ArgoCDReview

ArgoCDReview - ArgoCD CLI 审查与故障排查

用于审查应用、诊断问题和管理部署的全面 ArgoCD CLI 命令。

此技能提供一系列结构化命令，用于检查 ArgoCD 应用状态、比较实时清单与 Git 中的清单、查看部署历史，以及排查同步失败问题。

---

前提条件

• 已安装 ArgoCD CLI（argocd）
• 已登录 ArgoCD 服务器：argocd login <server>
• 如果使用多个集群，请设置上下文

# Verify connection
argocd version
argocd account get-user-info

---

快速参考命令

应用状态概览

# List all applications with status
argocd app list

# Get detailed app status (health, sync, resources)
argocd app get <app-name>

# Get app status as JSON (for scripting)
argocd app get <app-name> -o json

# Wide output with more columns
argocd app list -o wide

健康与同步状态

# Check sync status only
argocd app get <app-name> --show-operation

# List all OutOfSync applications
argocd app list --sync-status OutOfSync

# List all Degraded applications
argocd app list --health-status Degraded

# List applications by project
argocd app list --project <project-name>

---

工作流路由

工作流	触发词	文件
ReviewApplication	“review app”、“check app status”、“app health”	Workflows/ReviewApplication.md
DiffManifests	“diff app”、“compare live vs desired”、“what changed”	Workflows/DiffManifests.md
TroubleshootSync	“sync failed”、“why not syncing”、“sync errors”	Workflows/TroubleshootSync.md
RollbackApp	“rollback”、“revert deployment”、“previous version”	Workflows/RollbackApp.md
ReviewResources	“list resources”、“app resources”、“managed resources”	Workflows/ReviewResources.md

---

核心审查操作

1. 应用差异（实时状态与期望状态）

比较当前部署的内容与 Git 中声明的期望部署内容：

# Show diff between live state and Git
argocd app diff <app-name>

# Show diff with local manifests
argocd app diff <app-name> --local <repo-path>

# Server-side manifest generation (recommended for Helm/Kustomize)
argocd app diff <app-name> --local <repo-root> --server-side-generate

2. 查看应用清单

# Show desired manifests (from Git)
argocd app manifests <app-name>

# Show live manifests (from cluster)
argocd app manifests <app-name> --source live

# Get specific resource manifest
argocd app get-resource <app-name> \
  --kind Deployment \
  --name <deployment-name> \
  --namespace <namespace>

3. 部署历史

# View deployment history
argocd app history <app-name>

# Output example:
# ID  DATE                           REVISION
# 0   2025-01-10 10:00:00 +0000 UTC  abc1234 (HEAD)
# 1   2025-01-09 15:30:00 +0000 UTC  def5678 (v1.2.0)
# 2   2025-01-08 09:00:00 +0000 UTC  789abcd (v1.1.0)

4. 资源树与详情

# Show resource tree
argocd app resources <app-name>

# Show resource tree with health status
argocd app resources <app-name> --tree

# Get specific resource details
argocd app get-resource <app-name> \
  --resource-name <resource> \
  --kind <kind>

---

同步操作

安全同步（先预览）

# Preview what would sync (dry-run)
argocd app sync <app-name> --dry-run

# Sync single application
argocd app sync <app-name>

# Sync and wait for completion
argocd app sync <app-name> --wait

# Sync specific resources only
argocd app sync <app-name> --resource :Service:my-service
argocd app sync <app-name> --resource apps:Deployment:my-deployment

