Add install.sh end-to-end smoke gate against published release

Across v6 rc.1 → rc.5 the published install.sh asset was the agent
installer rather than the server installer, and the README's pinned
ed25519 key did not verify what the pipeline actually signed. The first
broke `bash install.sh --version` and the in-product Update button; the
second silently failed the README's secure-install ssh-keygen step.
Neither was caught by CI because every existing gate operated on the
local release/ build, the Docker image, or the helm chart — nothing
exercised the documented LXC/systemd install commands against the
published release URL.

scripts/validate-release.sh now catches asset-identity drift at build
time. This workflow catches the rest of the regression class — anything
that breaks the actual install at runtime — by running the documented
flow end-to-end against the published release.

What it does:
- Downloads install.sh, install.sh.sshsig, and the linux-amd64 tarball
  from releases/download/<tag>/.
- Extracts the README's pinned pulse-installer ed25519 key and runs the
  exact ssh-keygen -Y verify command from the README's secure-install
  snippet against the downloaded asset.
- Re-checks the server-installer banner, the --version) arg handler, and
  the absence of the agent banner — same pins as validate-release.sh, but
  now against what GitHub is actually serving (not just what was built
  locally).
- Boots jrei/systemd-debian:12 privileged, runs
  `bash install.sh --archive <tarball> --disable-auto-updates` from
  inside, waits for systemd pulse.service to become active, hits
  /api/health, and asserts /api/version reports the expected version.

--archive mode is used rather than --version so the workflow doesn't
depend on install.sh's self-refetch loop (the re-fetched bytes are the
ones we already validated). Auto-updates are disabled to avoid the timer
unit doing anything during the smoke run.

Triggers are workflow_dispatch + workflow_call only. Wire it into
create-release.yml after the next RC validates it green.

Pinned in build_release_assets_test.go so silent deletion or weakening
of any critical assertion (signature verify, banner check, /api/health
hit, version match) trips the test.
This commit is contained in:
rcourtman 2026-05-12 11:25:46 +01:00
parent b69c8c8007
commit 065ebdb276
2 changed files with 305 additions and 0 deletions

