Dask Kubernetes worker pod 给出错误状态

Dask Kubernetes worker pod giving error status

我正在关注 link: https://kubernetes.dask.org/en/latest/,到 Kubernetes 集群上的 运行 dask 阵列。在 运行 运行示例代码时,worker pod 显示如下错误状态:

步骤:

  1. 在 3 个节点(1 个主节点和 2 个工作节点)上安装了 Kubernetes。

  2. pip 安装 dask-kubernetes

  3. dask_example.py 代码为 运行 dask 数组(与 link 上给出的示例相同)

  4. Worker-spec.yml 包含 Pod 配置的文件(与 link 中给出的示例相同)

(base) [root@k8s-master example]# ls
dask_example.py  worker-spec.yml
(base) [root@k8s-master example]# nohup python dask_example.py &
[1] 3660
(base) [root@k8s-master example]# cat nohup.out
distributed.scheduler - INFO - Clear task state
distributed.scheduler - INFO -   Scheduler at:   tcp://172.16.0.76:40119
distributed.scheduler - INFO - Receive client connection: Client-df4caa18-0bc8-11ea-8e4c-12bd5ffa93ff
distributed.core - INFO - Starting established connection
(base) [root@k8s-master example]# kubectl get pods -o wide --all-namespaces
NAMESPACE     NAME                                 READY   STATUS    RESTARTS   AGE     IP             NODE           NOMINATED NODE   READINESS GATES
default       workerpod                            1/1     Running   0          70s     10.32.0.2      worker-node1   <none>           <none>
kube-system   coredns-5644d7b6d9-l4jsd             1/1     Running   0          8m19s   10.32.0.4      k8s-master     <none>           <none>
kube-system   coredns-5644d7b6d9-q679h             1/1     Running   0          8m19s   10.32.0.3      k8s-master     <none>           <none>
kube-system   etcd-k8s-master                      1/1     Running   0          7m16s   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-apiserver-k8s-master            1/1     Running   0          7m1s    172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-controller-manager-k8s-master   1/1     Running   0          7m27s   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-proxy-ctgj8                     1/1     Running   0          5m7s    172.16.0.114   worker-node2   <none>           <none>
kube-system   kube-proxy-f78bm                     1/1     Running   0          8m18s   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-proxy-ksk59                     1/1     Running   0          5m15s   172.16.0.31    worker-node1   <none>           <none>
kube-system   kube-scheduler-k8s-master            1/1     Running   0          7m2s    172.16.0.76    k8s-master     <none>           <none>
kube-system   weave-net-q2zwn                      2/2     Running   0          6m22s   172.16.0.76    k8s-master     <none>           <none>
kube-system   weave-net-r9tzs                      2/2     Running   0          5m15s   172.16.0.31    worker-node1   <none>           <none>
kube-system   weave-net-tm8xx                      2/2     Running   0          5m7s    172.16.0.114   worker-node2   <none>           <none>
(base) [root@k8s-master example]# kubectl get pods -o wide --all-namespaces
NAMESPACE     NAME                                 READY   STATUS    RESTARTS   AGE     IP             NODE           NOMINATED NODE   READINESS GATES
default       workerpod                            0/1     Error     0          4m23s   10.32.0.2      worker-node1   <none>           <none>
kube-system   coredns-5644d7b6d9-l4jsd             1/1     Running   0          11m     10.32.0.4      k8s-master     <none>           <none>
kube-system   coredns-5644d7b6d9-q679h             1/1     Running   0          11m     10.32.0.3      k8s-master     <none>           <none>
