The dashboard nobody looked at, and the alert that fired every day
An alert that fires daily has trained everyone to ignore it, which is worse than no alert.
practices
46 posts
Your error message is an API, and changing it breaks someone
Somebody is grepping your log line, matching on your message, or alerting on its text.
The test that broke on every refactor was testing the wrong thing
A test coupled to call order and internal structure fails when nothing is broken.
The migration that could not be rolled back, and nobody noticed until it mattered
A column drop is irreversible the moment it runs. Expand-contract exists because deploys and migrations do not move together.
The retry that turned a blip into an outage
Synchronised retries from every client are a self-inflicted thundering herd. Jitter is the mechanism, not a detail.
We shared 60% of the code and one platform used all of it
Measured per target rather than per line: the iOS app reached a third of the shared surface because the API was shaped around Android's lifecycle.
The third time this filled up, I diagnosed it as an outage
A consumable with no recorded lifetime is indistinguishable from a service that is down.
I applied my fix to 234 files at once. It broke 233 of them.
A mechanical sweep is one hypothesis executed N times. If the hypothesis is wrong, so is every edit.
Most code review approves code. Here is review that improves it
Line-level comments are the least valuable thing in a review. Three questions outrank all of them.
I created three duplicate posts because the only record was a file I didn't read
When an API has no list endpoint, your local record is not a convenience. It is the only state there is.