256
.github/workflows/install-sh-smoke.yml vendored Normal file
View file

@ -0,0 +1,256 @@
name: install.sh Smoke (Published Release)
# End-to-end smoke that exercises the documented Proxmox-LXC / systemd
# install flow against the published GitHub Release. Across v6 rc.1 → rc.5
# the published install.sh asset was actually the agent installer (rejecting
# --version) and the README's pinned signature key did not match the
# pipeline's actual signing key, so anyone running the secure-install
# snippet from the README hit one of two silent failure modes for ~30 days.
# The validate-release.sh content checks catch the asset-identity drift at
# build time; this workflow proves the documented commands actually install
# and boot Pulse end-to-end in a clean Linux environment.
#
# Triggers:
# - workflow_dispatch: run manually against any tag for verification.
# - workflow_call: callable from create-release.yml so future releases get
# the gate automatically once it's stable.
#
# What it does:
# 1. Downloads install.sh + install.sh.sshsig + the linux-amd64 tarball
# from the published release URL (not the local release/ dir).
# 2. Extracts the README's pinned ed25519 key and runs the README's exact
# ssh-keygen -Y verify command. Catches README/key drift.
# 3. Boots a privileged systemd Debian 12 container, runs
# `bash install.sh --archive <tarball> --enable-auto-updates` inside it
# (avoids the install.sh self-refetch loop that --version triggers; the
# re-fetched bytes are the same as what we already validated).
# 4. Waits for systemd pulse.service to be active and hits /api/health
# from inside the container.
# 5. Confirms /api/health reports the expected version.
on:
workflow_call:
inputs:
tag:
description: 'Release tag (e.g., v6.0.0-rc.6)'
required: true
type: string
version:
description: 'Version without v prefix (e.g., 6.0.0-rc.6)'
required: true
type: string
repository:
description: 'owner/repo to pull the published release from. Defaults to the workflow repository.'
required: false
type: string
default: ''
workflow_dispatch:
inputs:
tag:
description: 'Release tag (e.g., v6.0.0-rc.6)'
required: true
type: string
version:
description: 'Version without v prefix (e.g., 6.0.0-rc.6)'
required: true
type: string
repository:
description: 'owner/repo to pull from (leave blank for current repo)'
required: false
type: string
default: ''
permissions:
contents: read
concurrency:
group: install-sh-smoke-${{ inputs.tag }}
cancel-in-progress: false
jobs:
smoke:
runs-on: ubuntu-24.04
timeout-minutes: 15
steps:
- name: Checkout repository (for README key extraction)
uses: actions/checkout@34e114876b0b11c390a56381ad16ebd13914f8d5 # v4
- name: Resolve release repository
id: repo
env:
INPUT_REPO: ${{ inputs.repository }}
DEFAULT_REPO: ${{ github.repository }}
run: |
repo="${INPUT_REPO:-$DEFAULT_REPO}"
echo "repo=$repo" >> "$GITHUB_OUTPUT"
echo "Using release repository: $repo"
- name: Download published install.sh + sshsig + linux-amd64 tarball
env:
TAG: ${{ inputs.tag }}
REPO: ${{ steps.repo.outputs.repo }}
run: |
set -euo pipefail
mkdir -p smoke-workspace
cd smoke-workspace
base="https://github.com/${REPO}/releases/download/${TAG}"
echo "Pulling from ${base}/"
curl -fsSL --retry 5 --retry-delay 5 --retry-all-errors \
-o install.sh "${base}/install.sh"
curl -fsSL --retry 5 --retry-delay 5 --retry-all-errors \
-o install.sh.sshsig "${base}/install.sh.sshsig"
tarball="pulse-${TAG}-linux-amd64.tar.gz"
curl -fsSL --retry 5 --retry-delay 5 --retry-all-errors \
-o "${tarball}" "${base}/${tarball}"
echo "Downloaded:"
ls -la
- name: Verify install.sh signature with README's pinned key
env:
TAG: ${{ inputs.tag }}
run: |
set -euo pipefail
readme_key=$(grep -oE 'ssh-ed25519 [A-Za-z0-9+/=]+ pulse-installer' README.md | head -1)
if [ -z "$readme_key" ]; then
echo "::error::Could not extract pulse-installer key from README.md"
exit 1
fi
echo "README pins: $readme_key"
allowed_signers=$(mktemp)
printf 'pulse-installer %s\n' "$readme_key" > "$allowed_signers"
cd smoke-workspace
if ! ssh-keygen -Y verify \
-f "$allowed_signers" \
-I pulse-installer \
-n pulse-install \
-s install.sh.sshsig < install.sh; then
echo "::error::Published install.sh.sshsig does not verify against the README's pinned key."
echo "::error::Either README.md is pinning the wrong key or the pipeline signed with a different key."
rm -f "$allowed_signers"
exit 1
fi
rm -f "$allowed_signers"
echo "✓ install.sh signature verifies against README's pinned key"
- name: Assert install.sh is the Pulse server installer
run: |
set -euo pipefail
cd smoke-workspace
if ! grep -qE '^# Pulse Installer Script' install.sh; then
echo "::error::install.sh banner is not the Pulse server installer"
exit 1
fi
if grep -q 'Pulse Unified Agent Installer' install.sh; then
echo "::error::install.sh is the agent installer, not the server installer"
exit 1
fi
if ! grep -qE '^[[:space:]]*--version\)' install.sh; then
echo "::error::install.sh does not handle --version"
exit 1
fi
echo "✓ install.sh is the server installer with --version support"
- name: Run install.sh end-to-end in a privileged systemd container
env:
TAG: ${{ inputs.tag }}
VERSION: ${{ inputs.version }}
run: |
set -euo pipefail
tarball="pulse-${TAG}-linux-amd64.tar.gz"
container_name="pulse-install-smoke-$$"
# Cleanup on exit no matter what.
trap 'docker rm -f "${container_name}" >/dev/null 2>&1 || true' EXIT
# jrei/systemd-debian:12 is a community systemd-in-Docker image used
# for Ansible / Molecule testing — small, no Pulse-specific assumptions.
docker run -d --rm \
--name "${container_name}" \
--privileged \
--tmpfs /tmp --tmpfs /run \
-v /sys/fs/cgroup:/sys/fs/cgroup:rw \
-v "$(pwd)/smoke-workspace:/smoke" \
-p 7655:7655 \
jrei/systemd-debian:12
echo "Waiting for systemd to be ready inside the container..."
for i in $(seq 1 30); do
if docker exec "${container_name}" systemctl is-system-running --wait 2>/dev/null | grep -qE '^(running|degraded)$'; then
break
fi
if [ "$i" -eq 30 ]; then
docker logs "${container_name}" || true
echo "::error::systemd did not become ready inside the container"
exit 1
fi
sleep 2
done
echo "✓ systemd is up"
echo "Installing prerequisites inside container..."
docker exec "${container_name}" bash -lc 'apt-get update -qq && apt-get install -y -qq curl ca-certificates jq sudo'
echo "Running install.sh --archive against the published tarball..."
# docker exec without -t leaves stdin without a TTY, which install.sh's
# safe_read helper detects and falls through to defaults on every prompt.
docker exec "${container_name}" \
bash -lc "cd /smoke && bash install.sh --archive /smoke/${tarball} --disable-auto-updates"
echo "Waiting for pulse.service to become active..."
for i in $(seq 1 60); do
state=$(docker exec "${container_name}" systemctl is-active pulse 2>/dev/null || true)
if [ "$state" = "active" ]; then
break
fi
if [ "$i" -eq 60 ]; then
docker exec "${container_name}" systemctl status pulse --no-pager || true
docker exec "${container_name}" journalctl -u pulse --no-pager --lines=80 || true
echo "::error::pulse.service did not become active within 2 minutes"
exit 1
fi
sleep 2
done
echo "✓ pulse.service is active"
echo "Hitting /api/health..."
for i in $(seq 1 30); do
if docker exec "${container_name}" curl -fsS http://127.0.0.1:7655/api/health >/dev/null 2>&1; then
break
fi
if [ "$i" -eq 30 ]; then
docker exec "${container_name}" systemctl status pulse --no-pager || true
docker exec "${container_name}" journalctl -u pulse --no-pager --lines=80 || true
echo "::error::/api/health did not respond within 60 seconds of service activation"
exit 1
fi
sleep 2
done
echo "✓ /api/health responded 200"
echo "Confirming installed version matches ${VERSION} via /api/version..."
# /api/health intentionally does not include version; /api/version is
# the authoritative endpoint and is one of the canonical post-upgrade
# checks documented in docs/UPGRADE_v6.md.
version_payload=$(docker exec "${container_name}" curl -fsS http://127.0.0.1:7655/api/version)
echo "Version payload: ${version_payload}"
installed_version=$(echo "${version_payload}" | jq -r '.version // empty')
if [ -z "${installed_version}" ]; then
echo "::error::/api/version did not include a version field"
exit 1
fi
# Normalize: VERSION input is "6.0.0-rc.6", installed_version may be "v6.0.0-rc.6".
installed_version="${installed_version#v}"
if [ "${installed_version}" != "${VERSION}" ]; then
echo "::error::Installed version mismatch. Expected ${VERSION}, got ${installed_version}"
exit 1
fi
echo "✓ Installed version matches ${VERSION}"
- name: Smoke result
run: |
echo "::notice::install.sh smoke passed for tag ${{ inputs.tag }}"