kube-system   etcd-k8s-master                      1/1     Running   0          10m     172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-apiserver-k8s-master            1/1     Running   0          10m     172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-controller-manager-k8s-master   1/1     Running   0          10m     172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-proxy-ctgj8                     1/1     Running   0          8m20s   172.16.0.114   worker-node2   <none>           <none>
kube-system   kube-proxy-f78bm                     1/1     Running   0          11m     172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-proxy-ksk59                     1/1     Running   0          8m28s   172.16.0.31    worker-node1   <none>           <none>
kube-system   kube-scheduler-k8s-master            1/1     Running   0          10m     172.16.0.76    k8s-master     <none>           <none>
kube-system   weave-net-q2zwn                      2/2     Running   0          9m35s   172.16.0.76    k8s-master     <none>           <none>
kube-system   weave-net-r9tzs                      2/2     Running   0          8m28s   172.16.0.31    worker-node1   <none>           <none>
kube-system   weave-net-tm8xx                      2/2     Running   0          8m20s   172.16.0.114   worker-node2   <none>           <none>
(base) [root@k8s-master example]# cat nohup.out
distributed.scheduler - INFO - Clear task state
distributed.scheduler - INFO -   Scheduler at:   tcp://172.16.0.76:40119
distributed.scheduler - INFO - Receive client connection: Client-df4caa18-0bc8-11ea-8e4c-12bd5ffa93ff
distributed.core - INFO - Starting established connection
 (base) [root@k8s-master example]# kubectl describe pod workerpod
Name:         workerpod
Namespace:    default
Priority:     0
Node:         worker-node1/172.16.0.31
Start Time:   Wed, 20 Nov 2019 19:06:36 +0000
Labels:       app=dask
          dask.org/cluster-name=dask-root-99dcf768-4
          dask.org/component=worker
          foo=bar
          user=root
Annotations:  <none>
Status:       Failed
 IP:           10.32.0.2
 IPs:
   IP:  10.32.0.2
Containers:
  dask:
    Container ID:  docker://578dc575fc263c4a3889a4f2cb5e06cd82a00e03cfc6acfd7a98fef703421390
    Image:         daskdev/dask:latest
    Image ID:      docker-pullable://daskdev/dask@sha256:0a936daa94c82cea371c19a2c90c695688ab4e1e7acc905f8b30dfd419adfb6f
Port:          <none>
Host Port:     <none>
Args:
  dask-worker
  --nthreads
  2
  --no-bokeh
  --memory-limit
  6GB
  --death-timeout
  60
State:          Terminated
  Reason:       Error
  Exit Code:    1
  Started:      Wed, 20 Nov 2019 19:06:38 +0000
  Finished:     Wed, 20 Nov 2019 19:08:20 +0000
Ready:          False
Restart Count:  0
Limits:
  cpu:     2
  memory:  6G
Requests:
  cpu:     2
  memory:  6G
Environment:
  EXTRA_PIP_PACKAGES:      fastparquet git+https://github.com/dask/distributed
  DASK_SCHEDULER_ADDRESS:  tcp://172.16.0.76:40119
Mounts:
  /var/run/secrets/kubernetes.io/serviceaccount from default-token-p9f9v (ro)
Conditions:
  Type              Status
  Initialized       True
  Ready             False
  ContainersReady   False
  PodScheduled      True
Volumes:
  default-token-p9f9v:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-p9f9v
    Optional:    false
QoS Class:       Guaranteed
Node-Selectors:  <none>
Tolerations:     k8s.dask.org/dedicated=worker:NoSchedule
                 k8s.dask.org_dedicated=worker:NoSchedule
                 node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type    Reason     Age    From                   Message
  ----    ------     ----   ----                   -------
