Skip to content
~/blogwriting

Notes on distributed systems and the things that break in them

Mostly written after an incident, a migration, or an argument with a consumer group.

2 min read

Idempotency Is a Design Decision, Not a Retry Policy

Retries are inevitable in distributed systems. The question is whether your consumers were built to survive them.

kafkadistributed-systemsnode.js
2 min read

Long-Running Chain Listeners That Actually Stay Up

What it takes to keep a blockchain event listener alive across RPC failures, reorgs, and silent stalls.

blockchainethers.jsreliability
2 min read

Running Kafka on KRaft: What Changed and What Didn't

Notes from standing up a ZooKeeper-free Kafka cluster and keeping it under 100ms.

kafkainfrastructureobservability