View file

@ -880,6 +880,55 @@ func TestBuildReleasePackagesPulseMcpForAllPlatforms(t *testing.T) {
}
}
func TestInstallShSmokeWorkflowPresent(t *testing.T) {
// End-to-end install.sh smoke gate. validate-release.sh catches asset-
// identity drift at build time (right banner / right --version handler /
// right signature key). This workflow proves the documented secure-install
// commands actually install and boot Pulse end-to-end against the
// published GitHub Release, which is the surface that broke silently
// across v6 rc.1 → rc.5. Removing or weakening any of these assertions
// reopens that regression class.
workflowBytes, err := os.ReadFile(repoFile(".github", "workflows", "install-sh-smoke.yml"))
if err != nil {
t.Fatalf("read install-sh-smoke.yml: %v", err)
}
workflow := string(workflowBytes)
required := []string{
// Inputs and triggers.
`name: install.sh Smoke (Published Release)`,
`workflow_call:`,
`workflow_dispatch:`,
// Pull straight from the published release URL (not local release/).
`releases/download/${TAG}`,
`install.sh.sshsig`,
`pulse-${TAG}-linux-amd64.tar.gz`,
// README key extraction + actual ssh-keygen verify against the
// downloaded asset.
`grep -oE 'ssh-ed25519 [A-Za-z0-9+/=]+ pulse-installer' README.md`,
`ssh-keygen -Y verify \`,
`-I pulse-installer \`,
`-n pulse-install \`,
`-s install.sh.sshsig < install.sh`,
// Server-installer identity assertions, mirroring validate-release.sh.
`grep -qE '^# Pulse Installer Script' install.sh`,
`grep -q 'Pulse Unified Agent Installer' install.sh`,
`grep -qE '^[[:space:]]*--version\)' install.sh`,
// End-to-end install in a privileged systemd container.
`jrei/systemd-debian:12`,
`bash install.sh --archive /smoke/${tarball} --disable-auto-updates`,
`systemctl is-active pulse`,
`curl -fsS http://127.0.0.1:7655/api/health`,
// Authoritative version check via /api/version (not /api/health).
`curl -fsS http://127.0.0.1:7655/api/version`,
`Installed version mismatch. Expected`,
}
for _, needle := range required {
if !strings.Contains(workflow, needle) {
t.Fatalf("install-sh-smoke.yml missing required smoke step: %s", needle)
}
}
}
func repoFile(parts ...string) string {
root := filepath.Join("..", "..")
segments := append([]string{root}, parts...)