terradev-cli 5.6.0__tar.gz → 5.6.2__tar.gz

{terradev_cli-5.6.0 → terradev_cli-5.6.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: terradev-cli
-Version: 5.6.0
+Version: 5.6.2
 Summary: An imperative command-line-interface for AI workload orchestration.
 Author-email: Terradev Team <team@terradev.com>
 Maintainer-email: Terradev Team <team@terradev.com>
@@ -100,7 +100,7 @@ Requires-Dist: redis>=5.0.0; extra == "enterprise"
 Requires-Dist: PyJWT>=2.8.0; extra == "enterprise"
 Dynamic: license-file
 
-# Terradev CLI v5.6.0
+# Terradev CLI v5.6.2
 
 **An imperative command-line-interface for AI workload orchestration.**
 
@@ -118,6 +118,18 @@ Continued focus on lower cost, faster provisioning, and topology-aware execution
 
 Model agnostic. Dataset agnostic. GPU agnostic. Provider agnostic. The only thing Terradev is not agnostic about is correctness: it enforces topology, idempotency, and sequencing.
 
+**NOTES ON 5.6.1**
+
+Added **Database Connection System** and **Unified Observability & Scheduling** commands:
+
+- **Database Connection System** (`core/database_connection.py`): SQLite and PostgreSQL support with in-memory connection storage, connection ID system, auto-table creation for 4 standard metadata tables (dataset_versions, workflow_runs, idempotency_keys, node_executions), and database operations (query, upsert, get connection).
+
+- **Observe Command** (`terradev observe`): Unified monitoring pipeline that wires API Gateway traffic into W&B, Phoenix, and Cost Analytics with a shared trace ID across all three destinations. Commands: `terradev observe gateway`, `terradev observe status`.
+
+- **Schedule Command** (`terradev schedule`): Spot-aware scheduling for cost-optimized job execution with pricing window awareness (not just generic cron). Commands: `terradev schedule job`, `terradev schedule list`, `terradev schedule windows`.
+
+- **MCP Tools**: 5 new database tools for connection management and queries (create_sqlite_connection, create_postgresql_connection, query_database, upsert_database, get_database_connection).
+
 **NOTES ON 5.6.0**
 
 Added **API Gateway for inference serving** with OpenAI/Anthropic/custom API entry and exit points:

{terradev_cli-5.6.0 → terradev_cli-5.6.2}/README.md

@@ -1,4 +1,4 @@
-# Terradev CLI v5.6.0
+# Terradev CLI v5.6.2
 
 **An imperative command-line-interface for AI workload orchestration.**
 
@@ -16,6 +16,18 @@ Continued focus on lower cost, faster provisioning, and topology-aware execution
 
 Model agnostic. Dataset agnostic. GPU agnostic. Provider agnostic. The only thing Terradev is not agnostic about is correctness: it enforces topology, idempotency, and sequencing.
 
+**NOTES ON 5.6.1**
+
+Added **Database Connection System** and **Unified Observability & Scheduling** commands:
+
+- **Database Connection System** (`core/database_connection.py`): SQLite and PostgreSQL support with in-memory connection storage, connection ID system, auto-table creation for 4 standard metadata tables (dataset_versions, workflow_runs, idempotency_keys, node_executions), and database operations (query, upsert, get connection).
+
+- **Observe Command** (`terradev observe`): Unified monitoring pipeline that wires API Gateway traffic into W&B, Phoenix, and Cost Analytics with a shared trace ID across all three destinations. Commands: `terradev observe gateway`, `terradev observe status`.
+
+- **Schedule Command** (`terradev schedule`): Spot-aware scheduling for cost-optimized job execution with pricing window awareness (not just generic cron). Commands: `terradev schedule job`, `terradev schedule list`, `terradev schedule windows`.
+
+- **MCP Tools**: 5 new database tools for connection management and queries (create_sqlite_connection, create_postgresql_connection, query_database, upsert_database, get_database_connection).
+
 **NOTES ON 5.6.0**
 
 Added **API Gateway for inference serving** with OpenAI/Anthropic/custom API entry and exit points:

{terradev_cli-5.6.0 → terradev_cli-5.6.2}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "terradev-cli"
-version = "5.6.0"
+version = "5.6.2"
 description = "An imperative command-line-interface for AI workload orchestration."
 readme = "README.md"
 license = {text = "Apache-2.0"}

{terradev_cli-5.6.0 → terradev_cli-5.6.2}/terradev_cli/cli.py

@@ -4519,7 +4519,13 @@ def job(job_file, optimize):
         print(f"ERROR: Error loading job file: {e}")
 
