Run AIR CLI via DABs

acceptance/experimental/air/get-ai-runtime/output.txt

@@ -1,6 +1,6 @@
 
 === get (text)
->>> [CLI] experimental air get 123
+>>> [CLI] experimental air get run 123
 
 ╭─ Configuration ────────────────────────────────────────────────╮
 │                                                                │
@@ -35,7 +35,7 @@ Run URL:    [DATABRICKS_URL]/jobs/runs/123?o=[NUMID]
 MLflow URL: [DATABRICKS_URL]/ml/experiments/exp1/runs/run1
 
 === get (json)
->>> [CLI] experimental air get 123 -o json
+>>> [CLI] experimental air get run 123 -o json
 {
   "v": 1,
   "ts": "[TIMESTAMP]",

acceptance/experimental/air/get-ai-runtime/script

@@ -5,7 +5,7 @@ $CLI workspace mkdirs "/Workspace/Users/user@example.com/.air/cli_launch/my-exp/
 $CLI workspace import "/Workspace/Users/user@example.com/.air/cli_launch/my-exp/my-exp_abc/training_config.yaml" --file training_config.yaml --format AUTO &> LOG.import
 
 title "get (text)"
-trace $CLI experimental air get 123
+trace $CLI experimental air get run 123
 
 title "get (json)"
-trace $CLI experimental air get 123 -o json
+trace $CLI experimental air get run 123 -o json

acceptance/experimental/air/get/output.txt

@@ -1,6 +1,6 @@
 
 === get (text)
->>> [CLI] experimental air get 123
+>>> [CLI] experimental air get run 123
 
 ╭─ Configuration ────────────────────────────────────────────────╮
 │                                                                │
@@ -35,7 +35,7 @@ Run URL:    [DATABRICKS_URL]/jobs/runs/123?o=[NUMID]
 MLflow URL: [DATABRICKS_URL]/ml/experiments/exp1/runs/run1
 
 === get (json)
->>> [CLI] experimental air get 123 -o json
+>>> [CLI] experimental air get run 123 -o json
 {
   "v": 1,
   "ts": "[TIMESTAMP]",
@@ -52,13 +52,13 @@ MLflow URL: [DATABRICKS_URL]/ml/experiments/exp1/runs/run1
 }
 
 === invalid run id
->>> [CLI] experimental air get notanumber
+>>> [CLI] experimental air get run notanumber
 Error: invalid JOB_RUN_ID "notanumber": must be a positive integer
 
 Exit code: 1
 
 === invalid run id (json)
->>> [CLI] experimental air get notanumber -o json
+>>> [CLI] experimental air get run notanumber -o json
 {
   "v": 1,
   "ts": "[TIMESTAMP]",

acceptance/experimental/air/get/script

@@ -1,11 +1,11 @@
 title "get (text)"
-trace $CLI experimental air get 123
+trace $CLI experimental air get run 123
 
 title "get (json)"
-trace $CLI experimental air get 123 -o json
+trace $CLI experimental air get run 123 -o json
 
 title "invalid run id"
-errcode trace $CLI experimental air get notanumber
+errcode trace $CLI experimental air get run notanumber
 
 title "invalid run id (json)"
-errcode trace $CLI experimental air get notanumber -o json
+errcode trace $CLI experimental air get run notanumber -o json

acceptance/experimental/air/help/output.txt

@@ -11,8 +11,9 @@ Usage:
 
 Available Commands:
   cancel         Cancel one or more runs
-  get            Show status, configuration, and timing details for a specific run
-  list           List your active runs for the current profile (use --all-status for finished runs)
+  delete         Delete a resource (bundle)
+  get            Show details for a specific resource (run or bundle)
+  list           List a collection of resources (runs or bundles)
   logs           Stream or fetch logs for a run
   register-image Mirror a Docker image into the workspace registry
   run            Submit a training workload from a YAML config
@@ -29,17 +30,17 @@ Global Flags:
 Use "databricks experimental air [command] --help" for more information about a command.
 
