Files
GMW/.github/workflows/deploy.yml
T
asepharyana 4cb4904517 ci(deploy): fix attic fast path - public cache, extra-substituters, self-hosted client bootstrap
Root causes found by reproducing the 2026-08-10 run:
- attic 'gmw' cache was created private -> every narinfo/nix-cache-info
  read returned 401, so the VPS could never actually substitute from
  attic ('Substituted from Attic cache' was a false positive when the
  store path happened to be already present locally).
- VPS nix.conf used extra-trusted-substituters, which Determinate Nix
  never merges for nix-store CLI clients; extra-substituters (all
  users, no trust gate) fixes substitution (verified end-to-end:
  delete path -> nix-store --realise pulls from attic over HTTPS).
- runner bootstrap of the attic client depended on nix copy --from
  ssh:// (fragile, failed on runner); now the prebuilt attic client
  closure lives in the attic cache itself and the runner pulls it over
  HTTPS via extra-substituters configured in the Install Nix step.

Also surfaces bootstrap stderr on fallback for future debugging.
2026-08-10 18:29:00 +07:00

240 lines
10 KiB
YAML

name: Build & Deploy (Nix)
on:
push:
branches: [main]
workflow_dispatch:
concurrency:
group: gmw-deploy
cancel-in-progress: false
permissions:
contents: read
id-token: write
env:
VPS_HOST: ${{ secrets.VPS_HOST }}
VPS_USER: ${{ secrets.VPS_USER }}
jobs:
test:
runs-on: ubuntu-latest
steps:
- name: Checkout
uses: actions/checkout@v7
with:
fetch-depth: 0
submodules: false
- name: Setup Node
uses: actions/setup-node@v4
with:
node-version: 22
- name: Install pnpm
run: corepack enable && corepack prepare pnpm@11 --activate
- name: Install deps (backend)
working-directory: services/backend
run: pnpm install --ignore-scripts --no-frozen-lockfile
- name: Typecheck + test (backend)
working-directory: services/backend
run: |
./node_modules/.bin/tsc --noEmit
# e2e.test.ts requires a live backend (API_BASE) — run unit tests only
./node_modules/.bin/vitest run --exclude "src/e2e.test.ts"
- name: Install deps (discord-gateway)
working-directory: services/discord-gateway
run: pnpm install --ignore-scripts --no-frozen-lockfile
- name: Typecheck + test (discord-gateway)
working-directory: services/discord-gateway
run: |
./node_modules/.bin/tsc --noEmit
./node_modules/.bin/vitest run
- name: Biome check (all services)
run: |
cd services/backend && ./node_modules/.bin/biome check src/ tests/
cd ../discord-gateway && ./node_modules/.bin/biome check src/
build-and-deploy:
needs: test
runs-on: ubuntu-latest
strategy:
fail-fast: false
matrix:
service: [backend, discord-gateway, proxy, frontend]
steps:
- name: Checkout
uses: actions/checkout@v7
with:
fetch-depth: 0
submodules: false
- name: Install Nix
uses: DeterminateSystems/nix-installer-action@v22
with:
determinate: false
extra-conf: |
sandbox = false
accept-flake-config = true
# Attic binary cache as substituter on the runner: lets CI pull the
# prebuilt attic client (and any cached deps/builds) over HTTPS,
# no SSH round-trip needed. extra-substituters (NOT
# extra-trusted-substituters) is required — Determinate Nix never
# merges trusted-* substituters for nix-store CLI clients.
extra-substituters = https://attic.asepharyana.my.id/gmw
extra-trusted-public-keys = gmw:Fq2Anzuhkb+T/hftWnPcveHSi21/RzIgIOeG8pCJa88=
# NOTE: nix-installer-action unconditionally injects
# 'build-provenance-tags' into /etc/nix/nix.conf (a Determinate
# Nix-only setting). With determinate:false the runner's upstream
# nix warns 'unknown setting build-provenance-tags' on every
# invocation — benign, cosmetic. Switching determinate:true would
# silence it but changes the runner's nix flavor.
- name: Cache Nix
uses: DeterminateSystems/magic-nix-cache-action@v14
with:
use-flakehub: false
- name: Build ${{ matrix.service }}
id: build
run: |
nix build .#${{ matrix.service }} --impure --option sandbox false --print-build-logs
STORE_PATH=$(readlink result)
echo "store-path=$STORE_PATH" >> "$GITHUB_OUTPUT"
echo "Build OK ${{ matrix.service }}: $STORE_PATH"
- name: Setup SSH key
env:
SSH_KEY: ${{ secrets.SSH_PRIVATE_KEY }}
run: |
mkdir -p ~/.ssh
echo "$SSH_KEY" > ~/.ssh/id_ed25519
chmod 600 ~/.ssh/id_ed25519
sed -i 's/\r$//' ~/.ssh/id_ed25519
ssh-keygen -y -f ~/.ssh/id_ed25519 >/dev/null 2>&1 || { echo "SSH key invalid"; exit 1; }
ssh-keyscan -H "$VPS_HOST" >> ~/.ssh/known_hosts 2>/dev/null
# Push build result to Attic binary cache (attic.asepharyana.my.id) so
# the VPS can substitute it instead of a single-stream `nix copy ssh://`.
#
# Fast path: push DIRECTLY from the runner to the public attic endpoint
# (validated 2026-08-10: token auth over public HTTPS works without
# Tailscale). This skips the ~794MB closure SSH copy to the VPS that
# used to take 25+ minutes per new store path.
#
# The attic client is NOT in nixpkgs anymore and has no prebuilt
# releases, so we pull the same prebuilt closure the VPS uses
# (/nix/store/fygyy3yk4rqdknxkiwkqambpnhyax0k4-attic-0.1.0, ~52MB).
# The closure itself lives in the attic cache (pushed once from the
# VPS), so the runner bootstraps it over HTTPS via the configured
# extra-substituters — no SSH round-trip. If that fails we fall back
# to `nix copy --from ssh://`, then the old VPS-hop flow (SSH copy to
# VPS, then attic push from the VPS over Tailscale) so the deploy step
# always has a working closure path.
- name: Push to Attic cache
env:
ATTIC_TOKEN: ${{ secrets.ATTIC_TOKEN }}
run: |
if [ -z "$ATTIC_TOKEN" ]; then
echo "ATTIC_TOKEN not set; skipping attic push"
exit 0
fi
STORE_PATH="${{ steps.build.outputs.store-path }}"
ATTIC_DIR="/nix/store/fygyy3yk4rqdknxkiwkqambpnhyax0k4-attic-0.1.0"
ATTIC_BIN="$ATTIC_DIR/bin/attic"
attic_push_vps_hop() {
echo "Fallback: VPS-hop attic push"
# Copy closure to VPS (fast if attic already has it via substitute)
ssh "$VPS_USER@$VPS_HOST" "sudo /nix/var/nix/profiles/default/bin/nix-store --realise '$STORE_PATH'" 2>/dev/null \
|| nix copy --to "ssh://$VPS_USER@$VPS_HOST" "$STORE_PATH"
# Push from VPS → Attic over Tailscale (reliable for large payloads)
ssh "$VPS_USER@$VPS_HOST" "$ATTIC_BIN push imrnes-ts:gmw '$STORE_PATH' --jobs 4" \
|| echo "attic push failed (non-fatal; ssh copy fallback below)"