 
-@cli.command()
+@cli.group()
+def infer():
+    """Deploy and manage inference endpoints"""
+    pass
+
+
+@infer.command("deploy")
 @click.option("--model", "-m", required=True, help="Model name or path")
 @click.option(
     "--type", "-t", type=click.Choice(["llm", "embedding", "vision"]), help="Model type"
@@ -4534,7 +4540,7 @@ def job(job_file, optimize):
 @click.option("--region", "-r", help="Region preference")
 @click.option("--max-latency", type=float, help="Max latency in ms")
 @click.option("--max-cost", type=float, help="Max cost per request")
-def infer(model, type, provider, gpu_type, region, max_latency, max_cost):
+def infer_deploy_main(model, type, provider, gpu_type, region, max_latency, max_cost):
     """Deploy and manage inference endpoints"""
     print(f"Deploying inference for model: {model}")
 
@@ -4688,7 +4694,7 @@ def infer(model, type, provider, gpu_type, region, max_latency, max_cost):
         pass
 
 
-@cli.command()
+@infer.command("endpoint")
 @click.argument("model_path")
 @click.option("--name", "-n", required=True, help="Endpoint name (required)")
 @click.option(
@@ -4703,7 +4709,7 @@ def infer(model, type, provider, gpu_type, region, max_latency, max_cost):
 @click.option("--idle-timeout", type=int, default=300, help="Idle timeout in seconds")
 @click.option("--cost-optimize", is_flag=True, help="Enable cost optimization")
 @click.option("--dry-run", is_flag=True, help="Show deployment plan without deploying")
-def infer_deploy(
+def infer_endpoint(
     model_path,
     name,
     provider,
@@ -5217,7 +5223,7 @@ def run(gpu, image, cmd, mount, port, env, max_price, providers, keep_alive, dry
 # ═══════════════════════════════════════════════════════════════════════════════
 
 
-@cli.command("infer-status")
+@infer.command("status")
 @click.option(
     "--check", is_flag=True, help="Run live health probes before showing status"
 )
@@ -5288,7 +5294,7 @@ def infer_status(check):
             pass
 
 
-@cli.command("infer-failover")
+@infer.command("failover")
 @click.option(
     "--dry-run", is_flag=True, help="Show what would happen without executing failover"
 )
@@ -5363,7 +5369,7 @@ def infer_failover(dry_run):
         print("   OK: All primary endpoints healthy  no failover needed.")
 
 
-@cli.command("infer-route")
+@infer.command("route")
 @click.option("--model", "-m", help="Filter by model name")
 @click.option(
     "--strategy",
@@ -8276,7 +8282,13 @@ def hf_space(space_name, model_id, hardware, sdk, private, template, env, secret
 
 
 # Model Orchestrator Commands
-@cli.command()
+@cli.group()
+def orchestrator():
+    """Model orchestrator for multi-model inference"""
+    pass
+
+
+@orchestrator.command("start")
 @click.option("--gpu-id", default=0, help="GPU ID to use for orchestration")
 @click.option("--memory-gb", default=80.0, help="Total GPU memory in GB")
 @click.option(
@@ -8324,7 +8336,7 @@ def orchestrator_start(gpu_id, memory_gb, policy):
     asyncio.run(run_orchestrator())
 
 
-@cli.command()
+@orchestrator.command("register")
 @click.argument("model-id")
 @click.argument("model-path")
 @click.option(
@@ -8361,7 +8373,7 @@ def orchestrator_register(model_id, model_path, framework, priority, tags):
     print(f"  Tags: {', '.join(tag_set) if tag_set else 'None'}")
 
 
-@cli.command()
+@orchestrator.command("load")
 @click.argument("model-id")
 @click.option("--force", is_flag=True, help="Force loading even if memory is full")
 def orchestrator_load(model_id, force):
@@ -8385,7 +8397,7 @@ def orchestrator_load(model_id, force):
     asyncio.run(load_model())
 
 
-@cli.command()
+@orchestrator.command("evict")
 @click.argument("model-id")
 def orchestrator_evict(model_id):
     """Evict a model from GPU memory"""
@@ -8403,7 +8415,7 @@ def orchestrator_evict(model_id):
     asyncio.run(evict_model())
 
 
-@cli.command()
+@orchestrator.command("status")
 @click.option("--model-id", help="Get details for specific model")
 def orchestrator_status(model_id):
     """Get orchestrator and model status"""
@@ -8447,7 +8459,7 @@ def orchestrator_status(model_id):
                 print(f"  {state}: {', '.join(model_ids)}")
 
 
-@cli.command()
+@orchestrator.command("infer")
 @click.argument("model-id")
 def orchestrator_infer(model_id):
     """Test inference with a model"""
@@ -8466,7 +8478,13 @@ def orchestrator_infer(model_id):
     asyncio.run(test_inference())
 