  Normal  Scheduled  5m47s  default-scheduler      Successfully assigned default/workerpod to worker-node1
  Normal  Pulled     5m45s  kubelet, worker-node1  Container image "daskdev/dask:latest" already present on machine
  Normal  Created    5m45s  kubelet, worker-node1  Created container dask
  Normal  Started    5m45s  kubelet, worker-node1  Started container dask
(base) [root@k8s-master example]#
(base) [root@k8s-master example]# kubectl get events
LAST SEEN   TYPE     REASON                    OBJECT              MESSAGE
21m         Normal   Starting                  node/k8s-master     Starting kubelet.
21m         Normal   NodeHasSufficientMemory   node/k8s-master     Node k8s-master status is now: NodeHasSufficientMemory
21m         Normal   NodeHasNoDiskPressure     node/k8s-master     Node k8s-master status is now: NodeHasNoDiskPressure
21m         Normal   NodeHasSufficientPID      node/k8s-master     Node k8s-master status is now: NodeHasSufficientPID
21m         Normal   NodeAllocatableEnforced   node/k8s-master     Updated Node Allocatable limit across pods
21m         Normal   RegisteredNode            node/k8s-master     Node k8s-master event: Registered Node k8s-master in Controller
21m         Normal   Starting                  node/k8s-master     Starting kube-proxy.
18m         Normal   Starting                  node/worker-node1   Starting kubelet.
18m         Normal   NodeHasSufficientMemory   node/worker-node1   Node worker-node1 status is now: NodeHasSufficientMemory
18m         Normal   NodeHasNoDiskPressure     node/worker-node1   Node worker-node1 status is now: NodeHasNoDiskPressure
18m         Normal   NodeHasSufficientPID      node/worker-node1   Node worker-node1 status is now: NodeHasSufficientPID
18m         Normal   NodeAllocatableEnforced   node/worker-node1   Updated Node Allocatable limit across pods
18m         Normal   Starting                  node/worker-node1   Starting kube-proxy.
18m         Normal   RegisteredNode            node/worker-node1   Node worker-node1 event: Registered Node worker-node1 in Controller
17m         Normal   NodeReady                 node/worker-node1   Node worker-node1 status is now: NodeReady
18m         Normal   Starting                  node/worker-node2   Starting kubelet.
18m         Normal   NodeHasSufficientMemory   node/worker-node2   Node worker-node2 status is now: NodeHasSufficientMemory
18m         Normal   NodeHasNoDiskPressure     node/worker-node2   Node worker-node2 status is now: NodeHasNoDiskPressure
18m         Normal   NodeHasSufficientPID      node/worker-node2   Node worker-node2 status is now: NodeHasSufficientPID
18m         Normal   NodeAllocatableEnforced   node/worker-node2   Updated Node Allocatable limit across pods
18m         Normal   Starting                  node/worker-node2   Starting kube-proxy.
17m         Normal   RegisteredNode            node/worker-node2   Node worker-node2 event: Registered Node worker-node2 in Controller
17m         Normal   NodeReady                 node/worker-node2   Node worker-node2 status is now: NodeReady
14m         Normal   Scheduled                 pod/workerpod       Successfully assigned default/workerpod to worker-node1
14m         Normal   Pulled                    pod/workerpod       Container image "daskdev/dask:latest" already present on machine
14m         Normal   Created                   pod/workerpod       Created container dask
14m         Normal   Started                   pod/workerpod       Started container dask
(base) [root@k8s-master example]#