}
# ── Get an attic client on the runner ────────────────────────────
# Order: PATH → pull the prebuilt closure from the attic cache
# itself (extra-substituters configured in Install Nix step, HTTPS
# only, no SSH) → pull over ssh from the VPS → VPS-hop.
# The attic client closure is stored in the attic cache (pushed
# once from the VPS), so the fast path never depends on SSH.
ATTIC_BIN=""
if command -v attic >/dev/null 2>&1; then
ATTIC_BIN="$(command -v attic)"
elif nix-store --realise "$ATTIC_DIR" 2>/tmp/attic-bootstrap.err; then
echo "✅ Pulled attic client from attic cache (HTTPS substituter)"
ATTIC_BIN="$ATTIC_DIR/bin/attic"
elif nix copy --from "ssh://$VPS_USER@$VPS_HOST" "$ATTIC_DIR" 2>>/tmp/attic-bootstrap.err; then
echo "✅ Pulled attic client from VPS over ssh"
ATTIC_BIN="$ATTIC_DIR/bin/attic"
else
echo "attic client unavailable on runner; using VPS-hop flow"
echo "--- bootstrap errors (stderr) ---"
tail -5 /tmp/attic-bootstrap.err 2>/dev/null || true
attic_push_vps_hop
exit 0
fi
# ── Direct push: runner → attic public endpoint ──────────────────
mkdir -p "$HOME/.config/attic"
cat > "$HOME/.config/attic/config.toml" <<EOF
default-server = "pub"
[servers.pub]
endpoint = "https://attic.asepharyana.my.id"
token = "$ATTIC_TOKEN"
EOF
if "$ATTIC_BIN" push pub:gmw "$STORE_PATH" --jobs 4; then
echo "✅ Pushed $STORE_PATH to attic directly from runner"
else
echo "⚠️ Direct attic push failed; using VPS-hop flow"
attic_push_vps_hop
fi
# NOTE: env files /etc/gmw/backend.env & /etc/gmw/discord-gateway.env are
# managed MANUALLY on the VPS (source of truth). CI only builds & deploys.
- name: Deploy ${{ matrix.service }} to VPS
run: |
STORE_PATH="${{ steps.build.outputs.store-path }}"
echo "=== Copying ${{ matrix.service }}: $STORE_PATH ==="
if [ -n "${{ secrets.ATTIC_TOKEN }}" ] && ssh "$VPS_USER@$VPS_HOST" "sudo /nix/var/nix/profiles/default/bin/nix-store --realise '$STORE_PATH'" 2>/dev/null; then
echo "Substituted ${{ matrix.service }} from Attic cache"
else
echo "Attic substitute failed; falling back to ssh copy"
nix copy --to "ssh://$VPS_USER@$VPS_HOST" "$STORE_PATH"
fi
echo "=== Updating profile ==="
ssh "$VPS_USER@$VPS_HOST" "sudo /nix/var/nix/profiles/default/bin/nix-env --profile /nix/var/nix/profiles/gmw-${{ matrix.service }} --set '$STORE_PATH'"
echo "=== Restarting service ===\n"
ssh "$VPS_USER@$VPS_HOST" \
"sudo systemctl daemon-reload && sudo systemctl restart gmw-${{ matrix.service }} && for i in \$(seq 1 15); do state=\$(sudo systemctl is-active gmw-${{ matrix.service }} 2>/dev/null || echo inactive); [ \"\$state\" = \"active\" ] && break; sleep 2; done; echo \"final-state=\$state\"; [ \"\$state\" = \"active\" ]"
echo "✅ gmw-${{ matrix.service }} deployed"
cleanup:
# Bersihkan sampah Nix di VPS SETELAH semua deploy selesai: hapus generasi
# profile lama + nix store gc. Profil yang sedang dipakai tidak disentuh.
needs: build-and-deploy
if: always()
runs-on: ubuntu-latest
steps:
- name: Nix GC on VPS
env:
VPS_HOST: ${{ secrets.VPS_HOST }}
VPS_USER: ${{ secrets.VPS_USER }}
SSH_KEY: ${{ secrets.SSH_PRIVATE_KEY }}
run: |
mkdir -p ~/.ssh
echo "$SSH_KEY" > ~/.ssh/id_ed25519
chmod 600 ~/.ssh/id_ed25519
ssh-keyscan -H "$VPS_HOST" >> ~/.ssh/known_hosts 2>/dev/null
ssh "$VPS_USER@$VPS_HOST" "sudo /usr/local/bin/nix-gc-vps.sh" || echo "⚠️ Nix GC gagal (non-fatal)"