Skip to main content

Best Practices

Node​

  • monitor disk space usage, especially the database size - it can grow quickly, especially around network upgrades
  • monitor the memory usage - the node can use a lot of memory, especially during sync. Don't let it run too close to the limit
  • monitor the chain sync status - on average, the node should be able to sync two epochs per minute
  • monitor the number of peers - the more peers, the better, If a node has no peers, it cannot sync
  • monitor the logs for errors and warnings - they can indicate potential issues

Monitoring​

  • monitor the monitoring system - if the monitoring system goes down, you won't know if the node is down
  • set up alerts for critical metrics - disk space, memory usage, sync status, etc.
  • ensure the persistence of the monitoring system - if the monitoring system loses data, you won't be able to diagnose issues
  • don't expose monitoring endpoints to the internet - they are not secured and can leak sensitive information
  • don't set the log levels too high - this can generate a lot of data and slow down the node