更新-添加 pod 日志(根据 Dawid Kruk 的建议):

(base) [root@k8s-master example]# kubectl logs workerpod
+ '[' '' ']'
+ '[' -e /opt/app/environment.yml ']'
+ echo 'no environment.yml'
+ '[' '' ']'
+ '[' 'fastparquet git+https://github.com/dask/distributed' ']'
+ echo 'EXTRA_PIP_PACKAGES environment variable found.  Installing.'
+ /opt/conda/bin/pip install fastparquet git+https://github.com/dask/distributed
no environment.yml
EXTRA_PIP_PACKAGES environment variable found.  Installing.
Collecting git+https://github.com/dask/distributed
  Cloning https://github.com/dask/distributed to /tmp/pip-req-build-i3_1vo06
  Running command git clone -q https://github.com/dask/distributed /tmp/pip-req-build-i3_1vo06
  fatal: unable to access 'https://github.com/dask/distributed/': Could not resolve host: github.com
ERROR: Command errored out with exit status 128: git clone -q https://github.com/dask/distributed /tmp/pip-req-build-i3_1vo06 Check the logs for full command output.
+ exec dask-worker --nthreads 2 --no-bokeh --memory-limit 6GB --death-timeout 60
/opt/conda/lib/python3.7/site-packages/distributed/cli/dask_worker.py:252: UserWarning: The --bokeh/--no-bokeh flag has been renamed to --dashboard/--no-dashboard.
  "The --bokeh/--no-bokeh flag has been renamed to --dashboard/--no-dashboard. "
distributed.nanny - INFO -         Start Nanny at: 'tcp://10.32.0.2:45097'
distributed.worker - INFO -       Start worker at:      tcp://10.32.0.2:36389
distributed.worker - INFO -          Listening to:      tcp://10.32.0.2:36389
distributed.worker - INFO - Waiting to connect to:    tcp://172.16.0.76:43389
distributed.worker - INFO - -------------------------------------------------
distributed.worker - INFO -               Threads:                          2
distributed.worker - INFO -                Memory:                    6.00 GB
distributed.worker - INFO -       Local Directory:           /worker-55rpow8j
distributed.worker - INFO - -------------------------------------------------
distributed.worker - INFO - Waiting to connect to:    tcp://172.16.0.76:43389
distributed.worker - INFO - Waiting to connect to:    tcp://172.16.0.76:43389
distributed.worker - INFO - Waiting to connect to:    tcp://172.16.0.76:43389
distributed.worker - INFO - Waiting to connect to:    tcp://172.16.0.76:43389
distributed.worker - INFO - Waiting to connect to:    tcp://172.16.0.76:43389
distributed.nanny - INFO - Closing Nanny at 'tcp://10.32.0.2:45097'
distributed.worker - INFO - Stopping worker at tcp://10.32.0.2:36389
distributed.worker - INFO - Closed worker has not yet started: None
distributed.dask_worker - INFO - Timed out starting worker
distributed.dask_worker - INFO - End worker
(base) [root@k8s-master example]# git
usage: git [--version] [--help] [-C <path>] [-c <name>=<value>]
           [--exec-path[=<path>]] [--html-path] [--man-path] [--info-path]

在我看来,dask pod 与工作节点的连接存在问题,但我看到工作节点处于就绪状态,其他 pods (nginx) 在工作节点上 运行ning:

(base) [root@k8s-master example]# kubectl get nodes
NAME           STATUS   ROLES    AGE   VERSION
k8s-master     Ready    master   20h   v1.16.3
worker-node1   Ready    worker   20h   v1.16.2
worker-node2   Ready    worker   20h   v1.16.2
NAMESPACE     NAME                                 READY   STATUS    RESTARTS   AGE   IP             NODE           NOMINATED NODE   READINESS GATES
default       nginx-deployment-54f57cf6bf-b9nfd    1/1     Running   0          34s   10.32.0.2      worker-node1   <none>           <none>
default       nginx-deployment-54f57cf6bf-pnp59    1/1     Running   0          34s   10.40.0.0      worker-node2   <none>           <none>
default       workerpod                            0/1     Error     0          56m   10.32.0.2      worker-node1   <none>           <none>
kube-system   coredns-5644d7b6d9-l4jsd             1/1     Running   0          21h   10.32.0.4      k8s-master     <none>           <none>
kube-system   coredns-5644d7b6d9-q679h             1/1     Running   0          21h   10.32.0.3      k8s-master     <none>           <none>
kube-system   etcd-k8s-master                      1/1     Running   0          21h   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-apiserver-k8s-master            1/1     Running   0          21h   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-controller-manager-k8s-master   1/1     Running   0          21h   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-proxy-ctgj8                     1/1     Running   0          21h   172.16.0.114   worker-node2   <none>           <none>
kube-system   kube-proxy-f78bm                     1/1     Running   0          21h   172.16.0.76    k8s-master     <none>           <none>
kube-system   kube-proxy-ksk59                     1/1     Running   0          21h   172.16.0.31    worker-node1   <none>           <none>
kube-system   kube-scheduler-k8s-master            1/1     Running   0          21h   172.16.0.76    k8s-master     <none>           <none>
kube-system   weave-net-q2zwn                      2/2     Running   0          21h   172.16.0.76    k8s-master     <none>           <none>
kube-system   weave-net-r9tzs                      2/2     Running   0          21h   172.16.0.31    worker-node1   <none>           <none>
kube-system   weave-net-tm8xx                      2/2     Running   0          21h   172.16.0.114   worker-node2   <none>           <none>

