394 lines
11 KiB
Bash
Executable File
394 lines
11 KiB
Bash
Executable File
#!/bin/bash
|
|
# Test: State Recovery and Checkpoint Functionality
|
|
# Tests checkpoint creation, recovery, and rate limit handling
|
|
|
|
set -uo pipefail
|
|
# Note: Not using -e to allow collecting all test results
|
|
|
|
TEST_DIR=$(mktemp -d)
|
|
PASSED=0
|
|
FAILED=0
|
|
|
|
# Colors
|
|
RED='\033[0;31m'
|
|
GREEN='\033[0;32m'
|
|
YELLOW='\033[1;33m'
|
|
NC='\033[0m'
|
|
|
|
log_pass() { echo -e "${GREEN}[PASS]${NC} $1"; ((PASSED++)); }
|
|
log_fail() { echo -e "${RED}[FAIL]${NC} $1"; ((FAILED++)); }
|
|
log_test() { echo -e "${YELLOW}[TEST]${NC} $1"; }
|
|
|
|
cleanup() {
|
|
rm -rf "$TEST_DIR"
|
|
}
|
|
trap cleanup EXIT
|
|
|
|
cd "$TEST_DIR"
|
|
|
|
echo "========================================"
|
|
echo "Loki Mode State Recovery Tests"
|
|
echo "========================================"
|
|
echo ""
|
|
|
|
# Initialize structure
|
|
mkdir -p .loki/{state/{agents,checkpoints},queue,artifacts/backups}
|
|
|
|
# Create initial state
|
|
cat > .loki/state/orchestrator.json << 'EOF'
|
|
{
|
|
"version": "2.1.0",
|
|
"startupId": "test-session-001",
|
|
"phase": "development",
|
|
"agents": {"active":["eng-backend-01"],"idle":[],"failed":[],"totalSpawned":5},
|
|
"metrics": {"tasksCompleted":10,"tasksFailed":2,"deployments":0},
|
|
"circuitBreakers": {},
|
|
"lastCheckpoint": "",
|
|
"currentRelease": "0.1.0"
|
|
}
|
|
EOF
|
|
|
|
# Create agent state
|
|
cat > .loki/state/agents/eng-backend-01.json << 'EOF'
|
|
{
|
|
"id": "eng-backend-01",
|
|
"status": "active",
|
|
"currentTask": "task-042",
|
|
"tasksCompleted": 8,
|
|
"lastHeartbeat": "2025-01-15T10:30:00Z"
|
|
}
|
|
EOF
|
|
|
|
# Create queue state
|
|
cat > .loki/queue/pending.json << 'EOF'
|
|
{"tasks":[{"id":"task-043","type":"eng-frontend","priority":5}]}
|
|
EOF
|
|
cat > .loki/queue/in-progress.json << 'EOF'
|
|
{"tasks":[{"id":"task-042","type":"eng-backend","claimedBy":"eng-backend-01"}]}
|
|
EOF
|
|
|
|
# Test 1: Create checkpoint
|
|
log_test "Create checkpoint"
|
|
CHECKPOINT_DIR=".loki/state/checkpoints/$(date +%Y%m%d-%H%M%S)"
|
|
mkdir -p "$CHECKPOINT_DIR"
|
|
cp .loki/state/orchestrator.json "$CHECKPOINT_DIR/"
|
|
cp -r .loki/state/agents "$CHECKPOINT_DIR/"
|
|
cp -r .loki/queue "$CHECKPOINT_DIR/"
|
|
|
|
if [ -f "$CHECKPOINT_DIR/orchestrator.json" ] && [ -d "$CHECKPOINT_DIR/agents" ]; then
|
|
log_pass "Checkpoint created at $CHECKPOINT_DIR"
|
|
else
|
|
log_fail "Checkpoint creation failed"
|
|
fi
|
|
|
|
# Test 2: Update lastCheckpoint in state
|
|
log_test "Update lastCheckpoint timestamp"
|
|
python3 << EOF
|
|
import json
|
|
from datetime import datetime
|
|
|
|
with open('.loki/state/orchestrator.json', 'r') as f:
|
|
state = json.load(f)
|
|
|
|
state['lastCheckpoint'] = datetime.utcnow().isoformat() + 'Z'
|
|
|
|
with open('.loki/state/orchestrator.json', 'w') as f:
|
|
