Release Management Best Practices | Rollback & Feature Flags
Ship software releases confidently with feature flags, rollback plans, and release checklists. GitScrum tracks releases from development to production.
9 min read
Release management transforms completed development work into production-ready software that reaches users reliably. GitScrum's release tracking features help teams coordinate across milestones, track what's included in each release, and manage the handoff from development to deployment. The key is treating releases as planned events, not last-minute scrambles.
Release Strategies
| Strategy | Risk | Speed | Best For |
|---|---|---|---|
| Big Bang | High | Slow | Rare, major versions |
| Rolling | Medium | Medium | Continuous updates |
| Blue-Green | Low | Fast | Zero-downtime needed |
| Canary | Low | Medium | Risk-averse releases |
| Feature Flags | Low | Fast | Controlled rollouts |
Release Process
RELEASE WORKFLOW
RELEASE PIPELINE:
┌─────────────────────────────────────────────────┐
│ Code Build Test Stage │
│ │ │ │ │ │
│ ▼ ▼ ▼ ▼ │
│ ┌────┐ ┌────┐ ┌────┐ ┌────┐ │
│ │ PR │───▶│ CI │───▶│ QA │──────▶│Stag│ │
│ └────┘ └────┘ └────┘ └────┘ │
│ │ │
│ Approval ───────┤ │
│ ▼ │
│ ┌────┐ │
│ │Prod│ │
│ └────┘ │
│ │ │
│ Monitor + Verify │
└─────────────────────────────────────────────────┘
RELEASE STAGES:
┌─────────────────────────────────────────────────┐
│ 1. Development Complete │
│ ├── All features coded │
│ ├── Unit tests passing │
│ └── Code reviewed and merged │
│ │
│ 2. QA Validation │
│ ├── Integration tests passing │
│ ├── E2E tests passing │
│ └── Exploratory testing complete │
│ │
│ 3. Staging Deployment │
│ ├── Deployed to staging environment │
│ ├── Smoke tests passing │
│ └── Stakeholder sign-off │
│ │
│ 4. Production Release │
│ ├── Deploy to production │
│ ├── Verify deployment success │
│ └── Monitor for issues │
│ │
│ 5. Post-Release │
│ ├── Communicate release notes │
│ ├── Monitor metrics │
│ └── Respond to issues │
└─────────────────────────────────────────────────┘
Release Checklist
PRE-RELEASE CHECKLIST
BEFORE RELEASE:
┌─────────────────────────────────────────────────┐
│ Code & Tests: │
│ ☐ All code merged to release branch │
│ ☐ All tests passing (unit, integration, e2e) │
│ ☐ Code coverage meets threshold │
│ ☐ No critical security vulnerabilities │
│ ☐ Performance benchmarks acceptable │
│ │
│ Documentation: │
│ ☐ Release notes written │
│ ☐ API documentation updated │
│ ☐ Runbook updated if needed │
│ ☐ Known issues documented │
│ │
│ Stakeholders: │
│ ☐ Product Owner sign-off │
│ ☐ QA sign-off │
│ ☐ Support team notified │
│ ☐ Customer success team notified │
│ │
│ Infrastructure: │
│ ☐ Database migrations tested │
│ ☐ Feature flags configured │
│ ☐ Rollback plan documented │
│ ☐ On-call engineer identified │
└─────────────────────────────────────────────────┘
RELEASE DAY CHECKLIST:
┌─────────────────────────────────────────────────┐
│ ☐ Announce release window to team │
│ ☐ Verify staging is healthy │
│ ☐ Run final smoke tests │
│ ☐ Take production backup (if applicable) │
│ ☐ Deploy to production │
│ ☐ Run production smoke tests │
│ ☐ Verify key metrics are normal │
│ ☐ Monitor error rates for 15 minutes │
│ ☐ Announce release complete │
│ ☐ Update release tracking │
└─────────────────────────────────────────────────┘
Feature Flags
FEATURE FLAG STRATEGY
FEATURE FLAG TYPES:
┌─────────────────────────────────────────────────┐
│ Release Flag: │
│ ├── Purpose: Hide incomplete features │
│ ├── Lifespan: Until feature complete │
│ └── Example: new_dashboard_enabled │
│ │
│ Experiment Flag: │
│ ├── Purpose: A/B testing │
│ ├── Lifespan: Duration of experiment │
│ └── Example: checkout_v2_experiment │
│ │
│ Ops Flag: │
│ ├── Purpose: Control system behavior │
│ ├── Lifespan: Permanent │
│ └── Example: enable_rate_limiting │
│ │
│ Permission Flag: │
│ ├── Purpose: Feature access by plan/role │
│ ├── Lifespan: Permanent │
│ └── Example: advanced_analytics_enabled │
└─────────────────────────────────────────────────┘
ROLLOUT STRATEGY:
┌─────────────────────────────────────────────────┐
│ Phase 1: Internal (0-5%) │
│ ├── Enable for internal team │
│ ├── Duration: 2-3 days │
│ └── Monitor: All metrics │
│ │
│ Phase 2: Beta (5-20%) │
│ ├── Enable for beta users │
│ ├── Duration: 1 week │
│ └── Monitor: Error rates, feedback │
│ │
│ Phase 3: Gradual (20-100%) │
│ ├── Increase 10-20% per day │
│ ├── Duration: 1-2 weeks │
│ └── Monitor: Performance, errors │
│ │
