在 Kubernetes 中搭建 Harbor

0x01 Harbor 简介

Harbor 是一个开源且安全的容器镜像仓库,支持基于 RBAC 的访问权限控制,容器镜像漏洞扫描和镜像签名等安全特性。同时支持管理 Docker 镜像和 Helm Charts Repository,使开发者可以高效便捷且安全地管理各种云原生平台的镜像。

0x02 实验环境

本次实验基于 Docker Desktop 本地环境进行,具体情况如下:

Docker Desktop Community => 2.2.0.3(42716) stable

Docker Engine => 19.03.5

Kubernetes => v1.15.5

Helm => v3.1.2

关于 Helm 的安装过程和使用请参考 Helm

0x03 准备安装文件

首先下载 Harbor Chart 包到要安装的主机上:

$ git clone https://github.com/goharbor/harbor-helm
$ cd harbor-helm

创建对应的 Storage 目录用于绑定 Persistence Volume

$ mkdir $(pwd)/storage

修改 Harbor Chart 中 values.yaml 文件参数来改变部署配置:

expose:
  # 设置暴露服务的方式。将类型设置为 ingress、clusterIP或nodePort并补充对应部分的信息。
  # 由于是本地测试,所以采用绑定主机端口的方式进行服务暴露
  # 部署在集群中根据情况采用 Ingress 或者 LoadBalancer 方式暴露
  type: nodePort
  tls:
    # 是否开启 tls,注意:如果类型是 ingress 并且tls被禁用,则在pull/push镜像时,则必须包含端口
    # 本地测试故不开启 TLS, 线上环境建议开启
    enabled: false
    # 若需要使用自己的证书则需要在此填入对应 TLS Secret 资源名称, TLS Secret 生成请参考
    secretName: ""
    # 若暴露方式为 Ingress 且 Notory 服务需要使用单独的证书时填入对应 TLS Secret 资源名称
    notarySecretName: ""
    # common name 是用于生成证书的,当暴露类型不为 Ingress 且 secretName 为空的时候才需要
    commonName: ""
  ingress:
    hosts:
      core: core.harbor.domain
      notary: notary.harbor.domain
    # set to the type of ingress controller if it has specific requirements.
    # leave as `default` for most ingress controllers.
    # set to `gce` if using the GCE ingress controller
    # set to `ncp` if using the NCP (NSX-T Container Plugin) ingress controller
    controller: default
    annotations:
      ingress.kubernetes.io/ssl-redirect: "true"
      ingress.kubernetes.io/proxy-body-size: "0"
      nginx.ingress.kubernetes.io/ssl-redirect: "true"
      nginx.ingress.kubernetes.io/proxy-body-size: "0"
  clusterIP:
    # ClusterIP 服务名
    name: harbor
    ports:
      httpPort: 80
      httpsPort: 443
      notaryPort: 4443
  nodePort:
    # NodePort 服务名
    name: harbor
    ports:
      http:
        # 容器 HTTP 开放端口
        port: 80
        # 宿主机 HTTP 映射端口
        nodePort: 30002
      https:
        # 容器 HTTPS 开放端口
        port: 443
        # 宿主机 HTTPS 映射端口
        nodePort: 30003
      notary:
        # Notary 服务监听端口
        port: 4443
        # 宿主机 Notary 服务映射端口
        nodePort: 30004
  loadBalancer:
    # LoadBalancer 服务名
    name: harbor
    # 设置 LoadBalancer 绑定的 IP
    IP: ""
    # 相关服务端口开放
    ports:
      httpPort: 80
      httpsPort: 443
      notaryPort: 4443
    annotations: {}
    sourceRanges: []

# Harbor 核心服务外部访问 URL。主要用于:
# 1) 补全 portal 页面上面显示的 docker/helm 命令
# 2) 补全返回给 docker/notary 客户端的 token 服务 URL

# 格式:protocol://domain[:port]。
# 1) 如果 expose.type=ingress,"domain"的值就是 expose.ingress.hosts.core 的值
# 2) 如果 expose.type=clusterIP,"domain"的值就是 expose.clusterIP.name 的值
# 3) 如果 expose.type=nodePort,"domain"的值就是 k8s 节点的 IP 地址
# 如果在代理后面部署 Harbor,请将其设置为代理的 URL
# 本地环境故使用回环地址,线上环境按实际情况填写
externalURL: http://127.0.0.1