json.dump(state, f, indent=2)
|
|
|
|
print("UPDATED")
|
|
EOF
|
|
|
|
has_checkpoint=$(python3 -c "
|
|
import json
|
|
data = json.load(open('.loki/state/orchestrator.json'))
|
|
print('yes' if data.get('lastCheckpoint') else 'no')
|
|
")
|
|
|
|
if [ "$has_checkpoint" = "yes" ]; then
|
|
log_pass "lastCheckpoint timestamp updated"
|
|
else
|
|
log_fail "lastCheckpoint not set"
|
|
fi
|
|
|
|
# Test 3: Simulate crash and corrupt state
|
|
log_test "Detect corrupted state"
|
|
echo "corrupted{json" > .loki/state/orchestrator.json.corrupted
|
|
|
|
python3 << 'EOF'
|
|
import json
|
|
|
|
def is_valid_state(filepath):
|
|
try:
|
|
with open(filepath, 'r') as f:
|
|
data = json.load(f)
|
|
return isinstance(data, dict) and 'version' in data
|
|
except (json.JSONDecodeError, KeyError):
|
|
return False
|
|
|
|
is_valid = is_valid_state('.loki/state/orchestrator.json.corrupted')
|
|
print("CORRUPTED" if not is_valid else "VALID")
|
|
assert not is_valid, "Should detect corrupted state"
|
|
EOF
|
|
|
|
log_pass "Corrupted state detected"
|
|
|
|
# Test 4: Restore from checkpoint
|
|
log_test "Restore from checkpoint"
|
|
python3 << EOF
|
|
import json
|
|
import os
|
|
import shutil
|
|
from pathlib import Path
|
|
|
|
# Find latest checkpoint
|
|
checkpoints_dir = Path('.loki/state/checkpoints')
|
|
checkpoints = sorted(checkpoints_dir.iterdir(), reverse=True)
|
|
|
|
if checkpoints:
|
|
latest = checkpoints[0]
|
|
|
|
# Restore orchestrator state
|
|
if (latest / 'orchestrator.json').exists():
|
|
shutil.copy(latest / 'orchestrator.json', '.loki/state/orchestrator.json')
|
|
|
|
# Restore agent states
|
|
if (latest / 'agents').exists():
|
|
for agent_file in (latest / 'agents').iterdir():
|
|
shutil.copy(agent_file, f'.loki/state/agents/{agent_file.name}')
|
|
|
|
# Restore queue
|
|
if (latest / 'queue').exists():
|
|
for queue_file in (latest / 'queue').iterdir():
|
|
shutil.copy(queue_file, f'.loki/queue/{queue_file.name}')
|
|
|
|
print(f"RESTORED:{latest.name}")
|
|
else:
|
|
print("NO_CHECKPOINT")
|
|
EOF
|
|
|
|
# Verify restoration
|
|
restored_version=$(python3 -c "
|
|
import json
|
|
data = json.load(open('.loki/state/orchestrator.json'))
|
|
print(data.get('version', 'unknown'))
|
|
")
|
|
|
|
if [ "$restored_version" = "2.1.0" ]; then
|
|
log_pass "State restored from checkpoint"
|
|
else
|
|
log_fail "State restoration failed (version: $restored_version)"
|
|
fi
|
|
|
|
# Test 5: Orphaned task detection
|
|
log_test "Detect orphaned tasks"
|
|
python3 << 'EOF'
|
|
import json
|
|
from datetime import datetime, timedelta
|
|
|
|
CLAIM_TIMEOUT = 3600 # 1 hour
|
|
|
|
# Create an old claimed task
|
|
old_task = {
|
|
"id": "task-old-001",
|
|
"type": "eng-backend",
|
|
"claimedBy": "dead-agent-99",
|
|
"claimedAt": (datetime.utcnow() - timedelta(hours=2)).isoformat() + 'Z'
|
|
}
|
|
|
|
with open('.loki/queue/in-progress.json', 'r') as f:
|
|
