Sre

The core trick that cut our PR CI time from 22 minutes to 4 on average.

A six-month experience report. The value proposition is real; the downsides are too.

Enough formulas to write real alerts without spending a weekend in a textbook.

Structured logging is not the hard part. The hard part is everything around it.

I conflated these for years. Here is a cleaner way to think about them.

A service worked on one image and not another. The difference was invisible until we traced syscalls.

A core dump from production is a gift. Here is how I unwrap it.

Dev containers solve real problems but have their own operational tail.

noeviction is the default, and the default is dangerous when you thought you were running a cache.

A retrospective on how our team finally beat alert fatigue.

Global autovacuum settings are a lie. I tune per-table now.

A forgotten logical replication slot accumulated 380GB of WAL before we caught it. Here's what we changed.

One innocuous request_id label, 18M active series, and a very bad Friday.

Head sampling is simple. Tail sampling works. Here is a config we run in production without sadness.

A multi-day outage-adjacent incident caused by prepared statements not making it across pool boundaries.