🔧 DevOps

Rollback Strategies

Roll back failed deployments safely.

Kubernetes Rollback

# Instant rollback to previous revision
kubectl rollout undo deployment/gauzy-api

# Rollback to specific revision
kubectl rollout history deployment/gauzy-api
kubectl rollout undo deployment/gauzy-api --to-revision=3

# Check rollout status
kubectl rollout status deployment/gauzy-api

Docker Rollback

# Run previous image tag
docker service update --image ghcr.io/ever-co/gauzy-api:v1.2.3 gauzy-api

# Docker Compose - revert to previous tag
docker compose pull
docker compose up -d

Database Rollback

If the release included migrations:

# TypeORM
yarn typeorm migration:revert

# MikroORM
npx mikro-orm migration:down

Rollback Decision Tree

Diagram; its source is available below
Diagram source
graph TB
    Issue[Issue Detected] --> Severity{Severity?}
    Severity -->|Critical| Immediate[Immediate Rollback]
    Severity -->|Major| Assess[Assess Impact]
    Severity -->|Minor| Hotfix[Apply Hotfix]
    Assess -->|Wide impact| Immediate
    Assess -->|Limited| Hotfix
    Immediate --> DBCheck{DB Migrations?}
    DBCheck -->|Yes| MigRevert[Revert Migrations]
    DBCheck -->|No| ImageRevert[Revert Image Only]
    MigRevert --> Verify[Verify Health]
    ImageRevert --> Verify

Pre-Rollback Checklist

  • Confirm the issue is deployment-related
  • Identify the last known good version
  • Check if data migrations need reverting
  • Notify team and stakeholders
  • Have monitoring dashboards open

Post-Rollback

  • Verify service health
  • Check error rates returning to baseline
  • Document the rollback
  • Root cause analysis
  • Fix forward with new release

Edit this page

Copyright © 2023-Present Ever Co. LTD.Docs by Sourcey
Stack OverflowWebsiteDemo
↑↓ navigate↵ selectesc close