askrobots / dbbasic-object-primitive
Public
Python functions as REST objects, with clustering, load balancing and state replication; no containers or orchestration.
test_cluster_integration.py · 16.0 KB · 443 lines
| 1 | """ |
| 2 | Integration tests for 3-station cluster. |
| 3 | |
| 4 | Runs a full 3-station cluster locally on high ports (9001-9003) |
| 5 | to test distributed features without requiring SSH or remote machines. |
| 6 | |
| 7 | Tests: |
| 8 | - State replication across all 3 stations |
| 9 | - Log replication consistency |
| 10 | - Load balancing |
| 11 | - Failover scenarios |
| 12 | """ |
| 13 | |
| 14 | import pytest |
| 15 | import subprocess |
| 16 | import time |
| 17 | import tempfile |
| 18 | import shutil |
| 19 | import os |
| 20 | import requests |
| 21 | from pathlib import Path |
| 22 | |
| 23 | |
| 24 | class LocalCluster: |
| 25 | """Manages a 3-station cluster running locally on different ports.""" |
| 26 | |
| 27 | def __init__(self): |
| 28 | self.base_dir = None |
| 29 | self.stations = [] |
| 30 | self.server_processes = [] |
| 31 | self.heartbeat_processes = [] |
| 32 | |
| 33 | def setup(self): |
| 34 | """Start a 3-station cluster locally.""" |
| 35 | # Create temp directory for cluster |
| 36 | self.base_dir = Path(tempfile.mkdtemp(prefix="test_cluster_")) |
| 37 | print(f"Test cluster base: {self.base_dir}") |
| 38 | |
| 39 | # Define stations |
| 40 | self.stations = [ |
| 41 | {"id": "station1", "port": 9001, "is_master": True}, |
| 42 | {"id": "station2", "port": 9002, "is_master": False}, |
| 43 | {"id": "station3", "port": 9003, "is_master": False}, |
| 44 | ] |
| 45 | |
| 46 | # Get source directory (project root) |
| 47 | source_dir = Path(__file__).parent.parent.parent |
| 48 | |
| 49 | # Create station directories with necessary files |
| 50 | for station in self.stations: |
| 51 | station_dir = self.base_dir / station["id"] |
| 52 | station_dir.mkdir() |
| 53 | |
| 54 | # Create data directories |
| 55 | (station_dir / "data" / "cluster").mkdir(parents=True) |
| 56 | (station_dir / "data" / "state").mkdir(parents=True) |
| 57 | (station_dir / "data" / "logs").mkdir(parents=True) |
| 58 | (station_dir / "data" / "versions").mkdir(parents=True) |
| 59 | |
| 60 | # Copy entire src, api, examples directories (preserve structure) |
| 61 | for dir_name in ["src", "api", "examples"]: |
| 62 | src_path = source_dir / dir_name |
| 63 | if src_path.exists(): |
| 64 | shutil.copytree(src_path, station_dir / dir_name) |
| 65 | |
| 66 | # Copy scripts |
| 67 | for script in ["run_server.py", "cluster_heartbeat_daemon.py"]: |
| 68 | src_script = source_dir / script |
| 69 | if src_script.exists(): |
| 70 | shutil.copy(src_script, station_dir / script) |
| 71 | |
| 72 | station["dir"] = station_dir |
| 73 | |
| 74 | # Initialize cluster registry on master (station1) |
| 75 | self._init_registry() |
| 76 | |
| 77 | # Start all servers |
| 78 | for station in self.stations: |
| 79 | self._start_server(station) |
| 80 | |
| 81 | # Wait for servers to start |
| 82 | time.sleep(3) |
| 83 | |
| 84 | # Verify all servers are running |
| 85 | for station in self.stations: |
| 86 | url = f"http://localhost:{station['port']}/cluster/info" |
| 87 | resp = requests.get(url, timeout=5) |
| 88 | assert resp.status_code == 200, f"{station['id']} not responding" |
| 89 | print(f"✓ {station['id']} running on port {station['port']}") |
| 90 | |
| 91 | # Start heartbeat daemons (workers only, master doesn't send heartbeats) |
