This is a cache of https://docs.openshift.com/acs/3.74/operating/manage-network-policies.html. It is a snapshot of the page at 2024-11-29T17:40:47.153+0000.
Managing network policies | Operating | Red Hat Advanced Cluster Security for Kubernetes 3.74
    OpenShift docs are moving and will soon only be available at docs.redhat.com, the home of all Red Hat product documentation. Explore the new docs experience today.
  1. Documentation
    2. history
×

Managing network policies

A Kubernetes network policy is a specification of how groups of pods are allowed to communicate with each other and other network endpoints. These network policies are configured as YAML files. By looking at these files alone, it is often hard to identify whether the applied network policies achieve the desired network topology.

Red Hat Advanced Cluster Security for Kubernetes gathers all defined network policies from your orchestrator and provides functionality to make these policies easier to use.

To support network policy enforcement, Red Hat Advanced Cluster Security for Kubernetes provides:

  • Network graph

  • Network policy simulator

  • Network policy generator

  • Build-time network policy generator

Network graph view

The network graph provides visibility and control over:

  • The allowed network connections, as defined by the Kubernetes network policies.

  • The active communications paths among namespaces and deployments.

In the menu bar, choose the cluster for which you want to view information and at least one namespace.

In the Network Graph view, you can configure the type of connections you want to see. In the Flows section (upper left), select:

  • Active to view only active connections.

  • Allowed to view only allowed network connections.

  • All to view both active and allowed network connections.

In the Network Graph, click Legend to view information about the symbols in use and their meaning. The legend shows explanatory text for symbols representing namespaces, deployments, and connections on the network graph.

Clicking on an item in the Network Graph, such as a deployment or a namespace, opens a window that displays additional information. You can expand and collapse the window by selecting the arrow in the blue bar.

When you hover over:

  • a connection, you see information about the network flow, which includes active connections, port numbers, and protocols in use.

  • a deployment, you see information about ingress and egress connections, protocols, port numbers in use, and the direction of the network traffic between deployments.

Allowed network connections

Red Hat Advanced Cluster Security for Kubernetes processes all network policies in each secured cluster to show you which deployments can contact each other and which can reach external networks.

The network graph shows possible network connections as dashed lines.

Actual network flows

Red Hat Advanced Cluster Security for Kubernetes monitors running deployments and tracks traffic between them. The network graph shows observed network flows as solid lines.

Network baseline

Red Hat Advanced Cluster Security for Kubernetes discovers existing network flows and creates a baseline.

To view the network baseline for a deployment, select that deployment in the Network Graph view. The Network Flows details panel show both anomalous and baseline flows. From this panel, you can:

  • Mark network flows from the baseline as anomalous by selecting Mark as Anomalous.

  • Add network flows to the baseline from the anomalous flows by selecting Add to Baseline.

External entities and connections

The Network Graph view shows network connections between managed clusters and external sources. In addition, Red Hat Advanced Cluster Security for Kubernetes automatically discovers and highlights public Classless Inter-Domain Routing (CIDR) address blocks, such as Google Cloud, AWS, Azure, Oracle Cloud, and Cloudflare. Using this information, you can identify deployments with active external connections and determine if they are making or receiving unauthorized connections from outside your network.

By default, the external connections point to a common External Entities box and different CIDR address blocks in the Network Graph view. However, you can choose not to show auto-discovered CIDR blocks.

Red Hat Advanced Cluster Security for Kubernetes includes IP ranges for the following cloud providers:

  • Google Cloud

  • AWS

  • Azure

  • Oracle Cloud

  • Cloudflare

Red Hat Advanced Cluster Security for Kubernetes fetches and updates the cloud providers' IP ranges every 7 days. If you are using offline mode, you can update these ranges by installing new support packages.

Additional resources

Viewing network policies

Network policies specify how groups of pods are allowed to communicate with each other and with other network endpoints. Kubernetes NetworkPolicy resources use labels to select pods and define rules that specify what traffic is allowed to or from the selected pods. Red Hat Advanced Cluster Security for Kubernetes discovers and displays network policy information for all your Kubernetes clusters, namespaces, deployments, and pods, in the Network Graph view.

To view network policies and other related details for deployments in a namespace, you can select the namespace in Network Graph view.

The namespace details panel lists all deployments in the selected namespace. You can then hover over a deployment in the details panel and select the Navigate to deployment (arrow) icon that appears on the right to view deployment details.

Alternatively, you can directly select a deployment in the Network Graph view to view its details. The deployment details panel includes the Network Flows, Details, and Network Policies tabs.

You can select each tab to view related information.

