coredns不断重启，CrashLoopBackOff, [FATAL] plugin loop

【环境信息】

Ubuntu：16.04.6，桌面版本，x86 Atals800-3010

内核：kernel 4.15.0-45

docker版本：18.06.1

k8s: 1.12.2-00

【现象】

安装完后，查看到coredns没有起来，不断重启，CrashLoopBackOff,如下所示：

kubectl get pod --all-namespace

root@ubuntu-154:/home/# kubectl get pod --all-namespaces
NAMESPACE     NAME                                     READY   STATUS             RESTARTS   AGE
kube-system   calico-kube-controllers-bc8f7d57-rmtlw   1/1     Running            0          5m18s
kube-system   calico-node-zgdnw                        1/1     Running            0          5m18s
kube-system   coredns-576cbf47c7-22r69                 0/1     CrashLoopBackOff   3          78s
kube-system   coredns-576cbf47c7-9546j                 0/1     CrashLoopBackOff   3          57s
kube-system   etcd-ubuntu-154                          1/1     Running            0          6m46s
kube-system   kube-apiserver-ubuntu-154                1/1     Running            0          6m34s
kube-system   kube-controller-manager-ubuntu-154       1/1     Running            0          7m3s
kube-system   kube-proxy-2t4ld                         1/1     Running            0          7m38s
kube-system   kube-scheduler-ubuntu-154                1/1     Running            0          6m40s

查看pod的日志：

kubectl logs -n kube-system coredns-576cbf47c7-9546j

root@ubuntu-154:/home/# kubectl logs -n kube-system   coredns-576cbf47c7-9546j
.:53
2021/04/17 10:18:36 [INFO] CoreDNS-1.2.2
2021/04/17 10:18:36 [INFO] linux/amd64, go1.11, eb51e8b
CoreDNS-1.2.2
linux/amd64, go1.11, eb51e8b
2021/04/17 10:18:36 [INFO] plugin/reload: Running configuration MD5 = f65c4821c8a9b7b5eb30fa4fbc167769
2021/04/17 10:18:36 [FATAL] plugin/loop: Seen "HINFO IN 3162287063793932269.6822300986019194427." more than twice, loop detected

显示：[FATAL] plugin/loop

【原因】

根据报错，查看到对应cm的配置：

root@ubuntu-154:~# kubectl edit cm coredns -n kube-system

# Please edit the object below. Lines beginning with a '#' will be ignored,
# and an empty file will abort the edit. If an error occurs while saving this file will be
# reopened with the relevant failures.
#
apiVersion: v1
data:
  Corefile: |
    .:53 {
        errors
        health {
           lameduck 5s
        }
        ready
        kubernetes cluster.local in-addr.arpa ip6.arpa {
           pods insecure
           fallthrough in-addr.arpa ip6.arpa
           ttl 30
        }
        prometheus :9153
        forward . /etc/resolv.conf
        cache 30
        loop
        reload
        loadbalance
    }
kind: ConfigMap
metadata:
  creationTimestamp: "2021-04-08T06:33:43Z"
  name: coredns
  namespace: kube-system
  resourceVersion: "182"
  selfLink: /api/v1/namespaces/kube-system/configmaps/coredns
  uid: d037bd15-44e4-42c4-ac15-cec75f948e3a

官网的意思是：

CoreDNS启动后会通过宿主机的resolv.conf文件去获取上游DNS的信息，如果这个时候获取的DNS的服务器是本地地址的话，就会出现环路，从而被环路识别器识别出来。

【规避】

我在k8s的1.17.3这样配置没有问题，在1.12上出现了。删除loop后（:wq保存退出）,问题解决。

root@ubuntu-154:/home# kubectl get pod --all-namespaces
NAMESPACE     NAME                                     READY   STATUS    RESTARTS   AGE
kube-system   calico-kube-controllers-bc8f7d57-vsrzx   1/1     Running   0          41h
kube-system   calico-node-grxnm                        1/1     Running   0          41h
kube-system   coredns-576cbf47c7-9xv4q                 1/1     Running   0          41h
kube-system   coredns-576cbf47c7-bfxg9                 1/1     Running   0          41h
kube-system   etcd-ubuntu-154                          1/1     Running   0          41h
kube-system   kube-apiserver-ubuntu-154                1/1     Running   0          41h
kube-system   kube-controller-manager-ubuntu-154       1/1     Running   0          41h
kube-system   kube-proxy-2t4ld                         1/1     Running   0          41h
kube-system   kube-scheduler-ubuntu-154                1/1     Running   0          41h

【参考】

https://blog.csdn.net/weixin_43908373/article/details/84833783

https://blog.csdn.net/crystonesc/article/details/83387659