| 92 | for station in self.stations: |
| 93 | if not station["is_master"]: |
| 94 | self._start_heartbeat(station) |
| 95 | |
| 96 | # Wait for heartbeats to register |
| 97 | time.sleep(3) |
| 98 | |
| 99 | print("✓ Test cluster ready") |
| 100 | |
| 101 | def _init_registry(self): |
| 102 | """Initialize cluster registry on master station.""" |
| 103 | registry_path = self.stations[0]["dir"] / "data" / "cluster" / "stations.tsv" |
| 104 | |
| 105 | with open(registry_path, "w") as f: |
| 106 | # Write header |
| 107 | f.write("station_id\taddress\tport\tlast_heartbeat\tis_active\n") |
| 108 | |
| 109 | # Write all stations |
| 110 | for station in self.stations: |
| 111 | # Master starts as active, workers will activate via heartbeat |
| 112 | is_active = "True" if station["is_master"] else "False" |
| 113 | f.write(f"{station['id']}\tlocalhost\t{station['port']}\t0\t{is_active}\n") |
| 114 | |
| 115 | def _start_server(self, station): |
| 116 | """Start server process for a station.""" |
| 117 | env = os.environ.copy() |
| 118 | env["STATION_ID"] = station["id"] |
| 119 | env["SERVER_PORT"] = str(station["port"]) # For SelfLogger to query correct port |
| 120 | |
| 121 | # Workers need to know where the master is (for log replication) |
| 122 | if not station["is_master"]: |
| 123 | master_station = self.stations[0] # station1 is always master |
| 124 | env["MASTER_HOST"] = "localhost" |
| 125 | env["MASTER_PORT"] = str(master_station["port"]) |
| 126 | |
| 127 | # Add packages directory to PYTHONPATH so server can import dbbasic packages |
| 128 | source_dir = Path(__file__).parent.parent.parent |
| 129 | packages_dir = source_dir / "packages" |
| 130 | if "PYTHONPATH" in env: |
| 131 | env["PYTHONPATH"] = f"{packages_dir}:{env['PYTHONPATH']}" |
| 132 | else: |
| 133 | env["PYTHONPATH"] = str(packages_dir) |
| 134 | |
| 135 | cmd = [ |
| 136 | "python", |
| 137 | "run_server.py", |
| 138 | "--port", str(station["port"]) |
| 139 | ] |
| 140 | |
| 141 | log_file = open(station["dir"] / "server.log", "w") |
| 142 | |
| 143 | proc = subprocess.Popen( |
| 144 | cmd, |
| 145 | cwd=station["dir"], # Run from station directory |
| 146 | env=env, |
| 147 | stdout=log_file, |
| 148 | stderr=subprocess.STDOUT, |
| 149 | start_new_session=True # Detach from parent |
| 150 | ) |
| 151 | |
| 152 | self.server_processes.append((proc, log_file)) |
| 153 | station["server_pid"] = proc.pid |
| 154 | print(f"Started {station['id']} server (PID {proc.pid})") |
| 155 | |
| 156 | def _start_heartbeat(self, station): |
| 157 | """Start heartbeat daemon for a worker station.""" |
| 158 | env = os.environ.copy() |
| 159 | env["STATION_ID"] = station["id"] |
| 160 | |
| 161 | # Add packages directory to PYTHONPATH |
| 162 | source_dir = Path(__file__).parent.parent.parent |
| 163 | packages_dir = source_dir / "packages" |
| 164 | if "PYTHONPATH" in env: |
| 165 | env["PYTHONPATH"] = f"{packages_dir}:{env['PYTHONPATH']}" |
| 166 | else: |
| 167 | env["PYTHONPATH"] = str(packages_dir) |
| 168 | |
| 169 | master_station = self.stations[0] # station1 is master |
| 170 | master_addr = "localhost" |
| 171 | |
| 172 | cmd = [ |
| 173 | "python", |
| 174 | "cluster_heartbeat_daemon.py", |
| 175 | master_addr, |
| 176 | "--master-port", str(master_station["port"]), |
| 177 | "--station-port", str(station["port"]) |
| 178 | ] |
| 179 | |
| 180 | log_file = open(station["dir"] / "heartbeat.log", "w") |
| 181 | |
| 182 | proc = subprocess.Popen( |
| 183 | cmd, |