  • The Network Flows tab shows information about ingress and egress connections, protocols, and port numbers in use for that deployment.

  • The Details tab shows information about how the service is deployed, including orchestrator labels and annotations.

  • The Network Policies tab shows information about every network policy that applies to the deployment.

You need Red Hat Advanced Cluster Security for Kubernetes version 3.0.47 or newer to see information about ingress and egress connections, protocols, port numbers, and the direction of the network traffic.

Configuring CIDR blocks

You can specify custom CIDR blocks or configure displaying auto-discovered CIDR block in the Network Graph view.

Procedure

  1. In the Network Graph view, select Configure CIDR Blocks.

  2. Turn the Display auto-discovered CIDR blocks in Network Graph toggle off to hide auto-discovered CIDR blocks.

    When you hide the auto-discovered CIDR blocks, the auto-discovered CIDR blocks are hidden for all clusters, not only for the selected cluster on the top bar in the Network Graph view.

  3. Add custom CIDR addresses by adding CIDR Block Name and CIDR Address. To add more than one, select the Add icon.

  4. Click Update Configuration to save the changes.

Simulating network policies

Your current network policies might allow unneeded network communications. To simulate the effects of a new set of network policies, use the network policy simulator.

Procedure

  1. On the RHACS portal, select Network Graph from the left-hand navigation menu.

  2. Select one or more namespaces.

  3. On the Network Graph view header, select Network Policy Simulator.

  4. Select Upload and simulate network policy YAML and upload your proposed YAML file. The network graph view displays what your proposed network policies would achieve.

  5. To share your proposed policies with your team, select Share YAML. To apply your policy directly, select Apply Network Policies.

Directly applying network policies might cause problems for running applications. Always download and test the network policies in a development environment or test clusters before applying them to production workloads.

Generating network policies

A Kubernetes network policy controls which pods receive incoming network traffic, and which pods can send outgoing traffic. By using network policies to enable and disable traffic to or from pods, you can limit your network attack surface.

These network policies are YAML configuration files. It is often difficult to gain insights into the network flow and manually create these files. Red Hat Advanced Cluster Security for Kubernetes allows you to autogenerate these network policies based on the actual observed network communication flows in your environment.

You can generate network policies from the network graph view.

The generated policies apply to the deployments shown in the network graph and they allow all network traffic observed during the selected time.

Procedure

  1. In the RHACS portal, select Network Graph from the left-hand navigation menu.

  2. Select a cluster name from the menu on the top bar if the correct one is not already selected.

  3. Select one or more namespaces.

  4. If you want to generate policies for only some deployments, use the Add one or more deployment filters field to add criteria by which to filter deployments. If you do not add a filter, Red Hat Advanced Cluster Security for Kubernetes generates policies for all deployments in the cluster.

  5. Select an appropriate time from the menu on the top bar. If the selected time is too short it leaves out periodic or infrequent network communications.

  6. Select Network Policy Simulator.

  7. In the panel that opens, select Exclude ports & protocols if you do not want ports and protocols to be scoped in Red Hat Advanced Cluster Security for Kubernetes generated policies.

  8. Select Generate and simulate network policies. The generated network policy configuration YAML opens in the same panel, and the network graph shows the effects of the policies.

Saving generated policies

You can download and save the generated network policies from Red Hat Advanced Cluster Security for Kubernetes. Use this option to commit the policies into a version control system like Git.

Procedure

  • Select the Download YAML icon on the Network Policy Simulator panel.

Testing generated policies

After you download the network policies that Red Hat Advanced Cluster Security for Kubernetes generates, you can test them by applying them to your cluster.

Procedure

  1. To create policies using the saved YAML file, use the following command:

    $ oc create -f "<generated_file>.yaml" (1)
    1 If you use Kubernetes, enter kubectl instead of oc.

  2. If the generated policies cause problems, you can remove them by running the following command:

    $ oc delete -f "<generated_file>.yaml" (1)
    1 If you use Kubernetes, enter kubectl instead of oc.

Applying generated policies

You can apply the generated network policies from the RHACS portal.

Procedure

  • To directly apply the generated policies in the cluster from within Red Hat Advanced Cluster Security for Kubernetes, select Apply Network Policies.

Directly applying network policies might cause problems for running applications. Always download and test the network policies in a development environment or test clusters before applying them to production workloads.

Deleting generated policies

If you have applied generated policies directly and want to remove them, select the Revert most recently applied YAML icon on the Network Policy Simulator panel.