# 默认情况下开启数据持久化,在k8s集群中需要动态的挂载卷默认需要一个StorageClass对象。
# 如果你有已经存在可以使用的持久卷,需要在"storageClass"中指定你的 storageClass 或者设置 "existingClaim"。
#
# 对于存储 docker 镜像和 Helm charts 包,你也可以用 "azure"、"gcs"、"s3"、"swift" 或者 "oss",直接在 "imageChartStorage" 区域设置即可
persistence:
  enabled: true
 # 设置成"keep"避免在执行 helm 删除操作期间移除 PVC,留空则在 chart 被删除后删除 PVC
  resourcePolicy: "keep"
  persistentVolumeClaim:
    registry:
       # 使用一个存在的 PVC (必须在绑定前先手动创建)
       # 与其它组件共享 PVC 时请指定子路径分隔
      existingClaim: ""
      # 指定"storageClass",或者使用默认的 StorageClass 对象,设置成"-"禁用动态分配挂载卷
      storageClass: ""
      subPath: ""
      accessMode: ReadWriteOnce
      size: 5Gi
    chartmuseum:
      existingClaim: ""
      storageClass: ""
      subPath: ""
      accessMode: ReadWriteOnce
      size: 5Gi
    jobservice:
      existingClaim: ""
      storageClass: ""
      subPath: ""
      accessMode: ReadWriteOnce
      size: 1Gi
    # 若使用外部数据库服务,以下设置将被忽略
    database:
      existingClaim: ""
      storageClass: ""
      subPath: ""
      accessMode: ReadWriteOnce
      size: 1Gi
    # 若使用外部 Redis 服务,以下设置将被忽略
    redis:
      existingClaim: ""
      storageClass: ""
      subPath: ""
      accessMode: ReadWriteOnce
      size: 1Gi
    trivy:
      existingClaim: ""
      storageClass: ""
      subPath: ""
      accessMode: ReadWriteOnce
      size: 5Gi
  # 定义使用什么存储后端来存储镜像和 Charts 包,详细文档地址:https://github.com/docker/distribution/blob/master/docs/configuration.md#storage
  imageChartStorage:
    # 对镜像和 Chart 存储是否禁用跳转,对于一些不支持的后端(例如对于使用minio 的 s3 存储),需要禁用它。为了禁止跳转,只需要设置 disableredirect=true 即可,详细文档地址:https://github.com/docker/distribution/blob/master/docs/configuration.md#redirect
    disableredirect: false
    # 存储服务使用自签名证书时需要指定 caBundleSecretName
    # caBundleSecretName:

    # 指定存储类型:"filesystem", "azure", "gcs", "s3", "swift", "oss",在相应的区域填上对应的信息。
    # 如果你想使用 pv 则必须设置成"filesystem"类型
    # 本地测试故使用 filesystem 类型,路径为先前创建的目录方便管理,线上环境根据实际情况选择
    type: filesystem
    filesystem:
      rootdirectory: /Path-that-you-just-created-above
      #maxthreads: 100
    azure:
      accountname: accountname
      accountkey: base64encodedaccountkey
      container: containername
      #realm: core.windows.net
    gcs:
      bucket: bucketname
      # The base64 encoded json file which contains the key
      encodedkey: base64-encoded-json-key-file
      #rootdirectory: /gcs/object/name/prefix
      #chunksize: "5242880"
    s3:
      region: us-west-1
      bucket: bucketname
      #accesskey: awsaccesskey
      #secretkey: awssecretkey
      #regionendpoint: http://myobjects.local
      #encrypt: false
      #keyid: mykeyid
      #secure: true
      #v4auth: true
      #chunksize: "5242880"
      #rootdirectory: /s3/object/name/prefix
      #storageclass: STANDARD
    swift:
      authurl: https://storage.myprovider.com/v3/auth
      username: username
      password: password
      container: containername
      #region: fr
      #tenant: tenantname
      #tenantid: tenantid
      #domain: domainname
      #domainid: domainid
      #trustid: trustid
      #insecureskipverify: false
      #chunksize: 5M
      #prefix:
      #secretkey: secretkey
      #accesskey: accesskey
      #authversion: 3
      #endpointtype: public
      #tempurlcontainerkey: false
      #tempurlmethods:
    oss:
      accesskeyid: accesskeyid
      accesskeysecret: accesskeysecret
      region: regionname
      bucket: bucketname
      #endpoint: endpoint
      #internal: false
      #encrypt: false
      #secure: true
      #chunksize: 10M
      #rootdirectory: rootdirectory