| 184 | cwd=station["dir"], |
| 185 | env=env, |
| 186 | stdout=log_file, |
| 187 | stderr=subprocess.STDOUT, |
| 188 | start_new_session=True |
| 189 | ) |
| 190 | |
| 191 | self.heartbeat_processes.append((proc, log_file)) |
| 192 | station["heartbeat_pid"] = proc.pid |
| 193 | print(f"Started {station['id']} heartbeat (PID {proc.pid})") |
| 194 | |
| 195 | def teardown(self): |
| 196 | """Stop all processes and cleanup.""" |
| 197 | print("Tearing down test cluster...") |
| 198 | |
| 199 | # Kill all processes |
| 200 | for proc, log_file in self.server_processes + self.heartbeat_processes: |
| 201 | try: |
| 202 | proc.terminate() |
| 203 | proc.wait(timeout=5) |
| 204 | except subprocess.TimeoutExpired: |
| 205 | proc.kill() |
| 206 | finally: |
| 207 | log_file.close() |
| 208 | |
| 209 | # Clean up temp directory |
| 210 | if self.base_dir and self.base_dir.exists(): |
| 211 | shutil.rmtree(self.base_dir) |
| 212 | print(f"✓ Cleaned up {self.base_dir}") |
| 213 | |
| 214 | def get_url(self, station_id, path): |
| 215 | """Get full URL for a path on a station.""" |
| 216 | station = next(s for s in self.stations if s["id"] == station_id) |
| 217 | return f"http://localhost:{station['port']}{path}" |
| 218 | |
| 219 | def get_master_url(self, path): |
| 220 | """Get full URL for a path on the master station.""" |
| 221 | return self.get_url("station1", path) |
| 222 | |
| 223 | |
| 224 | @pytest.fixture(scope="module") |
| 225 | def cluster(): |
| 226 | """Pytest fixture that provides a running 3-station cluster.""" |
| 227 | c = LocalCluster() |
| 228 | c.setup() |
| 229 | yield c |
| 230 | c.teardown() |
| 231 | |
| 232 | |
| 233 | class TestClusterBasics: |
| 234 | """Test basic cluster operations.""" |
| 235 | |
| 236 | def test_all_stations_running(self, cluster): |
| 237 | """Verify all 3 stations are running.""" |
| 238 | for station in cluster.stations: |
| 239 | url = cluster.get_url(station["id"], "/cluster/info") |
| 240 | resp = requests.get(url, timeout=5) |
| 241 | assert resp.status_code == 200 |
| 242 | data = resp.json() |
| 243 | assert data["station_id"] == station["id"] |
| 244 | |
| 245 | def test_registry_has_all_stations(self, cluster): |
| 246 | """Verify cluster registry knows about all stations.""" |
| 247 | url = cluster.get_master_url("/cluster/stations") |
| 248 | resp = requests.get(url, timeout=5) |
| 249 | |
| 250 | # Print error details if request failed |
| 251 | if resp.status_code != 200: |
| 252 | print(f"\nError: {resp.status_code}") |
| 253 | print(f"Response: {resp.text}") |
| 254 | |
| 255 | assert resp.status_code == 200 |
| 256 | |
| 257 | data = resp.json() |
| 258 | assert data["status"] == "ok" |
| 259 | assert len(data["stations"]) == 3 |
| 260 | |
| 261 | station_ids = [s["station_id"] for s in data["stations"]] |
| 262 | assert "station1" in station_ids |
| 263 | assert "station2" in station_ids |
| 264 | assert "station3" in station_ids |
| 265 | |
| 266 | def test_workers_are_active(self, cluster): |
| 267 | """Verify worker stations registered via heartbeat.""" |
| 268 | # Give heartbeats time to register |
| 269 | time.sleep(5) |
| 270 | |
| 271 | url = cluster.get_master_url("/cluster/stations") |
| 272 | resp = requests.get(url, timeout=5) |
| 273 | data = resp.json() |
| 274 | |
| 275 | for station in data["stations"]: |
| 276 | if station["station_id"] in ["station2", "station3"]: |
| 277 | assert station["is_active"], f"{station['station_id']} should be active" |
