K8S-DNS-003 kubernetes network_error ai_generated true

plugin/loop: Loop (127.0.0.1:53) detected for zone ".", see https://coredns.io/plugins/loop#troubleshooting

ID: kubernetes/coredns-loop-detected

Also available as: JSON · Markdown · 中文
85%Fix Rate
87%Confidence
1Evidence
2023-04-05First Seen

Version Compatibility

VersionStatusIntroducedDeprecatedNotes
coredns 1.8.0 active — — —
coredns 1.9.0 active — — —
coredns 1.10.0 active — — —
kubernetes 1.22 active — — —
kubernetes 1.24 active — — —

Root Cause

CoreDNS is forwarding DNS queries to itself, creating a loop, usually because the node's /etc/resolv.conf points to localhost or the pod's DNS policy misconfigures upstream.

generic

中文

CoreDNS 将 DNS 查询转发给自己，造成循环，通常是因为节点的 /etc/resolv.conf 指向 localhost 或 Pod 的 DNS 策略配置了错误的上游。

Official Documentation

https://coredns.io/plugins/loop/

Workarounds

  1. 90% success Edit the CoreDNS ConfigMap (kube-system/coredns) to add the `loop` plugin and set a proper upstream DNS. For example, ensure the forward directive points to a valid external DNS (e.g., `forward . /etc/resolv.conf`) and not to localhost. 
    Edit the CoreDNS ConfigMap (kube-system/coredns) to add the `loop` plugin and set a proper upstream DNS. For example, ensure the forward directive points to a valid external DNS (e.g., `forward . /etc/resolv.conf`) and not to localhost.
  2. 85% success Check the node's /etc/resolv.conf for nameserver entries pointing to 127.0.0.1 or loopback addresses. If found, change them to a valid DNS server IP (e.g., 8.8.8.8) and restart kubelet. 
    Check the node's /etc/resolv.conf for nameserver entries pointing to 127.0.0.1 or loopback addresses. If found, change them to a valid DNS server IP (e.g., 8.8.8.8) and restart kubelet.

中文步骤

  1. 编辑 CoreDNS ConfigMap（kube-system/coredns）以添加 `loop` 插件并设置正确的上游 DNS。例如，确保 forward 指令指向有效的外部 DNS（例如 `forward . /etc/resolv.conf`），而不是 localhost。
  2. 检查节点的 /etc/resolv.conf 中是否有指向 127.0.0.1 或回环地址的 nameserver 条目。如果有，将其更改为有效的 DNS 服务器 IP（例如 8.8.8.8）并重启 kubelet。

Dead Ends

Common approaches that don't work:

  1. Delete the CoreDNS pod and let it restart. 95% fail

    The loop is caused by configuration, not a transient state; restarting will not fix the underlying issue.

  2. Set the node's /etc/resolv.conf to use a public DNS like 8.8.8.8. 60% fail

    This may break cluster-internal DNS resolution for services and pods.