Chrome pool issues

Chrome not starting

Symptoms

• failed to start chrome in logs
• Render Service fails to start

Causes

• Chrome not installed
• Wrong chrome.path in config
• Missing dependencies (Linux)
• Sandbox issues

Solutions

• Verify Chrome installation: google-chrome --version
• Check chrome.path configuration
• Install missing libraries (see Linux dependencies)
• Try --no-sandbox flag (not recommended for production)

Pool exhaustion

Symptoms

• no available chrome instances errors
• Requests queuing or timing out
• High latency spikes

Causes

• Pool size too small for traffic
• Renders taking too long
• Chrome instances not releasing

Solutions

• Increase chrome.pool_size
• Reduce render timeouts
• Check for stuck renders in logs
• Monitor chrome_pool_active metric

Memory issues

Symptoms

• OOM kills in system logs
• Chrome processes consuming excessive RAM
• System becoming unresponsive

Causes

• Pool size too large for available memory
• Memory leaks in long-running instances
• Heavy pages accumulating memory

Solutions

• Reduce chrome.pool_size
• Lower chrome.max_requests_per_instance
• Enable chrome.restart_on_memory_mb threshold
• Add swap space (temporary)

Instance restart problems

Symptoms

• Chrome instances cycling frequently
• Restart loops in logs

Causes

• max_requests_per_instance too low
• Chrome crashing on certain pages
• Resource limits too aggressive

Solutions

• Increase max_requests_per_instance
• Check for problematic URLs causing crashes
• Review Chrome flags configuration

Render Service registration failures

Symptoms

• failed to register in service registry
• Edge Gateway reports no render services

Causes

• Redis connectivity issues
• Duplicate service IDs
• Heartbeat failures

Solutions

• Verify Redis connection from Render Service
• Ensure unique server.id per instance
• Check heartbeat interval and timeout settings