in_progress = json.load(f)
|
|
|
|
in_progress['tasks'].append(old_task)
|
|
|
|
with open('.loki/queue/in-progress.json', 'w') as f:
|
|
json.dump(in_progress, f)
|
|
|
|
def find_orphaned_tasks(in_progress_tasks):
|
|
orphaned = []
|
|
now = datetime.utcnow()
|
|
|
|
for task in in_progress_tasks:
|
|
if task.get('claimedAt'):
|
|
claimed_at = datetime.fromisoformat(task['claimedAt'].replace('Z', '+00:00'))
|
|
age = (now.replace(tzinfo=claimed_at.tzinfo) - claimed_at).total_seconds()
|
|
if age > CLAIM_TIMEOUT:
|
|
orphaned.append(task['id'])
|
|
|
|
return orphaned
|
|
|
|
orphaned = find_orphaned_tasks(in_progress['tasks'])
|
|
print(f"ORPHANED:{len(orphaned)}")
|
|
assert len(orphaned) >= 1, "Should find orphaned task"
|
|
print("VERIFIED")
|
|
EOF
|
|
|
|
log_pass "Orphaned task detection works"
|
|
|
|
# Test 6: Re-queue orphaned tasks
|
|
log_test "Re-queue orphaned tasks"
|
|
python3 << 'EOF'
|
|
import json
|
|
from datetime import datetime, timedelta
|
|
|
|
CLAIM_TIMEOUT = 3600
|
|
|
|
with open('.loki/queue/in-progress.json', 'r') as f:
|
|
in_progress = json.load(f)
|
|
|
|
with open('.loki/queue/pending.json', 'r') as f:
|
|
pending = json.load(f)
|
|
|
|
now = datetime.utcnow()
|
|
requeued = []
|
|
|
|
for task in in_progress['tasks'][:]:
|
|
if task.get('claimedAt'):
|
|
claimed_at = datetime.fromisoformat(task['claimedAt'].replace('Z', '+00:00'))
|
|
age = (now.replace(tzinfo=claimed_at.tzinfo) - claimed_at).total_seconds()
|
|
|
|
if age > CLAIM_TIMEOUT:
|
|
# Re-queue: clear claim and move to pending
|
|
task['claimedBy'] = None
|
|
task['claimedAt'] = None
|
|
task['requeuedAt'] = now.isoformat() + 'Z'
|
|
task['requeueReason'] = 'claim_timeout'
|
|
|
|
pending['tasks'].append(task)
|
|
in_progress['tasks'].remove(task)
|
|
requeued.append(task['id'])
|
|
|
|
with open('.loki/queue/in-progress.json', 'w') as f:
|
|
json.dump(in_progress, f)
|
|
|
|
with open('.loki/queue/pending.json', 'w') as f:
|
|
json.dump(pending, f)
|
|
|
|
print(f"REQUEUED:{len(requeued)}")
|
|
EOF
|
|
|
|
log_pass "Orphaned tasks re-queued"
|
|
|
|
# Test 7: Rate limit backoff simulation
|
|
log_test "Rate limit exponential backoff"
|
|
python3 << 'EOF'
|
|
import time
|
|
import random
|
|
|
|
def calculate_backoff(attempt, base_delay=60, max_delay=3600):
|
|
"""Calculate exponential backoff with jitter"""
|
|
delay = min(base_delay * (2 ** attempt), max_delay)
|
|
jitter = random.uniform(0, delay * 0.1)
|
|
return delay + jitter
|
|
|
|
# Test backoff progression
|
|
delays = []
|
|
for attempt in range(5):
|
|
delay = calculate_backoff(attempt)
|
|
delays.append(int(delay))
|
|
print(f"Attempt {attempt}: {delay:.0f}s")
|
|
|
|
# Verify exponential growth
|
|
assert delays[0] >= 60, "Initial delay should be ~60s"
|
|
assert delays[1] >= 120, "Second delay should be ~120s"
|
|
assert delays[2] >= 240, "Third delay should be ~240s"
|
|
