togsh - extracts shell commands from github actions workflows

toghsh is a helper tool to translate github action workflows into equivalent shell scripts. toghsh cannot, and will not translate github actions into shell-script equivalent.

When I should use toghsh

In the narrow but relevant use case on which you run end-to-end (e2e) tests as github workflows, relying on shell commands to setup and run the suite, then toghsh can help you.

You can keep the master source of the e2e tests as github workflows, and use toghsh to generate a runnable shell script for your local/custom worker.

You are advised and strongly encouraged to look at act for a more comprehensive solution.

Usage

1. fetch the toghsh source tree
2. run make
3. copy the binary (_out/toghsh by default) in your $PATH
4. find the available jobs: togsh -L .github/workflows/$YOUR_YAML
5. run your job locally: togsh -J $JOB .github/workflows/$YOUR_YAML | bash -xe

Example

$ ./_out/toghsh --help
Usage of ./_out/toghsh:
  -J, --job-id string   process job
  -L, --list            list available jobs and exit
pflag: help requested

$ curl -L https://raw.githubusercontent.com/k8stopologyawareschedwg/resource-topology-exporter/master/.github/workflows/e2e.yml -o example.yaml
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  3518  100  3518    0     0  14242      0 --:--:-- --:--:-- --:--:-- 14242

$ ./_out/toghsh --list example.yaml 
job: "e2e-ic"
job: "e2e-ip"

$ ./_out/toghsh --job-id e2e-ic example.yaml 
### setup environment as per job "e2e-ic"
export RTE_CONTAINER_IMAGE=quay.io/k8stopologyawarewg/resource-topology-exporter:ci

### order=0 - ID="" - name="checkout sources"
### nothing to run

### order=1 - ID="go" - name="setup golang"
### nothing to run

### order=2 - ID="" - name="build test binary"
make build-e2e

### order=3 - ID="" - name="build image"
RTE_CONTAINER_IMAGE=${RTE_CONTAINER_IMAGE} RUNTIME=docker make image

### order=4 - ID="" - name="generate manifests"
RTE_CONTAINER_IMAGE=${RTE_CONTAINER_IMAGE} RTE_POLL_INTERVAL=10s make gen-manifests | tee rte-e2e.yaml

### order=5 - ID="" - name="create K8S kind cluster"
# kind is part of 20.04 image, see: https://github.com/actions/virtual-environments/blob/main/images/linux/Ubuntu2004-README.md
kind create cluster --config=hack/kind-config-e2e.yaml --image kindest/node:v1.21.1@sha256:69860bda5563ac81e3c0057d654b5253219618a22ec3a346306239bba8cfa1a6
kind load docker-image ${RTE_CONTAINER_IMAGE}

### order=6 - ID="" - name="deploy RTE"
# TODO: what about the other workers (if any)?
kubectl label node kind-worker node-role.kubernetes.io/worker=''
kubectl create -f rte-e2e.yaml

### order=7 - ID="" - name="cluster info"
kubectl get nodes
kubectl get pods -A
kubectl describe pod -l name=resource-topology || :
kubectl logs -l name=resource-topology -c resource-topology-exporter-container || :

### order=8 - ID="" - name="cluster ready"
hack/check-ds.sh
kubectl logs -l name=resource-topology -c resource-topology-exporter-container || :
kubectl get noderesourcetopologies.topology.node.k8s.io -A -o yaml

### order=9 - ID="" - name="run E2E tests"
export KUBECONFIG=${HOME}/.kube/config 
_out/rte-e2e.test -ginkgo.focus='\[(RTE|TopologyUpdater)\].*\[(Local|InfraConsuming)\]'