Update2 - 添加了 nslookup 输出(VAS 建议)参考:https://kubernetes.io/docs/tasks/administer-cluster/dns-debugging-resolution/#create-a-simple-pod-to-use-as-a-test-environment

(base) [root@k8s-master example]# kubectl exec -ti workerpod -- nslookup kubernetes.default
OCI runtime exec failed: exec failed: container_linux.go:345: starting container process caused "exec: \"nslookup\": executable file not found in $PATH": unknown
command terminated with exit code 126

如何向pod添加可执行文件?它是在我的主机上设置的。

(base) [root@k8s-master example]# nslookup github.com
Server:         172.31.0.2
Address:        172.31.0.2#53

Non-authoritative answer:
Name:   github.com
Address: 140.82.114.3

更新 3:dnsutils 的 nslookup(VAS 建议)

(base) [root@k8s-master example]# kubectl run dnsutils -it --rm=true --restart=Never --image=tutum/dnsutils cat /etc/resolv.conf
nameserver 10.96.0.10
search default.svc.cluster.local svc.cluster.local cluster.local ec2.internal
options ndots:5
pod "dnsutils" deleted
(base) [root@k8s-master example]# kubectl run dnsutils -it --restart=Never --image=tutum/dnsutils nslookup github.com
If you don't see a command prompt, try pressing enter.
;; connection timed out; no servers could be reached

pod default/dnsutils terminated (Error)
(base) [root@k8s-master example]# kubectl logs dnsutils

;; connection timed out; no servers could be reached

(base) [root@k8s-master example]#

更新 4:

(base) [root@k8s-master example]# kubectl exec -ti busybox -- nslookup kubernetes.default
Server:    10.96.0.10
Address 1: 10.96.0.10

nslookup: can't resolve 'kubernetes.default'
command terminated with exit code 1
(base) [root@k8s-master example]# kubectl get pods --namespace=kube-system -l k8s-app=kube-dns
NAME                       READY   STATUS    RESTARTS   AGE
coredns-5644d7b6d9-l4jsd   1/1     Running   0          25h
coredns-5644d7b6d9-q679h   1/1     Running   0          25h
(base) [root@k8s-master example]# kubectl get pods --namespace=kube-system -l k8s-app=kube-dns
NAME                       READY   STATUS    RESTARTS   AGE
coredns-5644d7b6d9-l4jsd   1/1     Running   0          25h
coredns-5644d7b6d9-q679h   1/1     Running   0          25h
(base) [root@k8s-master example]# for p in $(kubectl get pods --namespace=kube-system -l k8s-app=kube-dns -o name); do kubectl logs --namespace=kube-system $p; done
.:53
2019-11-20T19:01:42.161Z [INFO] plugin/reload: Running configuration MD5 = f64cb9b977c7dfca58c4fab108535a76
2019-11-20T19:01:42.161Z [INFO] CoreDNS-1.6.2
2019-11-20T19:01:42.161Z [INFO] linux/amd64, go1.12.8, 795a3eb
CoreDNS-1.6.2
linux/amd64, go1.12.8, 795a3eb
.:53
2019-11-20T19:01:41.862Z [INFO] plugin/reload: Running configuration MD5 = f64cb9b977c7dfca58c4fab108535a76
2019-11-20T19:01:41.862Z [INFO] CoreDNS-1.6.2
2019-11-20T19:01:41.862Z [INFO] linux/amd64, go1.12.8, 795a3eb
CoreDNS-1.6.2
linux/amd64, go1.12.8, 795a3eb
(base) [root@k8s-master example]# kubectl get service
NAME         TYPE        CLUSTER-IP   EXTERNAL-IP   PORT(S)   AGE
kubernetes   ClusterIP   10.96.0.1    <none>        443/TCP   26h
(base) [root@k8s-master example]# kubectl get ep kube-dns --namespace=kube-system
NAME       ENDPOINTS                                            AGE
kube-dns   10.32.0.3:53,10.32.0.4:53,10.32.0.3:53 + 3 more...   26h
(base) [root@k8s-master example]# kubectl -n kube-system edit configmap coredns
Edit cancelled, no changes made.