assert delays[4] <= 4000, "Should cap at max_delay"
|
|
|
|
print("VERIFIED")
|
|
EOF
|
|
|
|
log_pass "Exponential backoff works"
|
|
|
|
# Test 8: Full system recovery
|
|
log_test "Full system recovery simulation"
|
|
python3 << 'EOF'
|
|
import json
|
|
import os
|
|
from pathlib import Path
|
|
from datetime import datetime, timedelta
|
|
|
|
def recover_system():
|
|
"""Full system recovery procedure"""
|
|
recovery_log = []
|
|
|
|
# 1. Check orchestrator state
|
|
try:
|
|
with open('.loki/state/orchestrator.json', 'r') as f:
|
|
state = json.load(f)
|
|
recovery_log.append("Orchestrator state: OK")
|
|
except:
|
|
recovery_log.append("Orchestrator state: RESTORE FROM CHECKPOINT")
|
|
# Would restore here
|
|
|
|
# 2. Check agent states
|
|
agents_dir = Path('.loki/state/agents')
|
|
active_agents = []
|
|
dead_agents = []
|
|
|
|
for agent_file in agents_dir.glob('*.json'):
|
|
with open(agent_file, 'r') as f:
|
|
agent = json.load(f)
|
|
|
|
# Check heartbeat
|
|
if agent.get('lastHeartbeat'):
|
|
hb = datetime.fromisoformat(agent['lastHeartbeat'].replace('Z', '+00:00'))
|
|
age = (datetime.now(hb.tzinfo) - hb).total_seconds()
|
|
if age > 600: # 10 min heartbeat timeout
|
|
dead_agents.append(agent['id'])
|
|
else:
|
|
active_agents.append(agent['id'])
|
|
|
|
recovery_log.append(f"Active agents: {len(active_agents)}")
|
|
recovery_log.append(f"Dead agents: {len(dead_agents)}")
|
|
|
|
# 3. Re-queue tasks from dead agents
|
|
with open('.loki/queue/in-progress.json', 'r') as f:
|
|
in_progress = json.load(f)
|
|
|
|
requeued = 0
|
|
for task in in_progress['tasks'][:]:
|
|
if task.get('claimedBy') in dead_agents:
|
|
task['claimedBy'] = None
|
|
task['claimedAt'] = None
|
|
requeued += 1
|
|
|
|
with open('.loki/queue/in-progress.json', 'w') as f:
|
|
json.dump(in_progress, f)
|
|
|
|
recovery_log.append(f"Re-queued tasks: {requeued}")
|
|
|
|
# 4. Reset circuit breakers if cooldown expired
|
|
if 'circuitBreakers' in state:
|
|
for cb_name, cb in state['circuitBreakers'].items():
|
|
if cb.get('state') == 'open' and cb.get('cooldownUntil'):
|
|
cooldown = datetime.fromisoformat(cb['cooldownUntil'].replace('Z', '+00:00'))
|
|
if datetime.now(cooldown.tzinfo) > cooldown:
|
|
cb['state'] = 'half-open'
|
|
recovery_log.append(f"Circuit breaker {cb_name}: OPEN -> HALF-OPEN")
|
|
|
|
return recovery_log
|
|
|
|
log = recover_system()
|
|
for entry in log:
|
|
print(entry)
|
|
|
|
print("RECOVERY_COMPLETE")
|
|
EOF
|
|
|
|
log_pass "Full system recovery works"
|
|
|
|
echo ""
|
|
echo "========================================"
|
|
echo "Test Summary"
|
|
echo "========================================"
|
|
echo -e "${GREEN}Passed: $PASSED${NC}"
|
|
echo -e "${RED}Failed: $FAILED${NC}"
|
|
echo ""
|
|
|
|
if [ $FAILED -eq 0 ]; then
|
|
echo -e "${GREEN}All tests passed!${NC}"
|
|
exit 0
|
|
else
|
|
echo -e "${RED}Some tests failed!${NC}"
|
|
exit 1
|
|
fi
|