 
-@cli.command()
+@cli.group(name="warm-pool")
+def warm_pool():
+    """Warm pool manager for intelligent pre-warming"""
+    pass
+
+
+@warm_pool.command("start")
 @click.option(
     "--strategy",
     type=click.Choice(
@@ -8527,7 +8545,7 @@ def warm_pool_start(strategy, max_warm, min_warm):
     asyncio.run(run_warm_pool())
 
 
-@cli.command()
+@warm_pool.command("register")
 @click.argument("model-id")
 @click.option("--priority", default=0, help="Model priority for warming")
 def warm_pool_register(model_id, priority):
@@ -8541,7 +8559,7 @@ def warm_pool_register(model_id, priority):
     print(f"  Priority: {priority}")
 
 
-@cli.command()
+@warm_pool.command("status")
 def warm_pool_status():
     """Get warm pool manager status"""
     from terradev_cli.core.warm_pool_manager import WarmPoolManager, WarmPoolConfig
@@ -8563,7 +8581,13 @@ def warm_pool_status():
     print(f"  Cost saved: ${status['cost_saved_usd']:.2f}")
 
 
-@cli.command()
+@cli.group()
+def cost_scaler():
+    """Cost-aware scaling manager for inference optimization"""
+    pass
+
+
+@cost_scaler.command("start")
 @click.option(
     "--strategy",
     type=click.Choice(
@@ -8623,7 +8647,7 @@ def cost_scaler_start(strategy, budget, cost_per_gb):
     asyncio.run(run_cost_scaler())
 
 
-@cli.command()
+@cost_scaler.command("status")
 def cost_scaler_status():
     """Get cost scaler status and recommendations"""
     from terradev_cli.core.cost_scaler import CostScaler, CostConfig
@@ -8655,7 +8679,7 @@ def cost_scaler_status():
             print(f"    Potential savings: {rec['potential_savings']}")
 
 
-@cli.command()
+@cost_scaler.command("model-details")
 @click.argument("model-id")
 def cost_scaler_model_details(model_id):
     """Get cost details for a specific model"""
@@ -9499,7 +9523,13 @@ def preflight(nodes, ssh_user, ssh_key, provision_group, quick, fmt):
         print()
 
 
-@cli.command()
+@cli.group()
+def train():
+    """Launch distributed training jobs across provisioned GPU nodes"""
+    pass
+
+
+@train.command("start")
 @click.option(
     "--config",
     "-c",
@@ -9557,7 +9587,7 @@ def preflight(nodes, ssh_user, ssh_key, provision_group, quick, fmt):
     help="Output format: text (default) or json",
 )
 @click.argument("script_args", nargs=-1, type=click.UNPROCESSED)
-def train(
+def train_start(
     config_path,
     script,
     framework,
@@ -9880,7 +9910,7 @@ def checkpoint(action, job_id, step, checkpoint_id, dest, fmt):
         print(f"Deleted: {checkpoint_id}")
 
 
-@cli.command("train-status")
+@train.command("status")
 @click.option("--job-id", "-j", default="", help="Job ID (empty = all running)")
 @click.option(
     "--format", "-f", "fmt", type=click.Choice(["json", "text"]), default="text"
@@ -9954,7 +9984,7 @@ def train_status(job_id, fmt):
             print(f"\nTotal cost across all jobs: ${total:.2f}\n")
 
 
-@cli.command("train-stop")
+@train.command("stop")
 @click.option("--job-id", "-j", required=True, help="Job ID to stop")
 @click.option(
     "--format", "-f", "fmt", type=click.Choice(["json", "text"]), default="text"
@@ -9978,7 +10008,7 @@ def train_stop(job_id, fmt):
         print(f"Job {job_id}: {result.get('status', 'unknown')}")
 
 
-@cli.command("train-resume")
+@train.command("resume")
 @click.option("--job-id", "-j", required=True, help="Job ID to resume")
 @click.option(
     "--checkpoint-id", default="", help="Checkpoint to resume from (default: latest)"
@@ -16017,6 +16047,168 @@ def gateway(host, port, openai, no_openai, anthropic, no_anthropic, custom, no_c
         sys.exit(1)
 