 === list help
->>> [CLI] experimental air list --help
+>>> [CLI] experimental air list runs --help
 List your active runs for the current profile (use --all-status for finished runs)
 
 Usage:
-  databricks experimental air list [flags]
+  databricks experimental air list runs [flags]
 
 Flags:
       --all-status           Show runs in all states (default: active only)
       --all-users            Show runs from all users
       --filter stringArray   Filter runs, e.g. experiment=foo* (repeatable)
-  -h, --help                 help for list
+  -h, --help                 help for runs
       --limit int            Maximum number of runs to show (default 20)
 
 Global Flags:

acceptance/experimental/air/help/script

@@ -5,4 +5,4 @@ title "help"
 trace $CLI experimental air --help
 
 title "list help"
-trace $CLI experimental air list --help
+trace $CLI experimental air list runs --help

acceptance/experimental/air/list/output.txt

@@ -1,11 +1,11 @@
 
 === list (text)
->>> [CLI] experimental air list
-   Run ID           Experiment  Status     Started              Duration  MLflow              User                   Accelerators
-   [NUMID]  qwen-train  ● SUCCESS  [TIMESTAMP]       12s  …/runs/run1         [USERNAME]  8x H100
+>>> [CLI] experimental air list runs
+   Run ID           Experiment  Status     Started                    Duration  MLflow              User                   Accelerators
+   [NUMID]  qwen-train  ● RUNNING  [TIMESTAMP]  [DURATION]  …/runs/run1         [USERNAME]  8x H100
 
 === list (json)
->>> [CLI] experimental air list -o json
+>>> [CLI] experimental air list runs -o json
 {
   "v": 1,
   "ts": "[TIMESTAMP]",
@@ -15,21 +15,22 @@
         "run_id": "[NUMID]",
         "run_name": "qwen-train",
         "user": "[USERNAME]",
-        "status": "SUCCESS",
+        "status": "RUNNING",
         "started_at": "[TIMESTAMP]",
         "is_sweep": false
       }
     ]
   }
 }
 
-=== list --all-status (text, via AiTrainingService index)
->>> [CLI] experimental air list --all-status
-   Run ID           Experiment  Status     Started              Duration  MLflow              User                   Accelerators
-   [NUMID]  qwen-train  ● SUCCESS  [TIMESTAMP]       12s  …/runs/run1         [USERNAME]  8x H100
+=== list --all-status (text)
+>>> [CLI] experimental air list runs --all-status
+   Run ID           Experiment   Status     Started                    Duration  MLflow              User                   Accelerators
+   [NUMID]  qwen-train   ● RUNNING  [TIMESTAMP]  [DURATION]  …/runs/run1         [USERNAME]  8x H100
+   [NUMID]  llama-train  ● SUCCESS  [TIMESTAMP]          1m 11s  …/runs/run1         [USERNAME]  1x A10
 
 === list --all-status (json)
->>> [CLI] experimental air list --all-status -o json
+>>> [CLI] experimental air list runs --all-status -o json
 {
   "v": 1,
   "ts": "[TIMESTAMP]",
@@ -39,6 +40,14 @@
         "run_id": "[NUMID]",
         "run_name": "qwen-train",
         "user": "[USERNAME]",
+        "status": "RUNNING",
+        "started_at": "[TIMESTAMP]",
+        "is_sweep": false
+      },
+      {
+        "run_id": "[NUMID]",
+        "run_name": "llama-train",
+        "user": "[USERNAME]",
         "status": "SUCCESS",
         "started_at": "[TIMESTAMP]",
         "is_sweep": false

acceptance/experimental/air/list/script

@@ -1,11 +1,11 @@
 title "list (text)"
-trace $CLI experimental air list
+trace $CLI experimental air list runs
 
 title "list (json)"
-trace $CLI experimental air list -o json
+trace $CLI experimental air list runs -o json
 
-title "list --all-status (text, via AiTrainingService index)"
-trace $CLI experimental air list --all-status
+title "list --all-status (text)"
+trace $CLI experimental air list runs --all-status
 
 title "list --all-status (json)"
-trace $CLI experimental air list --all-status -o json
+trace $CLI experimental air list runs --all-status -o json

acceptance/experimental/air/list/test.toml

@@ -2,21 +2,26 @@
 [EnvMatrix]
 DATABRICKS_BUNDLE_ENGINE = []
 
