force-included.
-	includes, err := b.GetSyncIncludePatterns(ctx)
-	require.NoError(t, err)
-	assert.NotContains(t, includes, AiCodeSnapshotDir+"/*")
-
-	// Once the aicode mutator sets the flag, it is.
-	b.HasAiRuntimeCodeSnapshot = true
-	includes, err = b.GetSyncIncludePatterns(ctx)
-	require.NoError(t, err)
-	assert.Contains(t, includes, AiCodeSnapshotDir+"/*")
-}
-
 func TestBundleLocalStateDirOverride(t *testing.T) {
 	ctx := t.Context()
 	projectDir := t.TempDir()