| 278 | |
| 279 | |
| 280 | class TestStateReplication: |
| 281 | """Test state replication across stations.""" |
| 282 | |
| 283 | def test_state_replicates_to_all_stations(self, cluster): |
| 284 | """Write state on station1, verify it appears on all stations.""" |
| 285 | # Create a test object on station1 |
| 286 | test_value = int(time.time()) |
| 287 | |
| 288 | # Write to counter object on station1 |
| 289 | url = cluster.get_master_url("/objects/basics_counter") |
| 290 | resp = requests.post( |
| 291 | url, |
| 292 | json={"value": test_value}, |
| 293 | headers={"Content-Type": "application/json"}, |
| 294 | timeout=5 |
| 295 | ) |
| 296 | |
| 297 | # Debug: print response details |
| 298 | print(f"\nPOST {url}") |
| 299 | print(f"Status: {resp.status_code}") |
| 300 | print(f"Response: {resp.text[:500]}") |
| 301 | |
| 302 | assert resp.status_code == 200 |
| 303 | |
| 304 | # Wait for replication |
| 305 | time.sleep(2) |
| 306 | |
| 307 | # Debug: Check what's in the data directory |
| 308 | print(f"\nChecking data directories on station1...") |
| 309 | station1_data = cluster.stations[0]["dir"] / "data" |
| 310 | if station1_data.exists(): |
| 311 | import os |
| 312 | for root, dirs, files in os.walk(station1_data): |
| 313 | rel_path = os.path.relpath(root, station1_data) |
| 314 | print(f" {rel_path}/ - dirs: {dirs}, files: {files}") |
| 315 | else: |
| 316 | print(f" data/ directory doesn't exist!") |
| 317 | |
| 318 | # Read state file on all 3 stations |
| 319 | # NOTE: StateManager uses filename instead of full path ID, so "counter" not "basics_counter" |
| 320 | for station in cluster.stations: |
| 321 | state_file = station["dir"] / "data" / "state" / "counter" / "state.tsv" |
| 322 | |
| 323 | # State file should exist |
| 324 | assert state_file.exists(), f"State file missing on {station['id']}" |
| 325 | |
| 326 | # Read state value |
| 327 | with open(state_file) as f: |
| 328 | lines = f.readlines() |
| 329 | assert len(lines) >= 1 # Just the data line (no header) |
| 330 | |
| 331 | # Parse TSV (format: count\tVALUE\tTIMESTAMP) |
| 332 | data_line = lines[0].strip().split("\t") |
| 333 | actual_value = int(data_line[1]) |
| 334 | |
| 335 | assert actual_value == test_value, \ |
| 336 | f"State mismatch on {station['id']}: expected {test_value}, got {actual_value}" |
| 337 | |
| 338 | print(f"✓ State {test_value} replicated to all 3 stations") |
| 339 | |
| 340 | |
| 341 | class TestLogReplication: |
| 342 | """Test log replication across stations.""" |
| 343 | |
| 344 | def test_logs_replicate_to_all_stations(self, cluster): |
| 345 | """Write logs on station2, verify they appear on all stations.""" |
| 346 | # Get current log count on all stations |
| 347 | # NOTE: StateManager uses filename instead of full path ID, so "counter" not "basics_counter" |
| 348 | initial_counts = {} |
| 349 | for station in cluster.stations: |
| 350 | log_file = station["dir"] / "data" / "logs" / "counter" / "log.tsv" |
| 351 | if log_file.exists(): |
| 352 | with open(log_file) as f: |
| 353 | initial_counts[station["id"]] = len(f.readlines()) |
| 354 | else: |
| 355 | initial_counts[station["id"]] = 0 |
| 356 | |
| 357 | # Trigger some operations that generate logs (via station2) |
| 358 | url = cluster.get_url("station2", "/objects/basics_counter") |
| 359 | for i in range(5): |
| 360 | resp = requests.post( |
| 361 | url, |
| 362 | json={"value": 1000 + i}, |
| 363 | headers={"Content-Type": "application/json"}, |
