Looking for an example of configuring clab for connecting to SNMP from host VM

[danstadler-pdx]

Hi - I'm new to most of this but my goal is to use Containerlab to set up a few virtual routers, and then from the host VM (Ubuntu 20 in my case) be able to use tools like snmpwalk, to view the SNMP telemetry on the routers.

(Ultimately my goal is to set up snmp_exporter on the host VM so I can get the SNMP telemetry as Prometheus metrics, and feed them into Grafana. Full disclosure: I am an SE at Grafana Labs, and I'm taking myself on a learning curve with SNMP here.)

I have attached my full topology. A few key snippets:

  nodes:
    R1:
      kind: srl
      mgmt-ipv4: 172.100.100.11
      publish:
        - udp/161
        - udp/162
      env:
        # boot delay in seconds
        BOOT_DELAY: 30

    border0:
      kind: border0

I have successfully authenticated to border0 and stored a token locally.

When I try to deploy the lab, here is what I get about the publish section:

INFO[0000] Creating lab directory: /home/ubuntu/grafana-snmp/clab-ext
INFO[0000] Creating docker network: Name="statics", IPv4Subnet="172.100.100.0/24", IPv6Subnet="", MTU="1500"
INFO[0000] Creating container: "SSH"
INFO[0000] Creating container: "R2"
INFO[0000] Creating container: "R1"
INFO[0001] Creating virtual wire: R1:e1-1 <--> R2:e1-1
INFO[0001] Creating container: "border0"
INFO[0001] Running postdeploy actions for Nokia SR Linux 'R1' node
INFO[0001] Running postdeploy actions for Nokia SR Linux 'R2' node
INFO[0001] Creating border0.com tunnels...
ERRO[0001] failed to run postdeploy task for node border0: border0.com does not support socket type "udp". Supported types are "ssh|tls|http|https"

Is there an easy example someone can provide about how to accomplish this goal?
thanks

danstadler-pdx-topology-yml.txt

[hellt]

Hi @danstadler-pdx
You don't need to use publish to just be able to connect to the nodes running on the same host. Just remove publish and BOOT_DELAY and border0 node from your lab and you should be able to go

[hellt]

Likely AWS uses it for metrics collection, over a standard port.

[danstadler-pdx]

ah. makes sense.

[danstadler-pdx]

I got back to this a bit more today. I am slightly stumped and wondering if you could offer me a cheat sheet set of quick steps to follow.

Here's what I can do no problem:

• create a topology with 2 Ubuntu 20.04 VMs, and on one of them, configure snmpd, then snmpwalk it from the other Ubuntu VM

I can also add an SRLinux VM to the topology, and update its snmpd.conf so that I can snmpwalk it from itself (localhost). What I can't quite figure out:

• how to snmpwalk the SRL VM, from the Ubuntu VM

When I try that, I get:
Timeout: No Response from 172.80.80.13

(I can ping to 172.80.80.13 - but I just can't get snmpwalk to connect.

Am I just missing something simple?

thanks thanks thanks
dan

[danstadler-pdx]

to close this out: thank you for the great help over on Discord. https://discord.com/channels/854262008589058058/854262008589058061/1108397848514535475

Long story short, I now understand a lot more about the following:

enter candidate
set / system snmp community public
set / system snmp network-instance mgmt
set / system snmp network-instance mgmt admin-state enable
commit now

as well as the documentation here:
https://documentation.nokia.com/srlinux/23-3/books/system-mgmt/configuring-snmp-system-mgmt.html

[hellt]

great to hear it worked out @danstadler-pdx
We are open to collaboration with grafana on networking stuff, so it would be nice to see you using containerlab for internal tests/dev