│ Phase 4: General Availability │
│ ├── Enable for all users │
│ ├── Remove feature flag (cleanup) │
│ └── Archive flag documentation │
└─────────────────────────────────────────────────┘
Rollback Planning
ROLLBACK PROCEDURES
WHEN TO ROLLBACK:
┌─────────────────────────────────────────────────┐
│ Automatic rollback triggers: │
│ ├── Error rate > 5% (up from baseline 0.1%) │
│ ├── Latency > 3x baseline │
│ ├── Health checks failing │
│ └── Key business metric drop > 10% │
│ │
│ Manual rollback decision: │
│ ├── Customer-reported critical bugs │
│ ├── Data integrity issues │
│ └── Security vulnerabilities discovered │
└─────────────────────────────────────────────────┘
ROLLBACK STEPS:
┌─────────────────────────────────────────────────┐
│ 1. Confirm rollback decision │
│ └── On-call engineer + manager approval │
│ │
│ 2. Communicate rollback starting │
│ └── #incidents channel, stakeholders │
│ │
│ 3. Execute rollback │
│ ├── For code: Deploy previous version │
│ ├── For flags: Disable feature flag │
│ └── For data: Run reversal script │
│ │
│ 4. Verify rollback success │
│ ├── Health checks passing │
│ ├── Error rates normalized │
│ └── Key metrics stabilized │
│ │
│ 5. Communicate rollback complete │
│ └── Update status, notify stakeholders │
│ │
│ 6. Post-incident │
│ ├── Document what happened │
│ ├── Root cause analysis │
│ └── Plan for re-release │
└─────────────────────────────────────────────────┘
Release Communication
RELEASE COMMUNICATION PLAN
INTERNAL COMMUNICATION:
┌─────────────────────────────────────────────────┐
│ Pre-Release (1 week before): │
│ ├── Release contents summary │
│ ├── Timeline and deployment window │
│ └── Known risks and mitigation │
│ │
│ Release Day: │
│ ├── "Deploying now" announcement │
│ ├── "Deploy complete, monitoring" update │
│ └── "Release successful" confirmation │
│ │
│ Post-Release: │
│ └── Release notes distributed │
└─────────────────────────────────────────────────┘
EXTERNAL COMMUNICATION:
┌─────────────────────────────────────────────────┐
│ For Significant Releases: │
│ │
│ Customer email: │
│ ├── What's new (user benefits) │
│ ├── How to use new features │
│ └── Where to get help │
│ │
│ Changelog/Release notes: │
│ ├── New features │
│ ├── Improvements │
│ ├── Bug fixes │
│ └── Breaking changes (if any) │
│ │
│ Documentation updates: │
│ ├── New feature documentation │
│ ├── Updated screenshots │
│ └── API documentation │
└─────────────────────────────────────────────────┘
RELEASE NOTES FORMAT:
┌─────────────────────────────────────────────────┐
│ Version 2.5.0 - March 15, 2025 │
│ │
│ ✨ New Features: │
│ • Dashboard: Real-time analytics view │
│ • Reports: Custom date range selection │
│ • API: Bulk operations endpoint │
│ │
│ 🔧 Improvements: │
│ • Search: 50% faster results │
│ • Export: Support for CSV and Excel │
│ │
│ 🐛 Bug Fixes: │
│ • Fixed: Timezone display issue │
│ • Fixed: Email notification delays │
│ │
│ ⚠️ Breaking Changes: │
│ • API v1 deprecated (sunset: June 2025) │
└─────────────────────────────────────────────────┘
Hotfix Process
HOTFIX WORKFLOW
HOTFIX CRITERIA:
┌─────────────────────────────────────────────────┐
│ Qualifies as hotfix: │
│ ├── Production outage │
│ ├── Security vulnerability │
│ ├── Data corruption risk │
│ └── Major feature completely broken │
│ │
│ Does NOT qualify: │
│ ├── Minor bugs │
│ ├── Feature requests │
│ ├── Performance improvements │
│ └── UI polish │
└─────────────────────────────────────────────────┘
HOTFIX PROCESS:
┌─────────────────────────────────────────────────┐
│ 1. Triage (15 min) │
│ ├── Confirm severity justifies hotfix │
│ ├── Identify owner │
│ └── Communicate to stakeholders │
│ │
│ 2. Development (1-4 hours) │
│ ├── Branch from production tag │
│ ├── Minimal fix only (no extras) │
│ ├── Write test covering the issue │
│ └── Fast-track code review │
│ │
│ 3. Testing (30-60 min) │
│ ├── Verify fix works │
│ ├── Regression test affected area │
│ └── Deploy to staging, smoke test │
│ │
│ 4. Deploy (30 min) │
│ ├── Deploy to production │
│ ├── Verify fix in production │
│ └── Monitor for 15 minutes │
│ │
│ 5. Cleanup │
│ ├── Merge hotfix to main branch │
│ ├── Document in incident log │
│ └── Schedule proper fix if needed │
└─────────────────────────────────────────────────┘
Best Practices
Anti-Patterns
✗ Manual deployments with no script
✗ No rollback capability
✗ "Big bang" releases with months of changes
✗ No monitoring after deployment
✗ Hotfixes without code review
✗ No communication to stakeholders