Backend Observability and Monitoring cookbook

10 recipes · 65 steps · about 67 minutes of reading

Fix a health check that checks nothing
My /health route just says OK. It never checks the database.
BeginnerRead 4 minDoing: about 30 minutes5 steps
Replace print debugging with structured JSON logs
I still find bugs by scrolling through print() output, and once two things run at once I can't tell which line came from which run.
BeginnerRead 5 minDoing: a couple of hours6 steps
Give every request a correlation ID that survives your stack
Two users hit the same bug at once, and I can't tell whose logs are whose.
Beginner to intermediateRead 5 minDoing: a few hours5 steps
Your dashboard is missing the four golden signals
My dashboard is full of CPU and memory graphs. None of them tell me if the API itself is slow or throwing errors right now.
IntermediateRead 6 minDoing: an afternoon6 steps
Put your cache hit rate on the same dashboard as your latency
My cache made things faster once. I have no way to know if it still does.
IntermediateRead 5 minDoing: an afternoon6 steps
Expose consumer lag as a number instead of a guess
A queue that feels backed up, with no real number behind the feeling.
IntermediateRead 7 minDoing: an afternoon7 steps
Add distributed tracing so a slow request explains itself
An endpoint is slow and I genuinely can't tell if it's my code, a slow query, or the network in between.
Intermediate to advancedRead 10 minDoing: a day8 steps
Alert fatigue from paging on causes instead of symptoms
I get paged for a CPU spike that always fixes itself within a minute. By the third false alarm this week, I've stopped trusting the pager at all.
Intermediate to advancedRead 8 minDoing: an afternoon8 steps
Set an error budget and let it decide when to slow down
Ship the next feature, or spend the week on reliability? Nobody on the team has a number to settle it.
Intermediate to advancedRead 8 minDoing: an afternoon8 steps
A single label is multiplying your metric cardinality
I added one label to filter by, and now the monitoring bill has a number nobody recognizes.
AdvancedRead 9 minDoing: an afternoon6 steps

Cookbook badge

Finish all 10 recipes to earn this.

In these roadmaps

Backend Developerstep 13 of 15

Backend Observability and Monitoring cookbook

Fix a health check that checks nothing

Replace print debugging with structured JSON logs

Give every request a correlation ID that survives your stack

Your dashboard is missing the four golden signals

Put your cache hit rate on the same dashboard as your latency

Expose consumer lag as a number instead of a guess

Add distributed tracing so a slow request explains itself

Alert fatigue from paging on causes instead of symptoms

Set an error budget and let it decide when to slow down

A single label is multiplying your metric cardinality

In these roadmaps

Backend Observability and Monitoring cookbook

Fix a health check that checks nothing

Replace print debugging with structured JSON logs

Give every request a correlation ID that survives your stack

Your dashboard is missing the four golden signals

Put your cache hit rate on the same dashboard as your latency

Expose consumer lag as a number instead of a guess

Add distributed tracing so a slow request explains itself

Alert fatigue from paging on causes instead of symptoms

Set an error budget and let it decide when to slow down

A single label is multiplying your metric cardinality

In these roadmaps