# 组件镜像拉取规则
imagePullPolicy: IfNotPresent

# 用于预先设置镜像拉取 Secret
imagePullSecrets:
#  - name: docker-registry-secret
#  - name: internal-registry-secret

# 针对绑定了 PV 的 Deployments 的容器更新规则(jobservice, registry and chartmuseum): "RollingUpdate(滚动更新)" or "Recreate(重新床创建)。"Recreate" 方式不支持 "RWM" 类型的卷存储
updateStrategy:
  type: RollingUpdate

# 日志登记
logLevel: info

# Harbor admin 管理员初始密码
harborAdminPassword: "Harbor12345"
# 用于加密的一个 secret key,必须是一个16位的字符串
secretKey: "1234567890123456"

# 用于更新 Clair 漏洞库的代理设置
proxy:
  httpProxy:
  httpsProxy:
  noProxy: 127.0.0.1,localhost,.local,.internal
  components:
    - core
    - jobservice
    - clair

# 下面为 Harbor 相关组件的设置,暂不需要定制其设置

# 若暴露方式为 "Ingress",下列的 Nginx 将不被使用
  image:
    repository: goharbor/nginx-photon
    tag: dev
  replicas: 1
  # resources:
  #  requests:
  #    memory: 256Mi
  #    cpu: 100m
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}

portal:
  image:
    repository: goharbor/harbor-portal
    tag: dev
  replicas: 1
# resources:
#  requests:
#    memory: 256Mi
#    cpu: 100m
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}

core:
  image:
    repository: goharbor/harbor-core
    tag: dev
  replicas: 1
  ## Liveness probe values
  livenessProbe:
    initialDelaySeconds: 300
# resources:
#  requests:
#    memory: 256Mi
#    cpu: 100m
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}
  # Secret is used when core server communicates with other components.
  # If a secret key is not specified, Helm will generate one.
  # Must be a string of 16 chars.
  secret: ""
  # Fill the name of a kubernetes secret if you want to use your own
  # TLS certificate and private key for token encryption/decryption.
  # The secret must contain keys named:
  # "tls.crt" - the certificate
  # "tls.key" - the private key
  # The default key pair will be used if it isn't set
  secretName: ""
  # The XSRF key. Will be generated automatically if it isn't specified
  xsrfKey: ""

jobservice:
  image:
    repository: goharbor/harbor-jobservice
    tag: dev
  replicas: 1
  maxJobWorkers: 10
  # The logger for jobs: "file", "database" or "stdout"
  jobLogger: file
# resources:
#   requests:
#     memory: 256Mi
#     cpu: 100m
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}
  # Secret is used when job service communicates with other components.
  # If a secret key is not specified, Helm will generate one.
  # Must be a string of 16 chars.
  secret: ""

registry:
  registry:
    image:
      repository: goharbor/registry-photon
      tag: dev

    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
  controller:
    image:
      repository: goharbor/harbor-registryctl
      tag: dev

    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
  replicas: 1
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}
  # Secret is used to secure the upload state from client
  # and registry storage backend.
  # See: https://github.com/docker/distribution/blob/master/docs/configuration.md#http
  # If a secret key is not specified, Helm will generate one.
  # Must be a string of 16 chars.
  secret: ""
  # If true, the registry returns relative URLs in Location headers. The client is responsible for resolving the correct URL.
  relativeurls: false
  credentials:
    username: "harbor_registry_user"
    password: "harbor_registry_password"
    # If you update the username or password of registry, make sure use cli tool htpasswd to generate the bcrypt hash
    # e.g. "htpasswd -nbBC10 $username $password"
    htpasswd: "harbor_registry_user:$2y$10$9L4Tc0DJbFFMB6RdSCunrOpTHdwhid4ktBJmLD00bYgqkkGOvll3m"

  middleware:
    enabled: false
    type: cloudFront
    cloudFront:
      baseurl: example.cloudfront.net
      keypairid: KEYPAIRID
      duration: 3000s
      ipfilteredby: none
      # The secret key that should be present is CLOUDFRONT_KEY_DATA, which should be the encoded private key
      # that allows access to CloudFront
      privateKeySecret: "my-secret"

