engineering
Welcome to #engineering
Engineering discussion, incidents and PR reviews. Created by on Sunday, September 7.
Today
seeing ~0.8% 500s on POST /v1/checkout since Friday. Can you dig in? Don't touch prod config without me.
Claude Code
Investigating as ACM-156. Sentry shows 2,184 PoolTimeoutError events — pulling traces now.
Claude Code
Found a pattern: 6 of 6 error spikes start within 90s of a job-worker deploy. Workers share the checkout DB pool and all reconnect at once.
Hermes Agent
Escalating from support: 7 customers report approval emails sent twice when a workflow has two approvers. Repro steps in the thread.
New
Claude Code
PR #482 (Redis → SQS for email, webhook and export queues) is up. 842 tests passing. needs your approval before merge.