containerlab

Containerlab Skill

Reference for deploying, inspecting, and tearing down containerlab-based SR Linux labs.

Environment

• Rootless: all commands run without sudo
• Runtime: Docker (rootless)
• Python tooling: use uv for any Python scripting
• Local execution: labs run on the local machine

Commands

Deploy

containerlab deploy -t <path>.clab.yml
containerlab deploy -t <path>.clab.yml --reconfigure   # destroy lab first, then redeploy clean
containerlab deploy -t <path>.clab.yml --node-filter <node1,node2>

Destroy

containerlab destroy -t <path>.clab.yml --cleanup
containerlab destroy --name <lab-name> --cleanup

Always use --cleanup to remove generated certs and artifacts.

Inspect

containerlab inspect --all                    # all running labs
containerlab inspect -t <path>.clab.yml       # specific topology
containerlab inspect --name <lab-name>        # by name

Output includes: node name, container ID, image, kind, state, management IPs, mapped ports.

Use --format json | jq to minimize output when scripting or reporting.

Expected JSON structure from containerlab inspect --format json:

{
  "<lab-name>": [
    {
      "lab_name": "<lab>",
      "labPath": "<relative-path>",
      "name": "clab-<lab>-<node>",
      "container_id": "<id>",
      "image": "<image>",
      "kind": "<kind>",
      "group": "<group>",
      "state": "running",
      "status": "Up 2 hours",
      "ipv4_address": "<cidr>",
      "ipv6_address": "<cidr>",
      "ports": {},
      "owner": "<user>"
    }
  ]
}

Top-level keys are lab names; values are arrays of containers.

Execute on Nodes

Container naming convention: clab--

# SR Linux CLI
docker exec -it clab-<lab>-<node> sr_cli "<command>"

# Shell access
docker exec -it clab-<lab>-<node> bash

# SSH (default creds: admin / NokiaSrl1!)
ssh admin@clab-<lab>-<node>

Using containerlab exec

For multi-node command execution:

# All nodes in a lab
containerlab exec -t <path>.clab.yml --cmd '<command>'

# Specific node by label
containerlab exec -t <path>.clab.yml --label clab-node-name=<node> --cmd '<command>'

# SR Linux CLI via containerlab exec
containerlab exec -t <path>.clab.yml --cmd 'sr_cli "show version"'

Default Topology

When a quick test lab is needed and the user has no custom topology, create this:

name: srlinux-lab
topology:
  nodes:
    srl1:
      kind: nokia_srlinux
      image: ghcr.io/nokia/srlinux:latest
    srl2:
      kind: nokia_srlinux
      image: ghcr.io/nokia/srlinux:latest
  links:
    - endpoints: ["srl1:e1-1", "srl2:e1-1"]

Custom Topologies

When the user provides a .clab.yml:

1. Validate YAML syntax and presence of topology.nodes and topology.links
2. Confirm the SR Linux image tag is pullable
3. If startup-config references files, confirm they exist

Conventions: use .clab.yml extension, place in project root or lab/ directory.

Operational Procedures

Deploy Procedure

1. Preflight

   containerlab version
   docker info --format '{{.ServerVersion}}'
   

2. Check for conflicts

   containerlab inspect --all --format json 2>/dev/null | jq -r 'keys[]'
   

   If same-name lab exists, ask user: redeploy or reuse?
3. Deploy

   containerlab deploy -t <path>.clab.yml
   

4. Verify (SR Linux takes 30-60s to boot)

   containerlab inspect -t <path>.clab.yml --format json | jq '[.[][] | {name, state, ipv4: .ipv4_address}]'
   docker exec clab-<lab>-<node> sr_cli "show version" 2>/dev/null
   

5. Report: return node names, container names, management IPs, and access instructions.

Destroy Procedure

1. Identify

   containerlab inspect --all --format json 2>/dev/null | jq -r 'keys[]'
   

2. Destroy

   containerlab destroy -t <path>.clab.yml --cleanup
   

3. Verify

   containerlab inspect --all --format json 2>/dev/null | jq length
   docker ps --filter "label=containerlab" --format "{{.Names}}" 2>/dev/null
   

SR Linux Reference

Default Credentials

• User: admin
• Password: NokiaSrl1!

Common CLI Commands

sr_cli "show version"
sr_cli "show system information"
sr_cli "show interface brief"
sr_cli "show network-instance default route-table all"

Interface Naming

• Containerlab: e1-1, e1-2, ...
• SR Linux CLI: ethernet-1/1, ethernet-1/2, ...
• Management: mgmt0 (accessible via Docker network)

Health Check

Use the state column from inspect output:

containerlab inspect -t <path>.clab.yml --format json | jq '[.[][] | {name, state}]'

Healthy nodes show "state": "running".

Context Efficiency

These practices minimize context window consumption when executing commands:

• Always use --format json | jq for inspect commands to extract only what is needed
• Redirect verbose output: containerlab deploy -t ... > /tmp/clab-deploy.log 2>&1
• Use targeted checks over broad scans: if you know the lab name, do not use --all
• Use sr_cli one-liners via docker exec rather than interactive SSH sessions
• Pipe long output to files and read only the relevant parts with head, tail, grep

Troubleshooting

Problem	Quick Fix
Image pull failure	docker pull ghcr.io/nokia/srlinux:latest, check auth/network
Node stuck starting	Check docker logs clab--, likely RAM (need ~2GB/node)
Cannot SSH	Wait 30-60s post-deploy, verify IP via containerlab inspect
Orphan containers after destroy	docker ps -a --filter label=containerlab then docker rm -f
Permission denied (rootless)	Verify docker info works, check ~/.config/containers/
Lab name already exists	containerlab inspect --all --format json | jq -r 'keys[]' to check; ask user to redeploy (--reconfigure) or destroy first
Topology YAML parse error	Validate YAML syntax; check for missing topology.nodes or topology.links keys
Docker daemon not running	Verify with docker info; for rootless check systemctl --user status docker
Node filter matches nothing	Verify node names in topology file match --node-filter values exactly