chartmuseum:
  enabled: true
  # Harbor defaults ChartMuseum to returning relative urls, if you want using absolute url you should enable it by change the following value to 'true'
  absoluteUrl: false
  image:
    repository: goharbor/chartmuseum-photon
    tag: dev
  replicas: 1
  # resources:
  #  requests:
  #    memory: 256Mi
  #    cpu: 100m
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}

clair:
  enabled: true
  clair:
    image:
      repository: goharbor/clair-photon
      tag: dev
    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
  adapter:
    image:
      repository: goharbor/clair-adapter-photon
      tag: dev
    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
  replicas: 1
  # The interval of clair updaters, the unit is hour, set to 0 to
  # disable the updaters
  updatersInterval: 12
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}

trivy:
  # enabled the flag to enable Trivy scanner
  enabled: true
  image:
    # repository the repository for Trivy adapter image
    repository: goharbor/trivy-adapter-photon
    # tag the tag for Trivy adapter image
    tag: dev
  # replicas the number of Pod replicas
  replicas: 1
  # debugMode the flag to enable Trivy debug mode with more verbose scanning log
  debugMode: false
  # vulnType a comma-separated list of vulnerability types. Possible values are `os` and `library`.
  vulnType: "os,library"
  # severity a comma-separated list of severities to be checked
  severity: "UNKNOWN,LOW,MEDIUM,HIGH,CRITICAL"
  # ignoreUnfixed the flag to display only fixed vulnerabilities
  ignoreUnfixed: false
  # gitHubToken the GitHub access token to download Trivy DB
  #
  # Trivy DB contains vulnerability information from NVD, Red Hat, and many other upstream vulnerability databases.
  # It is downloaded by Trivy from the GitHub release page https://github.com/aquasecurity/trivy-db/releases and cached
  # in the local file system (`/home/scanner/.cache/trivy/db/trivy.db`). In addition, the database contains the update
  # timestamp so Trivy can detect whether it should download a newer version from the Internet or use the cached one.
  # Currently, the database is updated every 12 hours and published as a new release to GitHub.
  #
  # Anonymous downloads from GitHub are subject to the limit of 60 requests per hour. Normally such rate limit is enough
  # for production operations. If, for any reason, it's not enough, you could increase the rate limit to 5000
  # requests per hour by specifying the GitHub access token. For more details on GitHub rate limiting please consult
  # https://developer.github.com/v3/#rate-limiting
  #
  # You can create a GitHub token by following the instructions in
  # https://help.github.com/en/github/authenticating-to-github/creating-a-personal-access-token-for-the-command-line
  gitHubToken: ""
  # skipUpdate the flag to disable Trivy DB downloads from GitHub
  #
  # You might want to set the value of this flag to `true` in test or CI/CD environments to avoid GitHub rate limiting issues.
  # If the value is set to `true` you have to manually download the `trivy.db` file and mount it in the
  # `/home/scanner/.cache/trivy/db/trivy.db` path.
  skipUpdate: false
  resources:
    requests:
      cpu: 200m
      memory: 512Mi
    limits:
      cpu: 1
      memory: 1Gi

notary:
  enabled: true
  server:
    image:
      repository: goharbor/notary-server-photon
      tag: dev
    replicas: 1
    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
  signer:
    image:
      repository: goharbor/notary-signer-photon
      tag: dev
    replicas: 1
    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
  nodeSelector: {}
  tolerations: []
  affinity: {}
  ## Additional deployment annotations
  podAnnotations: {}
  # Fill the name of a kubernetes secret if you want to use your own
  # TLS certificate authority, certificate and private key for notary
  # communications.
  # The secret must contain keys named ca.crt, tls.crt and tls.key that
  # contain the CA, certificate and private key.
  # They will be generated if not set.
  secretName: ""