 
+# Observe Command - Unified Monitoring Pipeline
+@cli.group()
+def observe():
+    """Unified observability pipeline - W&B, Phoenix, Cost Analytics with shared trace ID"""
+    pass
+
+
+@observe.command("gateway")
+@click.argument("gateway_endpoint")
+@click.option("--wandb-project", help="W&B project name")
+@click.option("--wandb-entity", help="W&B entity/team")
+@click.option("--phoenix-endpoint", help="Phoenix endpoint URL")
+@click.option("--enable-cost-analytics/--disable-cost-analytics", default=True, help="Enable cost analytics tracking")
+@click.option("--duration", default=3600, help="Observation duration in seconds")
+@click.option("--sample-rate", default=1.0, help="Sampling rate (0.0-1.0)")
+def observe_gateway(
+    gateway_endpoint,
+    wandb_project,
+    wandb_entity,
+    phoenix_endpoint,
+    enable_cost_analytics,
+    duration,
+    sample_rate
+):
+    """Observe API Gateway traffic across W&B, Phoenix, and Cost Analytics"""
+    import asyncio
+    
+    async def run_observe():
+        from terradev_cli.core.observe import observe_gateway_traffic
+        
+        result = await observe_gateway_traffic(
+            gateway_endpoint=gateway_endpoint,
+            wandb_project=wandb_project,
+            wandb_entity=wandb_entity,
+            phoenix_endpoint=phoenix_endpoint,
+            enable_cost_analytics=enable_cost_analytics,
+            duration_seconds=duration,
+            sample_rate=sample_rate
+        )
+        
+        print(f"\n📊 Observability Summary:")
+        print(f"   Trace ID: {result['trace_id']}")
+        print(f"   Active Destinations: {', '.join(result['active_destinations'])}")
+        print(f"   Start Time: {result['start_time']}")
+        
+        return result
+    
+    try:
+        result = asyncio.run(run_observe())
+        return 0 if result else 1
+    except Exception as e:  # noqa: BLE001
+        print(f"ERROR: Observability pipeline failed: {e}")
+        return 1
+
+
+@observe.command("status")
+@click.argument("trace_id")
+def observe_status(trace_id):
+    """Get status of an observability trace"""
+    import asyncio
+    
+    async def run_status():
+        from terradev_cli.core.observe import observe_status
+        
+        result = await observe_status(trace_id)
+        
+        print(f"\n📊 Trace Status: {trace_id}")
+        print(f"   Overall Status: {result['status']}")
+        
+        for dest, info in result.items():
+            if dest != "trace_id" and dest != "status":
+                print(f"   {dest.upper()}: {info}")
+        
+        return result
+    
+    try:
+        result = asyncio.run(run_status())
+        return 0 if result else 1
+    except Exception as e:  # noqa: BLE001
+        print(f"ERROR: Failed to get trace status: {e}")
+        return 1
+
+
+# Schedule Command - Spot-Aware Scheduling
+@cli.group()
+def schedule():
+    """Spot-aware scheduling for cost-optimized job execution"""
+    pass
+
+
+@schedule.command("job")
+@click.argument("command")
+@click.argument("gpu_type")
+@click.option("--cron", help="Cron expression for recurring jobs")
+@click.option("--max-wait-hours", default=24, help="Maximum hours to wait for optimal pricing")
+@click.option("--job-name", help="Custom job name")
+@click.option("--prefer-current/--no-prefer-current", default=True, help="Prefer currently active pricing window")
+def schedule_job(command, gpu_type, cron, max_wait_hours, job_name, prefer_current):
+    """Schedule a job with spot pricing awareness"""
+    import asyncio
+    
+    async def run_schedule():
+        from terradev_cli.core.schedule import schedule_spot_job
+        
+        result = await schedule_spot_job(
+            command=command,
+            gpu_type=gpu_type,
+            cron_expression=cron,
+            max_wait_hours=max_wait_hours,
+            job_name=job_name,
+            prefer_current_window=prefer_current
+        )
+        
+        return result
+    
+    try:
+        result = asyncio.run(run_schedule())
+        return 0 if result.get("status") == "success" else 1
+    except Exception as e:  # noqa: BLE001
+        print(f"ERROR: Failed to schedule job: {e}")
+        return 1
+
+
+@schedule.command("list")
+def schedule_list_cmd():
+    """List all scheduled jobs"""
+    import asyncio
+    
+    async def run_list():
+        from terradev_cli.core.schedule import schedule_list
+        
+        result = await schedule_list()
+        return result
+    
+    try:
+        result = asyncio.run(run_list())
+        return 0
+    except Exception as e:  # noqa: BLE001
+        print(f"ERROR: Failed to list jobs: {e}")
+        return 1
+
+
+@schedule.command("windows")
+@click.option("--gpu-type", help="Filter by GPU type")
+def schedule_windows(gpu_type):
+    """Show available spot pricing windows"""
+    import asyncio
+    
+    async def run_windows():
+        from terradev_cli.core.schedule import schedule_pricing_windows
+        
+        result = await schedule_pricing_windows(gpu_type)
+        return result
+    
+    try:
+        result = asyncio.run(run_windows())
+        return 0
+    except Exception as e:  # noqa: BLE001
+        print(f"ERROR: Failed to get pricing windows: {e}")
+        return 1
+
+
 if __name__ == "__main__":
     cli()