Skip to content

More improvments to launcher's doc #49

More improvments to launcher's doc

More improvments to launcher's doc #49

# Tests launcher-based server-providing pods in a `kind` cluster
name: Launcher-Based E2E Test
on:
push:
paths:
- ".github/workflows/launcher-based-e2e-test.yml"
- Makefile
- cmd/dual-pods-controller/**
- cmd/test-requester/**
- cmd/launcher-populator/**
- inference_server/launcher/**
- dockerfiles/Dockerfile.launcher.benchmark
- pkg/**
- test/e2e/run-launcher-based.sh
- test/e2e/mkobjs.sh
pull_request:
branches:
- main
jobs:
debug:
name: print relevant info
runs-on: ubuntu-latest
steps:
- run: |
echo "github.actor=${{ github.actor }}"
echo "github.action_ref=${{ github.action_ref }}"
echo "github.event_name=${{ github.event_name }}"
echo "github.head_ref=$GITHUB_HEAD_REF"
echo "github.ref=${{ github.ref }}"
echo "github.ref_name=${{ github.ref_name }}"
echo "github.repository=${{ github.repository }}"
echo "github.repository_owner=${{ github.repository_owner }}"
echo "github.triggering_actor=${{ github.triggering_actor }}"
echo "GITHUB_ACTION_REF=$GITHUB_ACTION_REF"
run-launcher-test:
runs-on: ubuntu-22.04-arm
steps:
- uses: actions/setup-go@7a3fe6cf4cb3a834922a1244abfce67bcef6a0c5 # v6.2.0
with:
go-version: '1.24.2'
- name: Install ko
uses: ko-build/setup-ko@d006021bd0c28d1ce33a07e7943d48b079944c8d # v0.9
with:
version: v0.15.2
- name: Checkout code
uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
- name: Set up Docker Buildx
uses: docker/setup-buildx-action@8d2750c68a42422c14e847fe6c8ac0403b4cbd6f # v3.12.0
- name: Run launcher-based E2E test
run: test/e2e/run-launcher-based.sh
- name: show all pods
if: always()
run: kubectl get pods -A -o wide
- name: show test pods with labels
if: always()
run: kubectl get pods -L dual-pods.llm-d.ai/dual,dual-pods.llm-d.ai/sleeping,dual-pods.llm-d.ai/launcher-config-name
- name: show ReplicaSets
if: always()
run: kubectl get rs -A
- name: show dual-pods controller log
if: always()
run: kubectl logs deploy/fma-dual-pods-controller
- name: show launcher-populator log
if: always()
run: kubectl logs deploy/fma-launcher-populator || echo "launcher-populator not deployed"
- name: show GPU allocations
if: always()
run: kubectl get cm gpu-allocs -o yaml
- name: show GPU map
if: always()
run: kubectl get cm gpu-map -o yaml
- name: show InferenceServerConfigs
if: always()
run: kubectl get inferenceserverconfigs -o yaml
- name: show LauncherConfigs
if: always()
run: kubectl get launcherconfigs -o yaml
- name: show YAML of test pods
if: always()
run: kubectl get pods -o yaml
- name: show launcher pod logs
if: always()
run: |
for pod in $(kubectl get pods -l dual-pods.llm-d.ai/launcher-config-name -o name); do
echo "=== Logs for $pod ==="
kubectl logs $pod || echo "Failed to get logs for $pod"
done