database:
  # if external database is used, set "type" to "external"
  # and fill the connection informations in "external" section
  type: internal
  internal:
    image:
      repository: goharbor/harbor-db
      tag: dev
    # the image used by the init container
    initContainerImage:
      repository: busybox
      tag: latest
    # The initial superuser password for internal database
    password: "changeit"
    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
    nodeSelector: {}
    tolerations: []
    affinity: {}
  external:
    host: "192.168.0.1"
    port: "5432"
    username: "user"
    password: "password"
    coreDatabase: "registry"
    clairDatabase: "clair"
    notaryServerDatabase: "notary_server"
    notarySignerDatabase: "notary_signer"
    # "disable" - No SSL
    # "require" - Always SSL (skip verification)
    # "verify-ca" - Always SSL (verify that the certificate presented by the
    # server was signed by a trusted CA)
    # "verify-full" - Always SSL (verify that the certification presented by the
    # server was signed by a trusted CA and the server host name matches the one
    # in the certificate)
    sslmode: "disable"
  # The maximum number of connections in the idle connection pool.
  # If it <=0, no idle connections are retained.
  maxIdleConns: 50
  # The maximum number of open connections to the database.
  # If it <= 0, then there is no limit on the number of open connections.
  # Note: the default number of connections is 100 for postgre.
  maxOpenConns: 100
  ## Additional deployment annotations
  podAnnotations: {}

redis:
  # if external Redis is used, set "type" to "external"
  # and fill the connection informations in "external" section
  type: internal
  internal:
    image:
      repository: goharbor/redis-photon
      tag: dev
    # resources:
    #  requests:
    #    memory: 256Mi
    #    cpu: 100m
    nodeSelector: {}
    tolerations: []
    affinity: {}
  external:
    host: "192.168.0.2"
    port: "6379"
    # The "coreDatabaseIndex" must be "0" as the library Harbor
    # used doesn't support configuring it
    coreDatabaseIndex: "0"
    jobserviceDatabaseIndex: "1"
    registryDatabaseIndex: "2"
    chartmuseumDatabaseIndex: "3"
    clairAdapterIndex: "4"
    trivyAdapterIndex: "5"
    password: ""
  ## Additional deployment annotations
  podAnnotations: {}

0x04 使用 Helm 进行安装

这一步使用 Helm 将刚修改好的 Harbor Chart 安装到测试环境中:

$ helm install harbor .

随后通过 kubectl 查询容器情况:

$ kubectl get pods

各个组件启动成功,也可以查看各个 Deployment, Service, Persistence Volume 等资源状态,需要注意的是,由于 harbor-harbor-databaseStatefulset, 需要较长时间的初始化,会使得部分依赖其服务的组件在启动后出现 CrashLoopBackOff 的情况。

0x05 Let's Harbor

本地测试环境下,先前配置了 NodePort 的暴露方式,先查看一下对应 Service 的情况:

$ kubectl get svc

根据 harbor 的端口暴露可以得知,HTTP 服务已经绑定至宿主机的 30002 端口,故使用 http://127.0.0.1:30002/ 进入 Harbor Portal,输入 values 中预先定义好的账号(admin)和密码(Harbor12345)即可登录。

Harbor 的 GUI 管理界面还有诸多的功能,例如:镜像扫描、日志审计等,可以根据需要进行探索。

0x06 尝试使用 Docker Cli 推送镜像

$ docker login -u admin 127.0.0.1:30002/library
Password: 
Login Succeeded
$ docker tag busybox:latest 127.0.0.1:30002/library/busybox:latest
$ docker push 127.0.0.1:30002/library/busybox:latest
The push refers to repository [127.0.0.1:30002/library/busybox]
a6d503001157: Pushed 
latest: digest: sha256:afe605d272837ce1732f390966166c2afff5391208ddd57de10942748694049d size: 527

到这里证明上面我们的私有 Docker 镜像仓库搭建成功了,可以 pull/push 镜像,Harbor 还具有其他的一些功能,比如镜像复制,可以根据需要探索

0x07 尝试使用 Helm 推送 Helm Charts

用于测试推送的 Chart 是使用 Helm 新创建的。

# 安装 Helm Push 插件
$ helm plugin install https://github.com/chartmuseum/helm-push
Downloading and installing helm-push v0.8.1 ...
https://github.com/chartmuseum/helm-push/releases/download/v0.8.1/helm-push_0.8.1_darwin_amd64.tar.gz
Installed plugin: push

# 添加私有 Helm Chart Repo
$ helm repo add myharborlib http://127.0.0.1:30002/chartrepo/library
"myharborlib" has been added to your repositories

# 推送 Helm Chart
$ helm push test/ myharborlib -u admin -p Harbor12345
Pushing test-0.1.0.tgz to myharborlib...
Done.

到这里证明上面我们的私有 Chart Meseum 镜像仓库搭建成功了,可以 pull/push Helm Charts。


0 条评论

发表回复

Avatar placeholder

您的电子邮箱地址不会被公开。 必填项已用*标注