Procedure

  1. In the RHACS portal, select Network Graph from the left-hand navigation menu.

  2. Select a cluster name from the menu on the top bar if the correct one is not already selected.

  3. Select one or more namespaces.

  4. Select Network Policy Simulator.

  5. Select View active YAMLS.

  6. Select the Revert most recently applied YAML icon.

Deleting all autogenerated policies

You can delete all generated policies from your cluster that you have created by using Red Hat Advanced Cluster Security for Kubernetes.

Procedure

  • Run the following command:

    • On OpenShift Container Platform:

      $ oc get ns -o jsonpath='{.items[*].metadata.name}' | xargs -n 1 oc delete networkpolicies -l 'network-policy-generator.stackrox.io/generated=true' -n

    • On Kubernetes:

      $ kubectl get ns -o jsonpath='{.items[*].metadata.name}' | xargs -n 1 kubectl delete networkpolicies -l 'network-policy-generator.stackrox.io/generated=true' -n

Policy generation strategy

When you autogenerate network policies:

  • Red Hat Advanced Cluster Security for Kubernetes generates a single network policy for each deployment in the namespace. The pod selector for the policy is the pod selector of the deployment.

    • If a deployment already has a network policy, Red Hat Advanced Cluster Security for Kubernetes does not generate new policies or delete existing policies.

  • Generated policies only restrict traffic to existing deployments.

    • Deployments you create later will not have any restrictions unless you create or generate new network policies for them.

    • If a new deployment needs to contact a deployment with a network policy, you might need to edit the network policy to allow access.

  • Each policy has the same name as the deployment name, prefixed with stackrox-generated-. For example, the policy name for the deployment depABC in the generated network policy is stackrox-generated-depABC. All generated policies also have an identifying label.

  • Red Hat Advanced Cluster Security for Kubernetes generates a single rule allowing traffic from any IP address if:

    • The deployment has an incoming connection from outside the cluster within the selected time, or

    • The deployment is exposed through a node port or load balancer service.

  • Red Hat Advanced Cluster Security for Kubernetes generates one ingress rule for every deployment from which there is an incoming connection.

    • For deployments in the same namespace, this rule uses the pod selector labels from the other deployment.

    • For deployments in different namespaces, this rule uses a namespace selector. To make this possible, Red Hat Advanced Cluster Security for Kubernetes automatically adds a label, namespace.metadata.stackrox.io/name, to each namespace.

In rare cases, if a standalone pod does not have any labels, the generated policy allows traffic from or to the pod’s entire namespace.

Using build-time network policy generator

Build-time network policy generation is a Technology Preview feature only. Technology Preview features are not supported with Red Hat production service level agreements (SLAs) and might not be functionally complete. Red Hat does not recommend using them in production. These features provide early access to upcoming product features, enabling customers to test functionality and provide feedback during the development process.

For more information about the support scope of Red Hat Technology Preview features, see Technology Preview Features Support Scope.

The build-time network policy generator can automatically generate Kubernetes network policies based on application YAML manifests. You can use it to develop network policies as part of the CI/CD pipeline before deploying applications on your cluster.

Red Hat developed this feature in partnership with the developers of the NP-Guard project. First, the build-time network policy generator analyzes Kubernetes manifests in a local folder, including service manifests, config maps, and workload manifests such as Pod, Deployment, ReplicaSet, Job, DaemonSet, and StatefulSet. Then, it discovers the required connectivity and creates the Kubernetes network policies to achieve pod isolation. These policies allow no more and no less than the needed ingress and egress traffic.

Generating build-time network policies

The build-time network policy generator is included in the roxctl CLI. For the build-time network policy generation feature, roxctl CLI does not need to communicate with RHACS Central so you can use it in any development environment.

Prerequisites

  1. The build-time network policy generator recursively scans the directory you specify when you run the command. Therefore, before you run the command, you must already have service manifests, config maps, and workload manifests such as Pod, Deployment, ReplicaSet, Job, DaemonSet, and StatefulSet as YAML files in the specified directory.

  2. Verify that you can apply these YAML files as-is using the kubectl apply -f command. The build-time network policy generator does not work with files that use Helm style templating.

  3. Verify that the service network addresses are not hard-coded. Every workload that needs to connect to a service must specify the service network address as a variable. You can specify this variable by using the workload’s resource environment variable or in a config map.

  4. service network addresses must match the following official regular expression pattern:

    (http(s)?://)?<svc>(.<ns>(.svc.cluster.local)?)?(:<portNum>)? (1)
    1 In this pattern,

    • <svc> is the service name.

    • <ns> is the namespace where you defined the service.

    • <portNum> is the exposed service port number.

    Following are some examples that match the pattern:

    • wordpress-mysql:3306

    • redis-follower.redis.svc.cluster.local:6379

    • redis-leader.redis

    • http://rating-service.