-# Disable the on-disk run cache so --all-status output is deterministic across runs.
-[Env]
-DATABRICKS_CACHE_ENABLED = "false"
+# A running run's duration is measured against now (no end_time), so mask the
+# hours-minutes-seconds cell to keep the golden deterministic.
+[[Repls]]
+Old = '\d+h \d+m \d+s'
+New = "[DURATION]"
+Order = 20
 
 # The SDK occasionally probes host reachability with a HEAD request; stub it so
 # the test is deterministic.
 [[Server]]
 Pattern = "HEAD /"
 Response.Body = ''
 
-# `air list` reads Jobs runs/list directly (expand_tasks), filtering to AIR runs.
-# The current runs carry an ai_runtime_task (not modeled by the typed SDK), which
-# is why the CLI parses the response raw. Both runs belong to the default user
-# (tester@databricks.com, from the built-in scim/v2/Me handler). The second run
-# has no AI task and must be filtered out.
+# `air list` reads Jobs runs/list (expand_tasks), filtering to AIR runs. It first
+# tries the server-side AI-Runtime filter (filter=runType=AI_RUNTIME); this fake
+# ignores query params and serves the same body, exercising the parse + client-side
+# filtering (AIR task-shape, active-only). All runs belong to the default user
+# (tester@databricks.com, from the built-in scim/v2/Me handler). qwen-train is
+# RUNNING so it shows in the active default; the terminated run shows only under
+# --all-status; the notebook run has no AI task and is filtered out.
 [[Server]]
 Pattern = "GET /api/2.2/jobs/runs/list"
 Response.Body = '''
@@ -27,8 +32,7 @@ Response.Body = '''
       "run_name": "qwen-train",
       "creator_user_name": "tester@databricks.com",
       "start_time": 1717608759000,
-      "end_time": 1717608771000,
-      "state": {"life_cycle_state": "TERMINATED", "result_state": "SUCCESS"},
+      "state": {"life_cycle_state": "RUNNING"},
       "tasks": [{
         "run_id": 334747067049497,
         "ai_runtime_task": {
@@ -37,6 +41,21 @@ Response.Body = '''
         }
       }]
     },
+    {
+      "run_id": 334747067049400,
+      "run_name": "llama-train",
+      "creator_user_name": "tester@databricks.com",
+      "start_time": 1717608700000,
+      "end_time": 1717608771000,
+      "state": {"life_cycle_state": "TERMINATED", "result_state": "SUCCESS"},
+      "tasks": [{
+        "run_id": 334747067049401,
+        "ai_runtime_task": {
+          "experiment": "/Users/tester@databricks.com/llama-train",
+          "deployments": [{"compute": {"accelerator_type": "GPU_1xA10", "accelerator_count": 1}}]
+        }
+      }]
+    },
     {
       "run_id": 999000999000,
       "run_name": "not-an-air-run",
@@ -55,33 +74,3 @@ Pattern = "GET /api/2.2/jobs/runs/get-output"
 Response.Body = '''
 {"ai_runtime_task_output": {"mlflow_experiment_id": "exp1", "mlflow_run_id": "run1"}}
 '''
-
-# `air list --all-status` scoped to the current user is served by the
-# AiTrainingService index: it returns cheap (job_run_id, submit_time) pairs, which
-# the CLI orders and then hydrates into full runs via runs/get.
-[[Server]]
-Pattern = "GET /api/2.0/ai-training/workflows"
-Response.Body = '''
-{"training_workflows": [{"job_run_id": "334747067049496", "submit_time": "2024-06-05T17:32:39Z"}]}
-'''
-
-# runs/get hydrates one index id into the same shape as a runs/list element.
-[[Server]]
-Pattern = "GET /api/2.2/jobs/runs/get"
-Response.Body = '''
-{
-  "run_id": 334747067049496,
-  "run_name": "qwen-train",
-  "creator_user_name": "tester@databricks.com",
-  "start_time": 1717608759000,
-  "end_time": 1717608771000,
-  "state": {"life_cycle_state": "TERMINATED", "result_state": "SUCCESS"},
-  "tasks": [{
-    "run_id": 334747067049497,
-    "ai_runtime_task": {
-      "experiment": "/Users/tester@databricks.com/qwen-train",
-      "deployments": [{"compute": {"accelerator_type": "GPU_8xH100", "accelerator_count": 8}}]
-    }
-  }]
-}
-'''