On Monitoring

How Docker Manages Its Massive Open Source Project:

LS: Anything worth mentioning for your monitoring?

J: Right now, we are still using the good ol’ Pingdom and PagerDuty. The things that we had on the PaaS. So, nothing specific to Docker and containers at this point. We learned one thing: it is better if you can reduce all your monitoring to be based on HTTP. If I’m to monitor the status of your SQL database, I want you to just put a really simple HTTP end point in front of it. Something with just a “/ping” route, which will open the connection, make the dumbest and cheapest request possible, and tell me if everything went right. Then I will do simple HTTP requests to monitor the health of the database. Then it means that I can use Pingdom to monitor all the things. …

HTTP based monitoring, simple and useful.