Procedure

  1. Verify that the build-time network policy generation feature is available by running the help command:

    $ roxctl generate netpol -h

  2. Generate the policies by using the generate netpol command:

    $ roxctl generate netpol <folder-path> (1)
    1 Specify the path of the folder that has the Kubernetes manifests.

The roxctl generate netpol command supports the following options:

Option

Description

-h, --help

View the help text for the netpol command.

-d, --output-dir <dir>

Save the generated policies into a target folder. One file per policy.

-f, --output-file <filename>

Save and merge the generated policies into a single YAML file.

--fail

Fail on the first encountered error. The default value is false.

--remove

Remove the output path if it already exist.

--strict

Treat warnings as errors. The default value is false.

After generating the policies, you must inspect them for completeness and accuracy, in case any relevant network address was not specified as expected in the YAML files. Most importantly, verify that required connections are not blocked by the isolating policies. To help with this inspection, you can use RHACS to simulate the generated network policies .

Red Hat recommends applying network policies to the cluster as part of the workload deployment using automation. You can follow a GitOps approach by submitting the generated policies using Pull Requests, providing the team an opportunity to review the policies before deploying them as part of the pipeline.

Using network baselining

In Red Hat Advanced Cluster Security for Kubernetes, you can minimize your risks by using network baselining. It is a proactive approach to keep your infrastructure secure. Red Hat Advanced Cluster Security for Kubernetes first discovers existing network flows and creates a baseline, and then it treats network flows outside of this baseline as anomalous.

  • To use the Network baseline feature, you must use the Red Hat Advanced Cluster Security for Kubernetes version 3.0.54 or newer.

  • To enable alerts on baseline violations, you must use Red Hat Advanced Cluster Security for Kubernetes version 3.0.56 or newer.

When you install Red Hat Advanced Cluster Security for Kubernetes, there is no default network baseline. As Red Hat Advanced Cluster Security for Kubernetes discovers network flows, it creates a baseline and then it adds all discovered network flows to it.

  • When Red Hat Advanced Cluster Security for Kubernetes discovers new network activity, it adds that network flow to the network baseline.

  • Network flows do not show up as anomalous flows and do not trigger any violations.

After the discovery phase:

  • Red Hat Advanced Cluster Security for Kubernetes stops adding network flows to the network baselines.

  • New network flows (which are not in the network baseline) show up as anomalous flows but they do not trigger any violations.

Viewing network baselines

You can view network baselines from the network graph view.

Procedure

  1. On the Network Graph view, select one or more namespaces.

  2. Select a deployment.

  3. The Network Flows details panel show both anomalous and baseline flows. From this panel, you can:

    • mark network flows from the baseline as anomalous by selecting Mark as Anomalous, or

    • add network flows to baseline from the anomalous flows by selecting Add to Baseline.

Enabling alerts on baseline violations

You can configure Red Hat Advanced Cluster Security for Kubernetes to detect anomalous network flows and trigger violations.

You need Red Hat Advanced Cluster Security for Kubernetes version 3.0.56 or newer to enable alerts on baseline violations.
Procedure

  1. On the Network Graph view, select a deployment.

  2. In the network flow details panel, select Baseline Settings.

  3. Turn on the Alert on baseline violations toggle.

    • Once you turn on the Alert on baseline violations toggle anomalous network flows trigger violations.

    • You can turn off the Alert on baseline violations toggle to stop receiving violations for anomalous network flows.

Configuring network baselining time frame

You can use the ROX_NETWORK_BASELINE_OBSERVATION_PERIOD and the ROX_BASELINE_GENERATION_DURATION environment variables to configure the observation period and the network baseline generation duration.

Procedure

  • Set the ROX_NETWORK_BASELINE_OBSERVATION_PERIOD and the ROX_BASELINE_GENERATION_DURATION environment variables:

    $ oc -n stackrox set env deploy/central \ (1)
  ROX_NETWORK_BASELINE_OBSERVATION_PERIOD=<value> (2)
    1 If you use Kubernetes, enter kubectl instead of oc.
    2 Value must be time units, for example: 300ms, -1.5h, or 2h45m. Valid time units are ns, us or µs, ms, s, m, h. 
    $ oc -n stackrox set env deploy/central \ (1)
  ROX_BASELINE_GENERATION_DURATION=<value> (2)
    1 If you use Kubernetes, enter kubectl instead of oc.
    2 Value must be time units, for example: 300ms, -1.5h, or 2h45m. Valid time units are ns, us or µs, ms, s, m, h.