Engineering / Operations / Release
Production Deployment Guide
On this page
- Overview
- Prerequisites
- Deployment Steps
- Verification Checklist
- Rollback
Overview
This guide describes the standard procedure for deploying application releases to the production environment. All deployments must follow the change management process and require approval from the release manager before promotion beyond staging.
Note: Maintenance windows are scheduled every Tuesday and Thursday between 02:00–04:00 UTC. Emergency deployments require VP Engineering approval and a paired SRE on-call.
Prerequisites
Before initiating a deployment, confirm the following conditions are met:
- All CI pipeline checks pass on the release branch
- QA sign-off documented in Jira ticket
REL-2026-0847 - Database migration scripts reviewed and validated in staging
- Rollback plan documented and timed (target < 15 minutes)
- Grafana Infrastructure Overview dashboard open and healthy
Deployment Steps
Freeze & announcePost in #releases and set change freeze for dependent services.
Trigger pipelineRun GitLab job
deploy/production from tag v2.4.1.Watch canaryConfirm error rate < 0.5% and p99 latency within SLO for 10 minutes.
Full rolloutPromote remaining replicas. Update ServiceNow change to Implemented.
Caution: Do not run schema migrations concurrently with blue/green cutover. Complete DDL first, then shift traffic.
Verification Checklist
Validate authentication flows, webhook delivery, and critical read/write paths. Confirm Splunk dashboards show no new ERROR spikes related to platform-api.
Rollback
If canary metrics breach thresholds, abort the pipeline and redeploy the previous known-good tag. Escalate via ServiceNow P1 if customer impact exceeds 5 minutes.