5 Tips for Resolving Specific Domain Names Inside Containers

Follow our public account 「Wonderful Linux World」

Set as 「Starred」, bringing you the best of Linux every day!

In this article, we will explore several ways to specify the resolution results of specific domain names inside containers. To facilitate demonstration, we first create a Deployment configuration file.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: busybox-deployment
  labels:
    app: busybox
spec:
  replicas: 1
  selector:
    matchLabels:
      app: busybox
  template:
    metadata:
      labels:
        app: busybox
    spec:
      containers:
      - name: busybox
        image: busybox
        args:
        - /bin/sh
        - -c
        - "while true; do echo Hello, Kubernetes!; sleep 10;done"

This deployment will create one busybox pod, and the container will print “Hello, Kubernetes!” to the console every 10 seconds.

TL;DR

Modify /etc/hosts

Modifying /etc/hosts is the most traditional method, directly changing the corresponding file inside the container to achieve domain name resolution, effective at the Pod level. Due to its poor maintainability (it needs to be manually modified every time the pod restarts), it is not recommended for production environments.

For example, we can add a record like this in /etc/hosts

250.250.250.250 four-250

/ # ping four-250
PING four-250 (250.250.250.250): 56 data bytes

Add HostAliases Records

HostAliases is a field in Kubernetes Pod configuration that provides additional records for the /etc/hosts file of the Pod container. This can be very useful in certain situations, especially when you want to override the resolution results of a hostname or provide resolution for hostnames that are not available in the network.

This can be modified at the Pod, Replica, Deployment, or StatefulSet level, offering slightly better maintainability. For example, we can modify the above yaml to

apiVersion: apps/v1
kind: Deployment
metadata:
  name: busybox-deployment
  labels:
    app: busybox
spec:
  replicas: 3
  selector:
    matchLabels:
      app: busybox
  template:
    metadata:
      labels:
        app: busybox
    spec:
      hostAliases:
      - ip: "250.250.250.250"
        hostnames:
        - "four-250"
      containers:
      - name: busybox
        image: busybox
        args:
        - /bin/sh
        - -c
        - "while true; do echo Hello, Kubernetes!; sleep 10;done"

At this point, when we check the container’s /etc/hosts, we find that Kubernetes has automatically inserted a record Entries added by HostAliases. This is the implementation principle of HostAliases.

A write action is performed in the kubelet_pods code

func hostsEntriesFromHostAliases(hostAliases []v1.HostAlias) []byte {
 if len(hostAliases) == 0 {
  return []byte{}
 }

 var buffer bytes.Buffer
 buffer.WriteString("\n")
 buffer.WriteString("# Entries added by HostAliases.\n")
 // for each IP, write all aliases onto single line in hosts file
 for _, hostAlias := range hostAliases {
  buffer.WriteString(fmt.Sprintf("%s\t%s\n", hostAlias.IP, strings.Join(hostAlias.Hostnames, "\t")))
 }
 return buffer.Bytes()
}

Coredns Configuration

We can modify the ConfigMap to achieve the purpose of having the container resolve specific domain names.

Change Coredns Configuration

We can modify the Coredns configuration with the following command:

kubectl edit cm coredns -n kube-system

Existing ConfigMap

Corefile: |
    .:53 {
        log
        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
        hosts {
           192.168.65.2 host.minikube.internal
           fallthrough
        }
        forward . /etc/resolv.conf {
           max_concurrent 1000
        }
        cache 30
        loop
        reload
        loadbalance
    }

Add specific records in hosts

250.250.250.250 four-250

If you have not configured the reload plugin, you will need to restart Coredns for the changes to take effect. The default reload time is 30 seconds, defined in the defaultInterval of plugin/reload/setup.go.

Custom DNS Policy

By modifying the DNS policy, you can direct the resolution of specific domain names for individual Pods/Deploys/StatefulSets to specific servers, as shown below, where you can add DNS servers and search domains to the pod

spec:
      dnsConfig:
        nameservers:
          - 1.2.3.4
        searches:
          - search.prefix
      containers:
      - name: busybox
        image: busybox
        args:
        - /bin/sh
        - -c
        - "while true; do echo Hello, Kubernetes!; sleep 10;done"

Using Third-Party DNS Plugins

Not recommended, using other DNS plugins for some fancy custom operations. Currently, Coredns is the mainstream in the industry and has no good alternatives.

This article is reproduced from: 「Huawei Cloud Community」, original text: https://url.hi-linux.com/vLxOd, copyright belongs to the original author. Welcome to contribute, submission email: [email protected].

🚀 Recently, we have established a technical exchange WeChat group. Many experts in the industry have already joined. Interested students can join us for technical exchange by replying directly to 「Join Group」 in the 「Wonderful Linux World」 public account.

📕 Follow「Wonderful Linux World」 public account to start an interesting new life! For more useful and fun software resources, visit https://666666.dev for free.

You might also like

Click the image below to read

Practical tips for gracefully shutting down Node, Java, Python, and other application containers in DockerClick the image above, ‘Meituan|Ele.me’ delivery red envelopes are available for free every day

For more interesting fresh news from the internet, follow the 「Wonderful Internet」 video account to learn all about it!