| 364 | timeout=5 |
| 365 | ) |
| 366 | assert resp.status_code == 200 |
| 367 | |
| 368 | # Wait for log replication |
| 369 | time.sleep(3) |
| 370 | |
| 371 | # First check if logs were created on station2 (where requests were made) |
| 372 | station2_log_file = cluster.stations[1]["dir"] / "data" / "logs" / "counter" / "log.tsv" |
| 373 | print(f"\nChecking if logs were created on station2...") |
| 374 | print(f"Log file path: {station2_log_file}") |
| 375 | print(f"Exists: {station2_log_file.exists()}") |
| 376 | |
| 377 | if station2_log_file.exists(): |
| 378 | with open(station2_log_file) as f: |
| 379 | lines = f.readlines() |
| 380 | print(f"Station2 has {len(lines)} log lines") |
| 381 | if len(lines) > 0: |
| 382 | print(f"First line: {lines[0][:100]}") |
| 383 | else: |
| 384 | # Log file doesn't exist on station2 - logging isn't working at all! |
| 385 | # Let's check what directories do exist |
| 386 | station2_data = cluster.stations[1]["dir"] / "data" |
| 387 | print(f"\nContents of {station2_data}:") |
| 388 | import os |
| 389 | for root, dirs, files in os.walk(station2_data): |
| 390 | rel_path = os.path.relpath(root, station2_data) |
| 391 | print(f" {rel_path}/ - dirs: {dirs}, files: {files}") |
| 392 | |
| 393 | # Require logs to exist on station2 first (origin station) |
| 394 | assert station2_log_file.exists(), "Logs not created on origin station (station2)" |
| 395 | |
| 396 | with open(station2_log_file) as f: |
| 397 | station2_count = len(f.readlines()) |
| 398 | assert station2_count > initial_counts["station2"], \ |
| 399 | f"No new logs on station2 (initial: {initial_counts['station2']}, final: {station2_count})" |
| 400 | |
| 401 | # Now check all stations have new logs |
| 402 | for station in cluster.stations: |
| 403 | log_file = station["dir"] / "data" / "logs" / "counter" / "log.tsv" |
| 404 | assert log_file.exists(), f"Log file missing on {station['id']}" |
| 405 | |
| 406 | with open(log_file) as f: |
| 407 | final_count = len(f.readlines()) |
| 408 | |
| 409 | # Should have at least some of the 5 new logs |
| 410 | # (may not be all due to fire-and-forget nature) |
| 411 | new_logs = final_count - initial_counts[station["id"]] |
| 412 | assert new_logs > 0, \ |
| 413 | f"{station['id']} has no new logs (initial: {initial_counts[station['id']]}, final: {final_count})" |
| 414 | |
| 415 | print("✓ Logs replicated to all stations") |
| 416 | |
| 417 | |
| 418 | class TestLoadBalancing: |
| 419 | """Test load balancing across stations.""" |
| 420 | |
| 421 | def test_stations_report_load_metrics(self, cluster): |
| 422 | """Verify stations report CPU and memory metrics.""" |
| 423 | # Wait for heartbeats with metrics |
| 424 | time.sleep(5) |
| 425 | |
| 426 | url = cluster.get_master_url("/cluster/stations") |
| 427 | resp = requests.get(url, timeout=5) |
| 428 | data = resp.json() |
| 429 | |
| 430 | for station in data["stations"]: |
| 431 | if station["is_active"]: |
| 432 | # Workers send metrics, master might not have them |
| 433 | if station["station_id"] != "station1": |
| 434 | assert "metrics" in station, \ |
| 435 | f"{station['station_id']} missing metrics dict" |
| 436 | metrics = station["metrics"] |
| 437 | assert "cpu_percent" in metrics or "memory_percent" in metrics, \ |
| 438 | f"{station['station_id']} missing load metrics in metrics dict" |
| 439 | |
| 440 | |
| 441 | if __name__ == "__main__": |
| 442 | # Allow running tests directly |
| 443 | pytest.main([__file__, "-v", "-s"]) |