Configuring an IP masquerade agent in Standard clusters

Standard

---

This page explains how to configure clusters created in Google Kubernetes Engine (GKE) Standard mode to perform IP masquerade with the ip-masq-agent. For more information about IP masquerading in GKE Autopilot mode, see Use Egress NAT Policy to configure IP masquerade in Autopilot clusters.

Before you begin

Before you start, make sure you have performed the following tasks:

• Enable the Google Kubernetes Engine API.
Enable Google Kubernetes Engine API
• If you want to use the Google Cloud CLI for this task, install and then initialize the gcloud CLI. If you previously installed the gcloud CLI, get the latest version by running gcloud components update.
  Note: For existing gcloud CLI installations, make sure to set the compute/region and compute/zone properties. By setting default locations, you can avoid errors in gcloud CLI like the following: One of [--zone, --region] must be supplied: Please specify location.

Checking ip-masq-agent status

This section shows you how to:

• Determine whether your cluster has an ip-masq-agent DaemonSet.
• Check the ip-masq-agent ConfigMap resource.

Check the ip-masq-agent DaemonSet

To check if your cluster is running the ip-masq-agent DaemonSet, use the Google Cloud CLI or the Google Cloud console.

gcloud

1. Get the credentials for your cluster:

   gcloud container clusters get-credentials CLUSTER_NAME
   

   Replace CLUSTER_NAME with the name of your cluster.

2. Search for the ip-masq-agent in the kube-system namespace:

   kubectl get daemonsets/ip-masq-agent -n kube-system
   

   If the ip-masq-agent DaemonSet exists, then the output is similar to the following:

   NAME            DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR   AGE
   ip-masq-agent   3         3         3       3            3           <none>          13d
   

   If the ip-masq-agent DaemonSet does not exist, then the output is similar to the following:

   Error from server (NotFound): daemonsets.apps "ip-masq-agent" not found
   

3. Check if the ip-masq-agent DaemonSet is running the latest version:

   kubectl get daemonsets/ip-masq-agent -n kube-system -o jsonpath='{.spec.template.spec.containers[].image}'
   

   This command must return the same container image as specified in Deploying the ip-masq-agent DaemonSet.

Console

1. Go to the Workloads page in the Google Cloud console.

   Go to Workloads

2. For Filter, do the following:

   1. Click to clear the Is system object: False filter.
   2. Filter the following properties:
      • Name: ip-masq-agent.
      • Cluster: the name of your cluster.

   If ip-masq-agent DaemonSet exists, then you can see the DaemonSet record in the table. If ip-masq-agent DaemonSet does not exist, then no rows are displayed.

To create the ip-masq-agent ConfigMap and deploy the ip-masq-agent DaemonSet, refer to Configuring and deploying the ip-masq-agent.

Checking the ip-masq-agent ConfigMap

To check if your cluster is running the ip-masq-agent ConfigMap, use the Google Cloud CLI or the Google Cloud console.

gcloud

1. Get the credentials for your cluster:

   gcloud container clusters get-credentials CLUSTER_NAME
   

   Replace CLUSTER_NAME with the name of your cluster.

2. Describe the ip-masq-agent ConfigMap in the kube-system namespace:

   kubectl describe configmaps/ip-masq-agent -n kube-system
   

   If the ip-masq-agent ConfigMap exists, then the output is similar to the following:

   Name:         ip-masq-agent
   Namespace:    kube-system
   Labels:       <none>
   Annotations:  <none>
   
   Data
   ====
   config:
   ----
   nonMasqueradeCIDRs:
     - 198.15.5.92/24
     - 10.0.0.0/8
   masqLinkLocal: false
   resyncInterval: 60s
   
   BinaryData
   ====
   
   Events:  <none>
   

   If the ip-masq-agent ConfigMap does not exist, then the output is similar to the following:

   Error from server (NotFound): configmaps "ip-masq-agent" not found
   

Console

1. Go to the Configuration page in the Google Cloud console.

   Go to Configuration

2. For Filter, do the following:

   1. Click to clear the Is system object: False filter.
   2. Filter the following properties:
      • Name: ip-masq-agent.
      • Cluster: the name of your cluster.

