Overview#

Node type srl represents node running Nokia SRLinux.

Configuration#

Example#

    nodes:
      srl-1:
        srl:
          chassis: ixr-h5-32d
          image: ghcr.io/nokia/srlinux:25.10
          memory: 4Gi
          cpu: 2.0

Image#

SRL node uses offical Nokia SRLinux image.

Chassis#

KNL supported following chassis models:

ixr-d1
ixr-d2l
ixr-d3l
ixr-d5
ixr-h2
ixr-h3
ixr-6e
ixr-10e
ixr-18e
ixr-x1b
ixr-x3b
ixr-h5-32d
sxr-1d-32d
ixs-a1

default model is ixr-d3l

Note
currently KNL only creates single card in a chassis, even in case of multi-slot chassis.

SRL ethernet interface mapping#

For a given SRL node, by default, the 1st ethernet interface (ethernet-1/1) is mapped to the 1st lab link it belongs to, 2nd interface (ethernet-1/2) is mapped to the 2nd lab link it belongs to ..so on. The first lab link is the first in lexicographical order of the link name.

In example below, link2 is before link9 in lexicographical order, so ethernet-1/1 of srl-1 connects to pod-2, ethernet-1/2 connects to pod-2

apiVersion: knl.kubenetlab.net/v1beta1
kind: Lab
metadata:
    name: srl
    namespace: knl-system
spec:
  links:
    link9:
      nodes:
      - node: srl-1
      - node: pod-1
    link2:
      nodes:
      - node: srl-1
      - node: pod-2

However if port is specified, then its value is used to mapping to the SRL interfaces, the format is e1-X, like e1-1 means ethernet-1/1.

apiVersion: knl.kubenetlab.net/v1beta1
kind: Lab
metadata:
    name: srl
    namespace: knl-system
spec:
  links:
    link9:
      nodes:
      - node: srl-1
        port: e1-3 # map to ethernet-1/3
      - node: pod-1
    link2:
      nodes:
      - node: srl-1
        port: e1-4 # map to ethernet-1/4
      - node: pod-2

License#

License is required for multi-slot chassis models, specified via license field in the node configuration, the value is name of a k8s secrete in knl-system hold the license with key license, the default secret name is srllic.

Persistent storage#

On each SRLinux node, the files include startup config.json under /etc/opt/srlinux/ are persistent on an auto created k8s PVC (size: 100MB) after lab is removed.

Memory & CPU#

Node configuraion also contains cpu and memory requests for the pod, which would be used as input for k8s to schedule the pod; minimal 4GB memory is required (which is also the default value), certain model requires more memory like 8GB. CPU request is optional.

Note
If the pod keep crashing/restarting, one possible reason is not having enough memory, increasing request memory in the node configuration could fix it.