Observability · Maintainer and contributor
Monti APM
Production observability across methods, database work, HTTP calls, background jobs, and asynchronous execution.
Observation
Application performance monitoring must follow work as it crosses methods, database operations, remote calls, queues, and asynchronous boundaries. Meteor 3’s move away from fibers made that context propagation problem especially important.
Approach
For around a year, I contributed across the agent’s instrumentation stack and helped migrate it to Meteor 3 alpha releases. The work covered method and publication timeouts, oplog diagnostics, HTTP and email behavior, background jobs, Redis Oplog, error handling, and TypeScript definitions.
Systems study
The Meteor 3 effort replaced fibers-based assumptions with Async Hooks and AsyncLocalStorage-compatible behavior. The main support pull request crossed 97 files and was accompanied by related changes in the agent’s core packages.
Significance
Observability is useful only when it preserves the causal story of a request. This work made newer asynchronous applications visible without discarding the operational vocabulary existing Monti APM users relied on.