File size: 11,992 Bytes
81e3673 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 | """
Soak Tests for Cache Stability Under Concurrent Load
Extended duration tests (30 minutes) to validate cache behavior:
- Concurrent operations (10 workers, 10000 ops each)
- Cache consistency (no race conditions, no data loss)
- TTL expiration (entries expire correctly)
- Cache unbounded growth detection
Cache issues detected:
- Race conditions in concurrent get/set operations
- Cache corruption (data loss, incorrect values)
- TTL not working (cache grows unbounded)
- Deadlocks under concurrent load
Tests use ThreadPoolExecutor for realistic concurrent access patterns.
"""
import gc
import pytest
import time
from concurrent.futures import ThreadPoolExecutor, as_completed
from core.governance_cache import GovernanceCache
@pytest.mark.soak
@pytest.mark.timeout(1800) # 30 minutes
def test_cache_concurrent_operations_30min(
memory_monitor,
enable_gc_control,
soak_test_config
):
"""
Soak test for cache stability under concurrent operations (30 minutes).
Validates:
- Cache remains consistent under concurrent load (10 workers)
- No race conditions or deadlocks
- No cache corruption (data loss, incorrect values)
- Zero cache errors (get returns None after set)
Test pattern:
- Run for 30 minutes (1800 seconds)
- Use ThreadPoolExecutor with 10 workers
- Each worker performs 10000 cache operations (set + get)
- Validate cache consistency: if get returns None after set, fail
- Log worker completions
- Force GC every 10 iterations
Duration: 30 minutes
Workers: 10 concurrent threads
Operations per worker: 10000 (set + get)
Failure: Any cache inconsistency detected
Race conditions detected:
- Get returns None after set (data loss)
- Incorrect value returned (corruption)
- Deadlock (test hangs)
"""
cache = GovernanceCache(max_size=10000, ttl_seconds=60)
process = memory_monitor["process"]
initial_memory = memory_monitor["initial_memory_mb"]
errors = []
iterations = 0
def worker_operations(worker_id: int):
"""
Worker function for concurrent cache operations.
Each worker performs 10000 cache operations:
- Set key with unique worker_id to avoid conflicts
- Get key immediately after set
- Validate value is not None (cache consistency check)
Raises AssertionError if cache inconsistency detected.
"""
for i in range(10000):
# Concurrent write with unique key (worker_id avoids conflicts)
cache.set(
agent_id=f"agent_{worker_id}_{i}",
action_type="test_action",
data={"allowed": True, "maturity_level": "AUTONOMOUS"}
)
# Concurrent read
result = cache.get(
agent_id=f"agent_{worker_id}_{i}",
action_type="test_action"
)
# Validate cache consistency (get should not return None after set)
if result is None:
raise AssertionError(
f"Cache inconsistency: key agent_{worker_id}_{i} not found after write"
)
return f"Worker {worker_id} completed"
start_time = time.time()
# Run concurrent workers for 30 minutes
while time.time() - start_time < 1800:
# Submit 10 concurrent workers
with ThreadPoolExecutor(max_workers=10) as executor:
futures = [
executor.submit(worker_operations, worker_id)
for worker_id in range(10)
]
# Wait for all workers to complete, collect errors
for future in as_completed(futures):
try:
result = future.result()
print(result)
except AssertionError as e:
errors.append(str(e))
except Exception as e:
errors.append(f"Unexpected error: {e}")
iterations += 1
# Force garbage collection every 10 iterations
if iterations % 10 == 0:
enable_gc_control["collect"]()
# Log memory growth every 60 iterations
if iterations % 60 == 0:
current_memory = process.memory_info().rss / 1024 / 1024
memory_growth = current_memory - initial_memory
print(
f"Iteration {iterations}: "
f"Memory growth = {memory_growth:.2f}MB, "
f"Errors = {len(errors)}"
)
# Fail fast if memory growth > 500MB
if memory_growth > soak_test_config["fail_fast_threshold_mb"]:
pytest.fail(
f"FAIL-FAST: Memory leak detected - {memory_growth:.2f}MB growth "
f"(threshold: {soak_test_config['fail_fast_threshold_mb']}MB)"
)
# Final validation: zero cache errors
assert len(errors) == 0, (
f"Cache inconsistencies detected: {len(errors)} errors\n"
f"Errors: {errors[:5]}" # Show first 5 errors
)
print(f"\n✅ Soak test complete: {iterations} iterations, zero cache errors")
@pytest.mark.soak
@pytest.mark.timeout(900) # 15 minutes
def test_cache_ttl_expiration_15min(memory_monitor, enable_gc_control):
"""
Soak test for cache TTL expiration (15 minutes).
Validates:
- Cache entries expire after TTL (60 seconds)
- Cache doesn't grow unbounded (TTL prevents memory leaks)
- Expired entries are not returned (get returns None)
Test pattern:
- Create cache with TTL=60 seconds
- Add 1000 entries per iteration
- Validate entries exist immediately after adding
- Wait 70 seconds, validate entries are gone (expired)
- Repeat for 15 minutes
- Monitor memory usage
Duration: 15 minutes
TTL: 60 seconds
Validation: Entries expire correctly, cache doesn't grow unbounded
Purpose: Detect TTL bugs that cause unbounded cache growth.