   If ip-masq-agent ConfigMap exists, then you can see the ConfigMap record in the table. If ip-masq-agent ConfigMap does not exist, then no rows are displayed.

If the cluster already has the ip-masq-agent ConfigMap, you can configure and deploy it.

Configuring and deploying the ip-masq-agent

This section shows you how to create or edit the ip-masq-agent ConfigMap and how to deploy or delete the ip-masq-agent DaemonSet. To determine which tasks you need to perform, you must first determine whether your cluster already has the ip-masq-agent ConfigMap and ip-masq-agent DaemonSet.

Creating the ip-masq-agent ConfigMap

The following steps show how to create the ip-masq-agent ConfigMap. If your cluster already has the ip-masq-agent ConfigMap, edit an existing ip-masq-agent ConfigMap instead.

1. Create a configuration file using the following template and save it locally. You can use any name for the local copy of this configuration file.

   nonMasqueradeCIDRs:
     - CIDR_1
     - CIDR_2
   masqLinkLocal: false
   resyncInterval: SYNC_INTERVALUNIT_OF_TIME
   

   Replace the following:

   • CIDR_1 and CIDR_2: the IP address ranges in CIDR format. When packets are sent to these destinations, your cluster does not masquerade IP address sources and preserves source Pod IP addresses. If you need more than two CIDRs, add more entries to the nonMasqueradeCIDRs list following the same format. At minimum, the nonMasqueradeCIDRs property should include the node and Pod IP address ranges of your cluster.

   • SYNC_INTERVAL: the amount of time after which each ip-masq-agent Pod checks the contents of the ip-masq-agent ConfigMap and writes any changes to its local /etc/config/ip-masq-agent file. Defaults to 60.

   • UNIT_OF_TIME: the unit of time for the resyncInterval. Valid values include s (for seconds) or ms (for milliseconds). Defaults to s.

   Set masqLinkLocal to false (the default) unless you need to enable masquerading for packets sent to link local IPv4 addresses. For more information, see Masquerading to link-local destinations.

   Caution: A custom nonMasqueradeCIDRs must at least include the cluster's node and Pod IP address range(s). This is especially important if you enable intranode visibility. If the Pod IP address range(s) are not present in nonMasqueradeCIDRs and you have intranode visibility turned on, Pods won't be able to properly communicate with each other.

2. Create the ConfigMap resource:

   kubectl create configmap ip-masq-agent \
      --namespace=kube-system \
      --from-file=config=LOCAL_CONFIG_FILE_PATH
   

   Replace LOCAL_CONFIG_FILE_PATH with the path to the config file you created in the previous step.

   Note: The ConfigMap content must be associated with a named key that the ip-masq-agent DaemonSet expects when reading the ip-masq-agent ConfigMap to parse it as a volume. In the following example, and for standard ip-masq-agent DaemonSets, the key is config. Specifying --from-file=config=LOCAL_CONFIG_FILE_PATH ensures that the IP masquerade configuration information is associated with the config key.

3. Describe the ip-masq-agent ConfigMap in the kube-system namespace:

   kubectl describe configmaps/ip-masq-agent -n kube-system
   

   The output is similar to the following:

   Name:         ip-masq-agent
   Namespace:    kube-system
   Labels:       <none>
   Annotations:  <none>
   
   Data
   ====
   config:
   ----
   nonMasqueradeCIDRs:
     - 198.15.5.92/24
     - 10.0.0.0/8
   masqLinkLocal: false
   resyncInterval: 60s
   
   BinaryData
   ====
   Events:  <none>
   
   

   This output includes the config parameter with your configuration changes. You can now deploy the ip-masq-agent DeamonSet.

Editing an existing ip-masq-agent ConfigMap

You can modify the contents of an existing ip-masq-agent ConfigMap by completing the following steps:

1. Open the ConfigMap in a text editor:

   kubectl edit configmap ip-masq-agent --namespace=kube-system
   

2. Edit the content of the ConfigMap file:

   apiVersion: v1
   data:
     config: |
       nonMasqueradeCIDRs:
         - CIDR_1
         - CIDR_2
       masqLinkLocal: false
       resyncInterval: SYNC_INTERVALUNIT_OF_TIME
   kind: ConfigMap
   metadata:
     name: ip-masq-agent
     namespace: kube-system
   

   Replace the following:

   • CIDR_1 and CIDR_2: the IP address ranges in CIDR format. When packets are sent to these destinations, your cluster does not masquerade IP address sources and preserves source Pod IP addresses. If you need more than two CIDRs, add more entries to the nonMasqueradeCIDRs list following the same format. At minimum, the nonMasqueradeCIDRs property should include the node and Pod IP address ranges of your cluster.

