Production Deployment Runbook
Problem
Section titled “Problem”Ship a change to murmur-web (website, admin UI, and the Go API) to
production without downtime.
Severity
Section titled “Severity”N/A — this is a routine operational runbook, not an incident response one.
Symptoms
Section titled “Symptoms”N/A.
Immediate Checks
Section titled “Immediate Checks”Confirm CI is green on the branch being deployed
All required checks pass on the commit you’re about to deploy — don’t deploy a red commit “to see.”
Confirm no one else is mid-deploy
Check the deploy channel/lock before starting.
Mitigation
Section titled “Mitigation”Start the cloud deploy profile
Trigger the Deploy workflow for the target commit (see
murmur-web/docs/DEPLOY.md for the exact trigger).
Wait for the new profile to report healthy
Don’t cut traffic over until health checks pass on the new instances.
Cut nginx over
cutover-live.conf routes /api/v1, /api/auth, and related paths to the
new instance.
Validation
Section titled “Validation”Smoke test the API
Hit GET /api/v1/catalog and confirm a 200 with the expected product list.
Watch error-rate telemetry for 15 minutes
See Patch API for what “healthy” looks like.
Escalation
Section titled “Escalation”If health checks don’t pass or error rate spikes after cutover, go straight to the Rollback Runbook rather than debugging live in production.
Follow-Up
Section titled “Follow-Up”If this deploy caused a user-visible incident, open an incident doc (see INC-0001 for the shape) and eventually a postmortem.