来自工作节点 1 的 nslookup

[root@worker-node1 ec2-user]# nslookup 10.96.0.10
Server:         172.31.0.2
Address:        172.31.0.2#53

Non-authoritative answer:
10.0.96.10.in-addr.arpa name = ip-10-96-0-10.ec2.internal.

Authoritative answers can be found from:

[root@worker-node1 ec2-user]# nslookup 10.96.0.1
Server:         172.31.0.2
Address:        172.31.0.2#53

Non-authoritative answer:
1.0.96.10.in-addr.arpa  name = ip-10-96-0-1.ec2.internal.

Authoritative answers can be found from:

[root@worker-node1 ec2-user]#

来自工作节点 2 的 nslookup

[root@worker-node2 ec2-user]# nslookup 10.96.0.10
Server:         172.31.0.2
Address:        172.31.0.2#53

Non-authoritative answer:
10.0.96.10.in-addr.arpa name = ip-10-96-0-10.ec2.internal.

Authoritative answers can be found from:

[root@worker-node2 ec2-user]# nslookup 10.96.0.1
Server:         172.31.0.2
Address:        172.31.0.2#53

Non-authoritative answer:
1.0.96.10.in-addr.arpa  name = ip-10-96-0-1.ec2.internal.

Authoritative answers can be found from:

更新:

从评论中可以看出,我也需要在回答中涵盖一些基本概念。

为确保 Kubernetes 集群正常运行,应满足一些要求。

  1. 所有 Kubernetes 节点必须有 full network connectivity.
    这意味着任何集群节点都应该能够使用任何网络协议和任何端口(在 tcp/udp 的情况下)与任何其他集群节点通信,而无需 NAT。一些云环境需要额外的自定义防火墙规则来实现这一点。 Calico example

  2. Kubernetes Pods 应该能够与调度在其他节点上的 pods 通信。
    此功能由 [CNI 网络附加组件] 提供。大多数流行的附加组件需要 Kubernetes 控制平面中的附加选项,通常由 kubeadm init --pod-network-cidr=a.b.c.d/16 命令行选项设置。请注意,不同网络附加组件的默认 IP 子网不相同。
    如果要为特定网络附加组件使用自定义 Pod 子网,则必须在将其应用到集群之前自定义网络附加组件部署 YAML 文件。
    通过从节点 CLI 或 Pod CLI 向调度在另一个节点上的 Pod 的任何 IP 地址发送 ICMP 或 curl 请求,可以轻松测试 Pod 间连接。 请注意,Service ClusterIP 不会响应 ICMP 请求,因为它只是一组 iptables 转发规则。 pods 的完整列表可以显示为使用以下命令:

    kubectl get pods --all-namespaces -o wide
    
  3. 对于服务发现功能,必须在 Kubernetes 集群中运行 DNS 服务。
    对于 v1.9 之前的 Kubernetes 版本通常是 kubedns,对于较新的集群是 coredns
    Kubernetes DNS 服务通常包含一个具有两个副本的部署和一个默认 IP 地址为 10.96.0.10 的 ClusterIP 服务。