   • SYNC_INTERVAL: the amount of time after which each ip-masq-agent Pod checks the contents of the ip-masq-agent ConfigMap and writes any changes to its local /etc/config/ip-masq-agent file. Defaults to 60.

   • UNIT_OF_TIME: the unit of time for the resyncInterval. Valid values include s (for seconds) or ms (for milliseconds). Defaults to s.

   Set masqLinkLocal to false (the default) unless you need to enable masquerading for packets sent to link local IPv4 addresses. For more information, see Masquerading to link-local destinations.

   Caution: A custom nonMasqueradeCIDRs must at least include the cluster's node and Pod IP address range(s). This is especially important if you enable intranode visibility. If the Pod IP address range(s) are not present in nonMasqueradeCIDRs and you have intranode visibility turned on, Pods won't be able to properly communicate with each other.

3. Describe the ip-masq-agent ConfigMap in the kube-system namespace:

   kubectl describe configmaps/ip-masq-agent -n kube-system
   

   The output is similar to the following:

   Name:         ip-masq-agent
   Namespace:    kube-system
   Labels:       <none>
   Annotations:  <none>
   
   Data
   ====
   config:
   ----
   nonMasqueradeCIDRs:
     - 198.15.5.92/24
     - 10.0.0.0/8
   masqLinkLocal: false
   resyncInterval: 60s
   
   BinaryData
   ====
   
   Events:  <none>
   

   This output includes the config parameter which matches the configuration value from the file you created.

Deploying the ip-masq-agent DaemonSet

After you create or edit your ip-masq-agent ConfigMap, deploy the ip-masq-agent DaemonSet.

1. Save the following manifest as a YAML file:

   apiVersion: apps/v1
   kind: DaemonSet
   metadata:
     name: ip-masq-agent
     namespace: kube-system
   spec:
     selector:
       matchLabels:
         k8s-app: ip-masq-agent
     template:
       metadata:
         labels:
           k8s-app: ip-masq-agent
       spec:
         hostNetwork: true
         containers:
         - name: ip-masq-agent
           image: gke.gcr.io/ip-masq-agent:v2.12.3-gke.4@sha256:b5db41ddaf863b660da330322714f668101482b528829c50c53229d901d11af5
           args:
           - --v=2
           - --logtostderr=false
           - --log_file=/dev/stdout
           - --log_file_max_size=0
           # The masq-chain must be IP-MASQ
           - --masq-chain=IP-MASQ
           # To non-masquerade reserved IP ranges by default,
           # uncomment the following line.
           # - --nomasq-all-reserved-ranges
           # Must be set to false when using Dataplane V2.
           - --random-fully=false
           securityContext:
             privileged: false
             capabilities:
               drop: ["ALL"]
               add: ["NET_ADMIN", "NET_RAW"]
             allowPrivilegeEscalation: false
             seccompProfile:
               type: RuntimeDefault
           volumeMounts:
           - name: config-volume
             mountPath: /etc/config
         volumes:
         - name: config-volume
           configMap:
             name: ip-masq-agent
             optional: true
             items:
             - key: config
               path: ip-masq-agent
         tolerations:
         - effect: NoSchedule
           operator: Exists
         - effect: NoExecute
           operator: Exists
         - key: "CriticalAddonsOnly"
           operator: "Exists"

   This manifest creates a volume named config-volume which is mounted as specified by the container's volumeMount.

   If you need to edit this manifest, consider the following conditions:

   • The volume name can be anything but must match the container's volumeMount name.

   • The ConfigMap name must match the name of the configMap referenced in the config-volume Volume in the Pod.

   • The name of the chain (--masq-chain) must be IP-MASQ. Otherwise, GKE does not override the default masquerading rules.

