How to monitor MongoDB
10 likes41,888 views
The document discusses monitoring tools and strategies for MongoDB, detailing key metrics such as oplog replication lag, replica state, lock percentage, and disk I/O utilization. It highlights potential issues that can cause replication lag and emphasizes the importance of monitoring non-critical metrics. Various monitoring tools and resources for setting up effective monitoring are also provided.
How to monitor MongoDB
- 1. How to monitor: MongoDB David Mytton Hangout on Air - Sept 2014 https://blog.serverdensity.com/monitor-mongodb/
- 2. David Mytton
- 5. Server Density Architecture ● ~100 servers - Ubuntu 12.04
- 6. Server Density Architecture ● ~100 servers - Ubuntu 12.04 ● 50:50 virtual/dedicated
- 7. Server Density Architecture ● ~100 servers - Ubuntu 12.04 ● 50:50 virtual/dedicated ● 200TB/m processed data
- 8. Server Density Architecture ● ~100 servers - Ubuntu 12.04 ● 50:50 virtual/dedicated ● 200TB/m processed data ● Nginx, Python, MongoDB
- 9. Server Density Architecture ● ~100 servers - Ubuntu 12.04 ● 50:50 virtual/dedicated ● 200TB/m processed data ● Nginx, Python, MongoDB ● Softlayer > 1TB RAM, 5TB SSDs
- 10. Key metrics ● Oplog replication lag ● Replica state ● Lock % ● Disk i/o % utilization
- 11. Oplog replication lag ● Replica sets: master/slave
- 12. Oplog replication lag ● Replica sets: master/slave ● Async i.e. eventually consistent
- 13. Oplog replication lag ● Replica sets: master/slave ● Async i.e. eventually consistent ● Write concern
- 15. Oplog replication lag https://blog.serverdensity.com/mongodb-benchmarks/
- 16. Oplog replication lag ● Replica sets: master/slave ● Async i.e. eventually consistent ● Write concern ● Falling behind
- 17. Reasons for repl falling behind ● Network problems
- 18. Reasons for repl falling behind ● Network problems ● Hardware problems
- 19. Reasons for repl falling behind ● Network problems ● Hardware problems ● Shard chunk migrations
- 20. Reasons for repl falling behind ● Network problems ● Hardware problems ● Shard chunk migrations ● MongoDB bugs
- 21. Replica state ● Primary / secondary
- 22. Replica state ● Primary / secondary ● Alert on state change
- 23. Lock % ● Database locking (2.6)
- 24. Lock % ● Database locking (2.6) ● Sometimes a problem:
- 25. Lock % ● Database locking (2.6) ● Sometimes a problem: ● Nearing 100%
- 26. Lock % ● Database locking (2.6) ● Sometimes a problem: ● Nearing 100% ● Constantly high
- 27. Lock % ● Database locking (2.6) ● Sometimes a problem: ● Nearing 100% ● Constantly high ● Slows replication
- 28. Disk i/o % utilization ● Hardware limits
- 29. Disk i/o % utilization ● Hardware limits ● Nearing 100%
- 30. Disk i/o % utilization ● Hardware limits ● Nearing 100% ● Constantly high
- 31. Disk i/o % utilization ● Hardware limits ● Nearing 100% ● Constantly high ● Spinning -> SSD
- 32. Disk i/o % utilization https://blog.serverdensity.com/mongodb-performance-ssds-vs-spindle-sas-drives/
- 33. Disk i/o % utilization https://blog.serverdensity.com/mongodb-benchmarks/
- 34. Disk i/o % utilization ● Hardware limits ● Nearing 100% ● Constantly high ● Spinning -> SSD ● Slow queries, hangs, slow repl
- 35. Non-critical metrics to watch ● Memory usage
- 36. Non-critical metrics to watch ● Memory usage ● Page faults
- 37. Non-critical metrics to watch ● Memory usage ● Page faults ● Connections
- 38. Non-critical metrics to watch ● Memory usage ● Page faults ● Connections ● Shard chunk distribution
- 39. Non-critical metrics to watch
- 40. Monitoring tools ● mongostat ● mongotop ● rs.status() ● sh.status()
- 41. rs.status()
- 42. sh.status()
- 43. Server Density
- 44. MMS
- 45. Summary ● Critical alerts on key metrics
- 46. Key metrics ● Oplog replication lag ● Replica state ● Lock % ● Disk i/o % utilization
- 47. Summary ● Critical alerts on key metrics
- 48. Summary ● Critical alerts on key metrics ● Watch non-critical
- 49. Summary ● Critical alerts on key metrics ● Watch non-critical ● Manual tools for real time
- 50. Summary ● Critical alerts on key metrics ● Watch non-critical ● Manual tools for real time ● Set up a monitoring product
- 51. Useful resources ● http://docs.mongodb.org/manual/administration/monitoring/ ● https://blog.serverdensity.com/monitor-mongodb ● https://blog.serverdensity.com
- 52. どもありがとうございます ＠davidmytton david@serverdensity.com blog.serverdensity.com