13、实践 - 13.14、kubedog：解决K8S中资源跟踪问题 - 《Kubernetes》

rollout track
follow
multitrack

Kubedog 是一个开源的 Golang 项目，使用 watch 方式对 Kubernetes 资源进行跟踪，能够方便的用于日常运维和 CI/CD 过程之中，项目中除了一个 CLI 小工具之外，还提供了一组 SDK，用户可以将其中的 Watch 功能集成到自己的系统之中。安装过程非常简单，在项目网页直接下载即可。

源码地址：https://github.com/werf/kubedog.git

kubedog主要使用以下三种方式进行资源跟踪：

follow
rollout
multitrack

分别对应三个命令：

kubedog follow
kubedog rollout track
kubedog multitrack

rollout track

在 Kubernetes 上运行应用时，通常的做法是使用 kubectl apply 提交 YAML 之后，使用 kubectl get -w 或者 watch kubectl get 之类的命令等待 Pod 启动。如果启动成功，则进行测试等后续动作；如果启动失败，就需要用 kubectl logs、kubectl describe 等命令来查看失败原因。kubedog 能在一定程度上简化这一过程。
例如使用 kubectl run 命令创建一个新的 Deployment 资源，并使用 kubedog 跟进创建进程：

$ kubectl run nginx --image=nginx22
...
deployment.apps/nginx created
$ kubedog rollout track deployment nginx
# deploy/nginx added
# deploy/nginx rs/nginx-6cc78cbf64 added
# deploy/nginx po/nginx-6cc78cbf64-8pnjz added
# deploy/nginx po/nginx-6cc78cbf64-8pnjz nginx error: ImagePullBackOff: Back-off pulling image "nginx22"
deploy/nginx po/nginx-6cc78cbf64-8pnjz nginx failed: ImagePullBackOff: Back-off pulling image "nginx22"
$ echo $?
130

很方便的看出，运行失败的状态及其原因，并且可以使用返回码来进行判断，方便在 Pipeline 中的运行。接下来可以使用 kubectl edit 命令编辑 Deployment，修改正确的镜像名称。然后再次进行验证：

$ kubectl edit deployment nginx
deployment.extensions/nginx edited
$ kubedog rollout track deployment nginx
# deploy/nginx added
# deploy/nginx rs/nginx-dbddb74b8 added
# deploy/nginx po/nginx-dbddb74b8-x4nkm added
# deploy/nginx event: po/nginx-dbddb74b8-x4nkm Pulled: Successfully pulled image "nginx"
# deploy/nginx event: po/nginx-dbddb74b8-x4nkm Created: Created container
# deploy/nginx event: po/nginx-dbddb74b8-x4nkm Started: Started container
# deploy/nginx event: ScalingReplicaSet: Scaled down replica set nginx-6cc78cbf64 to 0
# deploy/nginx become READY
$ echo $?
0

修改完成，重新运行 kubedog，会看到成功运行的情况，并且返回值也变成了 0。

follow

follow 命令的功能和 kubetail 的功能有少量重叠，可以用 Deployment/Job/Daemonset 等为单位，查看其中所有 Pod 的日志，例如前面用的 Nginx，如果有访问的话，就会看到如下结果：

$ kubedog follow deployment nginx
# deploy/nginx appears to be ready
# deploy/nginx rs/nginx-6cc78cbf64 added
# deploy/nginx new rs/nginx-dbddb74b8 added
# deploy/nginx rs/nginx-dbddb74b8(new) po/nginx-dbddb74b8-x4nkm added
# deploy/nginx rs/nginx-6cc54845d9 added
# deploy/nginx event: ScalingReplicaSet: Scaled up replica set nginx-6cc54845d9 to 1
# deploy/nginx rs/nginx-6cc54845d9(new) po/nginx-6cc54845d9-nhlvs added
# deploy/nginx event: po/nginx-6cc54845d9-nhlvs Pulling: pulling image "nginx:alpine"
# deploy/nginx event: po/nginx-6cc54845d9-nhlvs Pulled: Successfully pulled image "nginx:alpine"
# deploy/nginx event: po/nginx-6cc54845d9-nhlvs Created: Created container
# deploy/nginx event: po/nginx-6cc54845d9-nhlvs Started: Started container
# deploy/nginx event: ScalingReplicaSet: Scaled down replica set nginx-dbddb74b8 to 0
# deploy/nginx become READY
# deploy/nginx event: po/nginx-dbddb74b8-x4nkm Killing: Killing container with id docker://nginx:Need to kill Pod
>> deploy/nginx rs/nginx-dbddb74b8 po/nginx-dbddb74b8-x4nkm nginx
>> deploy/nginx rs/nginx-6cc54845d9(new) po/nginx-6cc54845d9-nhlvs nginx
127.0.0.1 - - [02/Jun/2019:11:35:08 +0000] "GET / HTTP/1.1" 200 612 "-" "Wget" "-"
127.0.0.1 - - [02/Jun/2019:11:35:11 +0000] "GET / HTTP/1.1" 200 612 "-" "Wget" "-"

multitrack

官方不建议使用follow和rollout模式，后续也不会更新；而推荐使用multitrack则直观、易读的方式来展示状态信息。
要使用multitrack，您需要将JSON结构传递给kubedog的stdin，如下：

cat << EOF | kubedog multitrack
{
  "Deployments": [
    {
      "ResourceName": "helloworld",
      "Namespace": "test"
    }
  ],
  "Deployments": [
    {
      "ResourceName": "helloworld1",
      "Namespace": "test"
    }
  ]
}
EOF
或
echo '{"Deployments": [{"ResourceName": "helloworld","Namespace": "test"}],"Deployments": [{"ResourceName": "helloworld1","Namespace": "test"}]}' | kubedog multitrack

我们在做CI/CD的时候，无论是通过Kubernetes CLI还是Kubernetes Continuous Deploy插件，在应用yaml后无法检查资源是否部署成功，只能通过kubectl手动检查，因此我们可以使用kubedog在pipeline完成这最后一步验证。