Considerations for monitoring system health

One of the things a system administrator learns early in his or her career is that it is better to find out about problems and fix them before the customer or the boss notices them. Log collection and performance graphs are essential when it comes to troubleshooting problems. Rarely, however, is someone watching every log message and every piece of performance data. There is just too much information and some administrators even have lives outside of the office. Fortunately, computers are very good at sifting through lots of data and rarely ask for time off.

Most of the tools that have been used for years to monitor server and service health still apply. However, a fully orchestrated environment offers some unique challenges when compared to the old standalone server model. For example, a service does not live on a single host. It could run anywhere in the cluster. In addition, the ability to quickly add or remove capacity to the Docker cluster...