"""
cache = GovernanceCache(max_size=10000, ttl_seconds=60)
process = memory_monitor["process"]
initial_memory = memory_monitor["initial_memory_mb"]
iterations = 0
start_time = time.time()
# Run TTL validation for 15 minutes
while time.time() - start_time < 900:
# Add 1000 entries
for i in range(1000):
cache.set(
agent_id=f"agent_{iterations}_{i}",
action_type="test_action",
data={"allowed": True, "maturity_level": "AUTONOMOUS"}
)
# Validate entries exist immediately
for i in range(100): # Check 100 random entries
result = cache.get(
agent_id=f"agent_{iterations}_{i}",
action_type="test_action"
)
assert result is not None, f"Entry not found immediately after set: agent_{iterations}_{i}"
print(f"Iteration {iterations}: Added 1000 entries, validated existence")
# Wait 70 seconds (TTL=60, +10 buffer)
time.sleep(70)
# Validate entries are expired (get returns None)
expired_count = 0
for i in range(100): # Check 100 random entries
result = cache.get(
agent_id=f"agent_{iterations}_{i}",
action_type="test_action"
)
if result is None:
expired_count += 1
# All entries should be expired
assert expired_count == 100, (
f"TTL expiration failed: {expired_count}/100 entries expired "
f"(expected: 100)"
)
print(f"Iteration {iterations}: All entries expired correctly (100/100)")
iterations += 1
# Force garbage collection
enable_gc_control["collect"]()
# Log memory growth
current_memory = process.memory_info().rss / 1024 / 1024
memory_growth = current_memory - initial_memory
print(f"Iteration {iterations}: Memory growth = {memory_growth:.2f}MB")
# Fail fast if memory growth > 500MB (indicates TTL not working)
if memory_growth > 500:
pytest.fail(
f"TTL expiration not working: cache grew {memory_growth:.2f}MB "
f"(entries should be expiring)"
)
# Final validation: memory growth should be reasonable (< 200MB for 15min)
final_memory = process.memory_info().rss / 1024 / 1024
memory_growth = final_memory - initial_memory
assert memory_growth < 200, (
f"Cache grew unbounded: {memory_growth:.2f}MB growth over 15 minutes "
f"(TTL expiration may not be working)"
)
print(f"\n✅ Soak test complete: {iterations} iterations, TTL expiration working correctly")
@pytest.mark.soak
@pytest.mark.timeout(1800) # 30 minutes
def test_cache_lru_eviction_stability_30min(memory_monitor, enable_gc_control):
"""
Soak test for cache LRU eviction stability (30 minutes).
Validates:
- LRU eviction works correctly (cache doesn't exceed max_size)
- Evicted entries are not returned (get returns None)
- Cache performance remains stable under load
Test pattern:
- Create cache with max_size=1000
- Add 2000 entries (exceeds max_size, triggers LRU eviction)
- Validate cache size <= max_size
- Verify oldest entries are evicted (get returns None)
- Repeat for 30 minutes
Duration: 30 minutes
Max size: 1000 entries
Validation: LRU eviction works, cache size stable
Purpose: Detect LRU eviction bugs that cause unbounded cache growth.
"""
cache = GovernanceCache(max_size=1000, ttl_seconds=60)
process = memory_monitor["process"]
initial_memory = memory_monitor["initial_memory_mb"]
iterations = 0
start_time = time.time()
# Run LRU eviction validation for 30 minutes
while time.time() - start_time < 1800:
# Add 2000 entries (exceeds max_size=1000)
for i in range(2000):
cache.set(
agent_id=f"agent_{iterations}_{i}",
action_type="test_action",
data={"allowed": True, "maturity_level": "AUTONOMOUS"}
)
# Validate oldest entries are evicted (first 1000 should be gone)
evicted_count = 0
for i in range(1000): # Check oldest 1000 entries
result = cache.get(
agent_id=f"agent_{iterations}_{i}",
action_type="test_action"
)
if result is None:
evicted_count += 1
# At least some entries should be evicted (LRU working)
# Not all 1000 will be evicted (TTL may not have expired yet)
# But at least 500 should be evicted due to LRU
assert evicted_count >= 500, (
f"LRU eviction not working: only {evicted_count}/1000 entries evicted "
f"(expected: >= 500)"
)
print(f"Iteration {iterations}: LRU eviction working ({evicted_count}/1000 evicted)")
iterations += 1
# Force garbage collection
enable_gc_control["collect"]()
# Log memory growth
current_memory = process.memory_info().rss / 1024 / 1024
memory_growth = current_memory - initial_memory
print(f"Iteration {iterations}: Memory growth = {memory_growth:.2f}MB")
# Fail fast if memory growth > 500MB (indicates LRU not working)
if memory_growth > 500:
pytest.fail(
f"LRU eviction not working: cache grew {memory_growth:.2f}MB "
f"(entries should be evicted)"
)
# Final validation: memory growth should be reasonable
final_memory = process.memory_info().rss / 1024 / 1024
memory_growth = final_memory - initial_memory
assert memory_growth < 200, (
f"Cache grew unbounded: {memory_growth:.2f}MB growth over 30 minutes "
f"(LRU eviction may not be working)"
)
print(f"\n✅ Soak test complete: {iterations} iterations, LRU eviction working correctly")
|