Distributed Systems with Node.js

Chapter 4. Observability

This chapter is dedicated to observing Node.js services that run on remote machines. Locally, tools like the debugger or console.log() make this a straightforward process. However, once a service is running in a faraway land, you’ll need to reach for a different set of tools.

When debugging locally, you’re usually concerned with a single request. You might ask yourself, “When I pass this value into a request, why do I get that value in the response?” By logging the inner workings of a function, you gain insight into why a function behaved in an unanticipated way. This chapter looks at technologies useful for debugging individual requests as well. “Logging with ELK” looks at log generation, which is a way to keep track of information on a per-request basis, much like you might print with console.log(). Later, “Distributed Request Tracing with Zipkin” looks at a tool for tracking requests as they’re passed around, associating related logs generated by different services.

You often need insight into situations that wouldn’t normally be considered a hard bug when dealing with production traffic. For example, you might have to ask, “Why are HTTP requests 100ms slower for users created before April 2020?” Such timing might not be worrying with a single request, but when such metrics are considered in aggregate over many requests, you’re able to spot trends of negative performance. “Metrics with Graphite, StatsD, and Grafana” covers this in more detail.

These ...