Rounds Daemon

Entry Point: rounds/main.py (composition root)

Command:

cd /home/austinsand/workspace/orchestrator/rounds
RUN_MODE=daemon python -m rounds.main

With Custom Configuration:

# Export environment variables from custom config
export $(cat /path/to/custom.env | xargs)
RUN_MODE=daemon python -m rounds.main

Key Components Activated:

1. Composition Root (rounds/main.py): Wires all adapters and services
2. Poll Service (rounds/core/poll_service.py): Orchestrates polling loop
3. Telemetry Adapter (rounds/adapters/telemetry/): Queries configured backend
4. Fingerprint Service (rounds/core/fingerprint.py): Generates signatures
5. Investigator (rounds/core/investigator.py): Orchestrates diagnosis
6. Store Adapter (rounds/adapters/store/sqlite.py): Persists signatures
7. Diagnosis Adapter (rounds/adapters/diagnosis/claude_code.py): LLM analysis
8. Notification Adapter (rounds/adapters/notification/): Reports findings

Environment Variables (from rounds/config.py):

Core settings:

• RUN_MODE=daemon (required)
• TELEMETRY_BACKEND=signoz|jaeger|grafana_stack
• STORE_BACKEND=sqlite (default)
• DIAGNOSIS_BACKEND=claude_code (default)
• NOTIFICATION_BACKEND=stdout|markdown|github_issue

Polling configuration:

• POLL_INTERVAL_SECONDS=60 (default)
• ERROR_LOOKBACK_MINUTES=5 (default)
• POLL_BATCH_SIZE=100 (default)

Budget controls:

• CLAUDE_CODE_BUDGET_USD=5.00 (per-diagnosis limit)
• DAILY_BUDGET_LIMIT=100.00 (daily spending cap)

Backend-specific:

• SIGNOZ_API_URL, SIGNOZ_API_KEY
• JAEGER_API_URL
• GRAFANA_STACK_URL, GRAFANA_API_KEY
• STORE_SQLITE_PATH=./signatures.db
• NOTIFICATION_OUTPUT_DIR=./diagnoses
• GITHUB_TOKEN, GITHUB_REPO

Examples

Example 1: Start daemon with SigNoz telemetry

# Set up environment
export TELEMETRY_BACKEND=signoz
export SIGNOZ_API_URL=https://signoz.example.com
export SIGNOZ_API_KEY=your-api-key-here
export POLL_INTERVAL_SECONDS=30
export NOTIFICATION_BACKEND=markdown
export NOTIFICATION_OUTPUT_DIR=/var/log/rounds/diagnoses

# Start daemon
cd /home/austinsand/workspace/orchestrator/rounds
RUN_MODE=daemon python -m rounds.main

Expected behavior:

• Polls SigNoz every 30 seconds for errors in the last 5 minutes
• Fingerprints new errors and stores signatures in ./signatures.db
• Triggers Claude Code diagnosis for new signatures
• Writes markdown reports to /var/log/rounds/diagnoses/

Example 2: Start daemon with GitHub issue notifications

# Set up environment
export TELEMETRY_BACKEND=jaeger
export JAEGER_API_URL=http://localhost:16686
export NOTIFICATION_BACKEND=github_issue
export GITHUB_TOKEN=ghp_xxxxxxxxxxxxx
export GITHUB_REPO=myorg/myapp
export DAILY_BUDGET_LIMIT=50.00

# Start daemon
cd /home/austinsand/workspace/orchestrator/rounds
RUN_MODE=daemon python -m rounds.main

Expected behavior:

• Polls Jaeger for errors
• Creates GitHub issues for new diagnoses in myorg/myapp
• Stops diagnosing new errors once daily budget reaches $50

Example 3: Development mode with stdout notifications

# Minimal config for local development
export TELEMETRY_BACKEND=signoz
export SIGNOZ_API_URL=http://localhost:3301
export NOTIFICATION_BACKEND=stdout
export POLL_INTERVAL_SECONDS=10
export ERROR_LOOKBACK_MINUTES=1

# Start daemon
cd /home/austinsand/workspace/orchestrator/rounds
RUN_MODE=daemon python -m rounds.main

Expected behavior:

• Fast polling (every 10 seconds)
• Short lookback window (1 minute)
• Prints diagnoses to stdout for immediate visibility
• Useful for testing telemetry integration

Example 4: Using dotenv file for configuration

# Create .env file
cat > /tmp/rounds.env << EOF
RUN_MODE=daemon
TELEMETRY_BACKEND=grafana_stack
GRAFANA_STACK_URL=https://grafana.example.com
GRAFANA_API_KEY=eyJrIjoiXXXXXX
POLL_INTERVAL_SECONDS=120
STORE_SQLITE_PATH=/var/lib/rounds/signatures.db
NOTIFICATION_BACKEND=markdown
NOTIFICATION_OUTPUT_DIR=/var/log/rounds
DAILY_BUDGET_LIMIT=200.00
EOF

# Load and run
export $(cat /tmp/rounds.env | xargs)
cd /home/austinsand/workspace/orchestrator/rounds
python -m rounds.main

Process Management

Systemd Service Example:

[Unit]
Description=Rounds Error Diagnosis Daemon
After=network.target

[Service]
Type=simple
User=rounds
WorkingDirectory=/home/austinsand/workspace/orchestrator/rounds
EnvironmentFile=/etc/rounds/daemon.env
ExecStart=/usr/bin/python3 -m rounds.main
Restart=on-failure
RestartSec=10

[Install]
WantedBy=multi-user.target

Docker Compose Example: