Skip to content

gitops-drift.md

Latest commit

聽

History

History
36 lines (26 loc) 路 1.37 KB

gitops-drift.md

File metadata and controls

36 lines (26 loc) 路 1.37 KB
Raw

GitOps Configuration Drift Management

Overview

This document describes the drift detection system that monitors for configuration drift between Git-declared manifests and the live Kubernetes cluster state.

Detection Strategy

The drift reporter (infrastructure/scripts/drift-reporter.sh) performs a weekly comparison:

  1. Reads all Kubernetes manifests from infrastructure/k8s/
  2. Applies each manifest with --dry-run=server to get the desired state
  3. Compares against the live cluster state via kubectl get
  4. Generates a JSON report with categorized drifts

Severity Classification

Severity Resource Types Action
Critical StatefulSet, DaemonSet Immediate alert, auto-remediation
High Deployment PR required, auto-issue created
Medium ConfigMap, Secret, Service, Ingress Weekly review
Low HPA, PDB, others Logged in report

Response Procedures

  1. Critical drift: Pager alert triggered, on-call investigates
  2. High drift: GitHub issue auto-created, assigned to infrastructure team
  3. Medium drift: Reviewed in weekly infrastructure sync
  4. Low drift: Noted in trend analysis

Remediation

To remediate drift:

  1. Update the Git manifests to match desired state
  2. Or apply the live changes back to Git
  3. Run bash infrastructure/scripts/drift-reporter.sh locally to verify