查看问题中的数据,我怀疑您可能是网络插件有问题。我将使用以下命令对其进行测试,这应该 return 在健康集群上取得成功结果:

# check connectivity from the master node to nginx pods:
k8s-master$ ping 10.32.0.2
k8s-master$ curl 10.32.0.2 
k8s-master$ ping 10.40.0.0
k8s-master$ curl 10.40.0.0

# check connectivity from other nodes to coredns pods and DNS Service:
worker-node1$ nslookup github.com 10.32.0.4
worker-node1$ nslookup github.com 10.32.0.3
worker-node1$ nslookup github.com 10.96.0.10

worker-node2$ nslookup github.com 10.32.0.4
worker-node2$ nslookup github.com 10.32.0.3
worker-node2$ nslookup github.com 10.96.0.10

故障排除网络插件是一个相当大的知识,要写在一个答案中,所以,如果您需要修复网络插件,请搜索现有答案并提出另一个问题,以防找不到合适的.


以下部分描述了如何在 Kubernetes 集群中检查 DNS 服务:

dnsPolicy ClusterFirst(默认)的情况下,任何与配置的集群域后缀不匹配的 DNS 查询(例如“www.kubernetes.io”)将转发到从节点继承的上游名称服务器。

如何检查集群节点上的 DNS 客户端配置:

$ cat /etc/resolv.conf
$ systemd-resolve --status

如何检查节点上的 DNS 客户端是否正常工作:

$ nslookup github.com

Server:         127.0.0.53
Address:        127.0.0.53#53

Non-authoritative answer:
Name:   github.com
Address: 140.82.118.4

如何获取 Kubernetes 集群 DNS 配置:

$ kubectl get svc,pods -n kube-system -o wide | grep dns 

service/kube-dns        ClusterIP   10.96.0.10      <none>        53/UDP,53/TCP,9153/TCP   185d   k8s-app=kube-dns

pod/coredns-fb8b8dccf-5jjv8                     1/1     Running   122        185d   10.244.0.16   kube-master2   <none>           <none>
pod/coredns-fb8b8dccf-5pbkg                     1/1     Running   122        185d   10.244.0.17   kube-master2   <none>           <none>

$ kubectl get configmap coredns -n kube-system -o yaml

apiVersion: v1
data:
  Corefile: |
    .:53 {
        errors
        health
        kubernetes cluster.local in-addr.arpa ip6.arpa {
           pods insecure
           upstream
           fallthrough in-addr.arpa ip6.arpa
        }
        prometheus :9153
        forward . /etc/resolv.conf
        cache 30
        loop
        reload
        loadbalance
    }
kind: ConfigMap
metadata:
  creationTimestamp: "2019-05-20T16:10:42Z"
  name: coredns
  namespace: kube-system
  resourceVersion: "1657005"
  selfLink: /api/v1/namespaces/kube-system/configmaps/coredns
  uid: d1598034-7b19-11e9-9137-42010a9c0004

如何检查集群DNS服务(coredns)是否正常工作:

$ nslookup github.com 10.96.0.10

Server:         10.96.0.10
Address:        10.96.0.10#53

Non-authoritative answer:
Name:   github.com
Address: 140.82.118.4

如何检查常规 pod 是否可以解析特定的 DNS 名称:

$ kubectl run dnsutils -it --rm=true --restart=Never --image=tutum/dnsutils cat /etc/resolv.conf

nameserver 10.96.0.10
search default.svc.cluster.local svc.cluster.local cluster.local 
options ndots:5

$ kubectl run dnsutils -it --rm=true --restart=Never --image=tutum/dnsutils nslookup github.com

Server:         10.96.0.10
Address:        10.96.0.10#53

Non-authoritative answer:
Name:   github.com
Address: 140.82.118.3

有关 DNS 故障排除的更多详细信息,请参阅 official documentation