NextGenTrigger Hardware

• Can be accessed by a Kubernetes cluster
• Thus the code needs to run in a container
• Might require explicit approval for non-NGT people (to be checked)
• Needs kubernetes installed and configured correctly (on personal machine or on lxplus)

How to Run Our Code

Create a file "session.yml" defining a pod. The marked lines you might have to change.

apiVersion: v1
kind: Pod
metadata:
  name: session-1
  labels:
    mount-eos: "true"
    inject-oauth2-token-pipeline: "true"
  annotations:
    sidecar.istio.io/inject: "false"
spec:
  containers:
  - name: session-1
    image: registry.cern.ch/ngt-wp1.7/wp1.7-soa-wrapper:latest
    command: ["sleep", "infinity"]
    resources:
      limits:
        nvidia.com/gpu: 1
    securityContext:
      runAsUser: 0 
      runAsGroup: 0

In the terminal, run the following commands to start the pod and enter an interactive session.

kubectl apply -f session.yml
kubectl exec -it session-1 -- /bin/bash

Now you can "git clone" your code and build it.

TODOs

• Persistent volume claim (PVC) to save stuff between sessions
• A secret for the ssh key when we use git in the container
• Edit code remotely e.g. with visual studio code
• Merge everything to a YAML file that works out of the box (maybe using a helm chart?)
• Find a setting that allows comparable benchmarks