fluxcd/flagger
Flagger
Flagger este un operator de livrare progresivă pentru Kubernetes care automatizează lansarea graduală a noilor versiuni software. Acesta funcționează ca un manager de lansări de tip canary și un sistem de direcționare a traficului, integrându-se cu service mesh-uri și ingress controllere pentru a gestiona distribuția cererilor între versiunile aplicației.
Proiectul se concentrează pe reducerea riscurilor în producție prin implementarea lansărilor canary, a deployment-urilor blue-green și a testării A/B. Utilizează o buclă de analiză bazată pe metrici pentru a monitoriza rata de succes și durata cererilor, promovând automat o lansare sau declanșând un rollback bazat pe starea sistemului dacă criteriile de sănătate nu sunt îndeplinite.
Instrumentul oferă validare suplimentară prin execuția de teste bazate pe webhook-uri și testare sintetică de sarcină pentru a verifica stabilitatea. Include, de asemenea, mecanisme pentru porți de lansare manuală, sincronizarea configurațiilor și notificări de evenimente către furnizori externi de comunicare.
Controller-ul este instalat în cluster folosind manageri de pachete standard, cu configurații pentru service mesh-urile și ingress controllerele suportate.
Features
- Progressive Delivery Strategies - Automates progressive delivery strategies, including canary and blue-green updates, to reduce production risk.
- A/B Testing - Performs A/B testing by splitting traffic between different versions to compare performance and user behavior.
- Blue-Green Deployments - Executes blue-green deployments by routing traffic between two identical environments to ensure version stability.
- Change Rollback Mechanisms - Automatically reverts traffic to the stable version when health metrics fail to meet required thresholds.
- Custom Resource Controllers - Provides a custom resource controller that monitors the Kubernetes API to automate progressive delivery workflows.
- Automated Rollout Managers - Manages application updates through incremental rollouts with automated rollback capabilities based on metric analysis.
- Deployment Strategies - Implements varied release patterns using weighted traffic, header-based routing, and traffic mirroring.
- Deployment Strategy Mappings - Implements a declarative system to map configuration files to specific deployment patterns like canary or blue-green.
- Automated Canary Analysis - Automates canary analysis by comparing real-time metrics against a stable baseline to decide on promotion.
- Automated Metric Analysis - Continuously compares real-time performance metrics between stable and canary versions to automate release decisions.
- Canary Release Managers - Acts as a canary release manager by gradually routing traffic and monitoring health to prevent outages.
- Kubernetes Operators - Implements the Kubernetes operator pattern using custom resources to automate progressive delivery workflows.
- Kubernetes Traffic Controllers - Integrates with service meshes and ingress controllers to automate network traffic routing within Kubernetes.
- Automated Release Validation - Provides automated promotion and rollback based on webhook-driven tests and real-time health metrics.
- Weighted Traffic Splitting - Updates service mesh and ingress configurations to incrementally shift request percentages between application versions.
- Release Health Monitoring - Monitors request success rates and durations to automatically promote or roll back releases based on health.
- Automated Test Execution - Executes automated test suites via webhooks during the rollout to verify delivery process stability.
- Release Quality Validation - Triggers external acceptance and load tests to ensure new versions meet established quality standards before full promotion.
- Workflow Approval Gates - Provides manual approval steps to pause or resume the promotion process for human oversight.
- Webhook Quality Gates - Triggers external HTTP endpoints to execute acceptance or load tests as quality gates during the rollout process.
- Rollout Approval Gates - Enables manual oversight of a release through a pause-and-resume approval mechanism for workload promotion.
- Synthetic Traffic Generators - Generates synthetic traffic against application versions to verify stability and performance during rollout.
- Kubernetes Ecosystem - Automates progressive delivery and canary deployments.
