TrafficControl is a CRD API that manages and manipulates the transmission of
Pod traffic. It allows users to mirror or redirect specific traffic originating
from specific Pods or destined for specific Pods to a local network device or a
remote destination via a tunnel of various types. It provides full visibility
into network traffic, including both north-south and east-west traffic.
You may be interested in using this capability if any of the following apply:
-
You want to monitor network traffic passing in or out of a set of Pods for purposes such as troubleshooting, intrusion detection, and so on.
-
You want to redirect network traffic passing in or out of a set of Pods to applications that enforce policies, and reject traffic to prevent intrusion.
This guide demonstrates how to configure TrafficControl to achieve the above
goals.
TrafficControl was introduced in v1.7 as an alpha feature. A feature gate,
TrafficControl must be enabled on the antrea-agent in the antrea-config
ConfigMap for the feature to work, like the following:
kind: ConfigMap
apiVersion: v1
metadata:
name: antrea-config
namespace: kube-system
data:
antrea-agent.conf: |
featureGates:
TrafficControl: trueA TrafficControl in Kubernetes is a REST object. Like all the REST objects, you
can POST a TrafficControl definition to the API server to create a new instance.
For example, supposing you have a set of Pods which contain a label app=web,
the following specification creates a new TrafficControl object named
"mirror-web-app", which mirrors all traffic from or to any Pod with the
app=web label and send them to a receiver running on "10.0.10.2" encapsulated
within a VXLAN tunnel:
apiVersion: crd.antrea.io/v1alpha2
kind: TrafficControl
metadata:
name: mirror-web-app
spec:
appliedTo:
podSelector:
matchLabels:
app: web
direction: Both
action: Mirror
targetPort:
vxlan:
remoteIP: 10.0.10.2The appliedTo field specifies the grouping criteria of Pods to which the
TrafficControl applies to. Pods can be selected cluster-wide using
podSelector. If set with a namespaceSelector, all Pods from Namespaces
selected by the namespaceSelector will be selected. Specific Pods from
specific Namespaces can be selected by providing both a podSelector and a
namespaceSelector. Empty appliedTo selects nothing. The field is mandatory.
The direction field specifies the direction of traffic that should be matched.
It can be Ingress, Egress, or Both.
The action field specifies which action should be taken for the traffic. It
can be Mirror or Redirect. For the Mirror action, targetPort must be
set to the port to which the traffic will be mirrored. For the Redirect
action, both targetPort and returnPort need to be specified, the latter of
which represents the port from which the traffic could be sent back to OVS and
be forwarded to its original destination. Once redirected, a packet should be
either dropped or sent back to OVS without modification, otherwise it would lead
to undefined behavior.
The targetPort field specifies the port to which the traffic should be
redirected or mirrored. There are five kinds of ports that can be used to
receive mirrored traffic:
ovsInternal: This specifies an OVS internal port on all Nodes. A Pod's
traffic will be redirected or mirrored to the OVS internal port on the same Node
that hosts the Pod. The port doesn't need to exist in advance, Antrea will
create the port if it doesn't exist. To use an OVS internal port, the name of
the port must be provided:
ovsInternal:
name: tap0device: This specifies a network device on all Nodes. A Pod's traffic will
be redirected or mirrored to the network device on the same Node that hosts the
Pod. The network device must exist on all Nodes and Antrea will attach it to the
OVS bridge if not already attached. To use a network device, the name of the
device must be provided:
device:
name: eno2geneve: This specifies a remote destination for a GENEVE tunnel. All
selected Pods' traffic will be redirected or mirrored to the destination via
a GENEVE tunnel. The remoteIP field must be provided to specify the IP address
of the destination. Optionally, the destinationPort field could be used to
specify the UDP destination port of the tunnel, or 6081 will be used by default.
If Virtual Network Identifier (VNI) is desired, the vni field can be specified
to an integer in the range 0-16,777,215:
geneve:
remoteIP: 10.0.10.2
destinationPort: 6081
vni: 1vxlan: This specifies a remote destination for a VXLAN tunnel. All
selected Pods' traffic will be redirected or mirrored to the destination via
a VXLAN tunnel. The remoteIP field must be provided to specify the IP address
of the destination. Optionally, the destinationPort field could be used to
specify the UDP destination port of the tunnel, or 4789 will be used by default.
If Virtual Network Identifier (VNI) is desired, the vni field can be specified
to an integer in the range 0-16,777,215:
vxlan:
remoteIP: 10.0.10.2
destinationPort: 4789
vni: 1gre: This specifies a remote destination for a GRE tunnel. All selected
Pods' traffic will be redirected or mirrored to the destination via a GRE
tunnel. The remoteIP field must be provided to specify the IP address of the
destination. If GRE key is desired, the key field can be specified to an
integer in the range 0-4,294,967,295:
gre:
remoteIP: 10.0.10.2
key: 1erspan: This specifies a remote destination for an ERSPAN tunnel. All
selected Pods' traffic will be mirrored to the destination via an ERSPAN tunnel.
The remoteIP field must be provided to specify the IP address of the
destination. If ERSPAN session ID is desired, the sessionID field can be
specified to an integer in the range 0-1,023. The version field must be
provided to specify the ERSPAN version: 1 for version 1 (type II), or 2 for
version 2 (type III).
For version 1, the index field can be specified to associate with the ERSPAN
traffic's source port and direction. An example of version 1 might look like
this:
erspan:
remoteIP: 10.0.10.2
sessionID: 1
version: 1
index: 1For version 2, the dir field can be specified to indicate the mirrored
traffic's direction: 0 for ingress traffic, 1 for egress traffic. The
hardwareID field can be specified as an unique identifier of an ERSPAN v2
engine. An example of version 2 might look like this:
erspan:
remoteIP: 10.0.10.2
sessionID: 1
version: 2
dir: 0
hardwareID: 4The returnPort field should only be set when the action is Redirect. It is
similar to the targetPort field, but meant for specifying the port from which
the traffic will be sent back to OVS and be forwarded to its original
destination.
In this example, we will mirror all Pods' traffic and send them to a remote destination via a GENEVE tunnel:
apiVersion: crd.antrea.io/v1alpha2
kind: TrafficControl
metadata:
name: mirror-all-to-remote
spec:
appliedTo:
podSelector: {}
direction: Both
action: Mirror
targetPort:
geneve:
remoteIP: 10.0.10.2In this example, we will redirect traffic of all Pods in the Namespace prod to
OVS internal ports named tap0 configured on Nodes that these Pods run on.
The returnPort configuration means, if the traffic is sent back to OVS from
OVS internal ports named tap1, it will be forwarded to its original
destination. Therefore, if an intrusion prevention system or a network firewall
is configured to capture and forward traffic between tap0 and tap1, it can
actively scan forwarded network traffic for malicious activities and known
attack patterns, and drop the traffic determined to be malicious.
apiVersion: crd.antrea.io/v1alpha2
kind: TrafficControl
metadata:
name: redirect-prod-to-local
spec:
appliedTo:
namespaceSelector:
matchLabels:
kubernetes.io/metadata.name: prod
direction: Both
action: Redirect
targetPort:
ovsInternal:
name: tap0
returnPort:
ovsInternal:
name: tap1With the TrafficControl capability, Antrea can be used with threat detection
engines to provide network-based IDS/IPS to Pods. We provide a reference
cookbook on how to implement IDS using Suricata. For more information, refer to
the cookbook.