选择性同步

# Sync by label selector
argocd app sync -l app.kubernetes.io/instance=my-app

# Sync multiple apps
argocd app sync app1 app2 app3

# Sync excluding specific resources
argocd app sync <app-name> --resource '!apps:Deployment:skip-this'

强制同步（谨慎使用）

# Force sync (replaces resources)
argocd app sync <app-name> --force

# Force sync with prune (deletes removed resources)
argocd app sync <app-name> --prune --force

# Sync with replace strategy
argocd app sync <app-name> --replace

---

回滚操作

# View history first
argocd app history <app-name>

# Rollback to specific history ID
argocd app rollback <app-name> <history-id>

# Rollback without confirmation
argocd app rollback <app-name> <history-id> --yes

# Rollback and wait for completion
argocd app rollback <app-name> <history-id> --timeout 300

---

故障排查命令

同步失败诊断

# Get detailed sync operation status
argocd app get <app-name> --show-operation

# View sync operation result
argocd app get <app-name> -o json | jq '.status.operationState'

# Check for resource errors
argocd app resources <app-name> --tree | grep -E "(Degraded|Progressing|Missing)"

应用日志

# View logs for app's pods
argocd app logs <app-name>

# Follow logs
argocd app logs <app-name> -f

# Logs for specific container
argocd app logs <app-name> \
  --container <container-name>

# Logs for specific group/kind
argocd app logs <app-name> \
  --kind Deployment \
  --name <deployment-name>

资源事件

# Get Kubernetes events for app resources
kubectl get events -n <namespace> --sort-by='.lastTimestamp' | grep <app-name>

---

常见诊断场景

场景 1：应用卡在“Progressing”状态

# 1. Check what's progressing
argocd app resources <app-name> --tree

# 2. Find stuck resources (usually Deployments)
argocd app get <app-name> -o json | jq '.status.resources[] | select(.health.status == "Progressing")'

# 3. Check pod events
kubectl describe deployment <deployment-name> -n <namespace>
kubectl get pods -n <namespace> | grep <app-name>

场景 2：应用显示“OutOfSync”，但无法同步

# 1. Check diff
argocd app diff <app-name>

# 2. Check for ignored differences
argocd app get <app-name> -o json | jq '.spec.ignoreDifferences'

# 3. Check sync policy
argocd app get <app-name> -o json | jq '.spec.syncPolicy'

# 4. Force refresh from Git
argocd app get <app-name> --refresh

场景 3：健康状态为“Degraded”

# 1. Find unhealthy resources
argocd app resources <app-name> --tree

# 2. Get resource details
argocd app get-resource <app-name> \
  --kind <kind> \
  --name <name>

# 3. Check pod status
kubectl get pods -n <namespace> -l app.kubernetes.io/instance=<app-name>
kubectl describe pod <pod-name> -n <namespace>

---

多源应用

对于具有多个源的应用：

# Sync with specific source revisions
argocd app sync <app-name> \
  --revisions v1.0.0 --source-positions 1 \
  --revisions v2.0.0 --source-positions 2

# Sync by source name
argocd app sync <app-name> \
  --revisions v1.0.0 --source-names helm-chart \
  --revisions main --source-names values-repo

---

最佳实践检查清单

• 同步前务必运行 argocd app diff
• 在生产环境中执行同步操作时使用 --dry-run
• 回滚前检查部署历史
• 排查健康状况时查看资源树
• 使用 JSON 输出进行脚本处理：-o json
• 为同步操作设置适当的超时时间

---

示例

示例 1：审查故障应用

User: "Why is my-app not healthy?"
→ Run argocd app get my-app --show-operation
→ Check argocd app resources my-app --tree
→ Identify Degraded resources
→ Investigate with kubectl describe

示例 2：比较实时状态与期望状态

User: "What changed in production?"
→ Run argocd app diff my-app
→ Shows line-by-line diff
→ Identify configuration drift

示例 3：部署不当后回滚

User: "Rollback my-app to previous version"
→ Run argocd app history my-app
→ Identify target history ID
→ Run argocd app rollback my-app <id>
→ Verify with argocd app get my-app

---

相关工具

• kubectl：直接检查集群
• k9s：交互式 Kubernetes 仪表板
• stern：多 Pod 日志跟踪
• kubectx/kubens：切换上下文和命名空间