   • DaemonSet Pods read from the ip-masq-agent file. The ip-masq-agent file content is the value of the config key in the ConfigMap.

   • If you use non-masquerade reserved IP ranges by default, uncomment the - --nomasq-all-reserved-ranges line in the arg section.

2. Deploy the DaemonSet:

   kubectl apply -f LOCAL_FILE_PATH
   

   Replace LOCAL_FILE_PATH with the path to the file you created in the previous step.

You can manually update the ip-masq-agent DaemonSet you created. To learn more, see Updating a DaemonSet.

Deleting the ip-masq-agent

This section shows you how to delete the ip-masq-agent DaemonSet and the ip-masq-agent ConfigMap. Deleting the ip-masq-agent doesn't revert the existing IP masquerading settings on the nodes.

Deleting the ip-masq-agent DaemonSet

If you manually created the ip-masq-agent DaemonSet, you can delete it by running the following command:

kubectl delete daemonsets ip-masq-agent -n kube-system

Note: If the ip-masq-agent DaemonSet was deployed automatically, GKE will automatically recreate it.

Deleting the ip-masq-agent ConfigMap

To completely delete the ip-masq-agent ConfigMap, run the following command:

kubectl delete configmap ip-masq-agent -n kube-system

Troubleshooting

The following sections provide troubleshooting advice.

General troubleshooting

The following steps help you diagnose issues with IP address masquerading:

• Confirm the status of the ip-masq-agent. If the ConfigMap is not defined, traffic to all default destinations is not masqueraded and preserves the Pod IP address. Traffic to other destinations preserves the node IP address.
• Confirm the version of the ip-masq-agent image specified in the DaemonSet. If the version of the ip-masq-agent DaemonSet is not the latest, follow the deployment steps to update the DaemonSet.
• Check whether the IP-MASQ chain is correctly populated in the NAT IP tables by running the sudo iptables -t nat -L IP-MASQ command in the impacted node. If the nonMasqueradeCIDRs defined in the ConfigMap is not showing up on the NAT IP Tables, verify that there are no typos in the config file that was used to create the ConfigMap.
• Confirm that the destination allows both the node and Pod IP address ranges.
• If traffic is not accessible from the node or Pod, run a connectivity test.

Issue: Pod IP address changes to node IP address

While using an IP masquerade agent, you might notice the Pod's source IP address unexpectedly use the node's IP address when your Pods communicate with external destinations.

The issue is caused by a missing or incomplete custom Source Network Address Translation (SNAT) list. When you use an IP masquerade agent, the default SNAT of the cluster is used when the ConfigMap is missing or it doesn't contain a nonMasqueradeCIDRs list. When a packet leaves a Pod, the default SNAT changes the source IP address from the Pod's IP address to the node's internal IP address. To learn more about SNAT, see IP masquerade agent.

To resolve the issue, configure a custom SNAT list by defining a nonMasqueradeCIDRs list in the ip-masq-agent ConfigMap:

1. Open the ip-masq-agent ConfigMap:

   kubectl edit configmap ip-masq-agent --namespace=kube-system
   

2. Review the nonMasqueradeCIDRs list within the ConfigMap. The nonMasqueradeCIDRs list should be present and complete. For example:

   ...
   nonMasqueradeCIDRs:
     - 35.100.0.0/16
   ...
   

3. If the list is missing or incomplete, add or modify the nonMasqueradeCIDRs list to include the destination IP ranges for which you want to preserve source Pod IP addresses. This list should also include the addresses that you want to use Cloud NAT.

   If you don't want any external traffic to use SNAT, set the nonMasqueradeCIDRs field to 0.0.0.0/0. For example:

   ...
   nonMasqueradeCIDRs:
     - 0.0.0.0/0
   ...
   

   When your traffic doesn't use SNAT, all packets sent from Pods retain the Pod's IP address as their source IP address, regardless of destination.

4. Check the source IP address of outbound packets from your Pods. To check this address, take a packet capture. If that's not feasible, you can check the IP address of the node, which should be the source IP address of outbound packets subject to SNAT, with the following command:

   kubectl get nodes -o wide
   

What's next

• Learn about Alias IPs.
• Read the GKE network overview.
• Learn about configuring authorized networks.