Files
lancedb/.github/workflows/npm-publish.yml
Will Jones ff50e698cf ci: cut Actions cost by moving builds to free runners and fixing caches (#3735)
Standard GitHub-hosted runners are free on public repos, so all Actions
spend here is on the `*-8x-*` / `4x` larger runners. Measured over 30
days at current (post-Jan-2026) larger-runner rates, that is ~$1,400/mo,
and `npm-publish` is ~70% of it.

## Changes

**Fat LTO was forcing builds onto large runners.** `[profile.release]`
in `.cargo/config.toml` sets `lto = "fat"` with `codegen-units = 1`,
which is single-threaded and the peak-memory step. The macOS
`npm-publish` build was 111 of its 113 minutes in one `napi build` step,
making it the critical path of the whole publish pipeline. The ThinLTO
override already applied to Windows now covers macOS too, and both
Windows builds move from `windows-2025-8x-x64` to the free standard
`windows-2025`.

**The npm-publish cargo cache never existed.** There are zero caches
with its key prefix. The key was static, so `actions/cache` (which only
writes on a miss) could never refresh it, and a multi-GB release
`target/` per target could never fit the repo's 10 GB budget anyway. Now
caches only the crate registry, keyed on `Cargo.lock`. The docker builds
also mounted `.cargo/registry/*` while the cache saved `.cargo-cache`,
so containers re-downloaded the registry every run.

**Cache eviction thrash.** Repo cache usage is 10.4 GB against GitHub's
10 GB cap, so every PR run evicted main's warm entries. `rust.yml` and
`nodejs.yml` now restore everywhere but only save from `main`.

**npm-publish moves to nightly + tags** instead of every push to main
(~90/month). The cross-compiled targets do need watching, so
`report-failure` now fires on scheduled runs, and dedupes onto an
existing open issue rather than filing one per night.

**rust.yml aarch64-pc-windows-msvc** cross-compiled its tests and then
skipped them, paying full codegen and link cost for a compile check.
`windows-11-arm` is now GA and free on public repos, so it builds and
tests natively. Its test step also passes `--target` — without it cargo
used `target/ci/` rather than `target/<triple>/ci/` and rebuilt the
entire dependency graph a second time.

**pypi-publish.yml had no concurrency group**, so force-pushes left a
~74 minute Windows job running.

## What is cost vs. wall-clock

| Change | Cost | Wall-clock |
|---|---|---|
| Windows npm-publish → free runners | **−$570/mo** | slower per job
(8→4 cores) |
| npm-publish nightly | **−$125/mo** | — |
| pypi-publish concurrency | small | — |
| macOS ThinLTO | $0 (already free) | **−~50 min** per release |
| rust aarch64 Windows native | $0 (already free) | **−~25 min** |
| rust `--target` on test step | $0 | large, avoids a second full build
|
| rust-cache `save-if` | small | faster via real cache hits |

## Risks

- The two Windows builds now have 4 cores instead of 8 and ~14 GB of
free disk. If they fail, it is most likely disk rather than memory;
fallback is `windows-2025-4x-x64`, which still halves that line.
- `windows-11-arm` has a thinner toolset (choco/vcpkg/protoc under
emulation) and this enables a test step that has never run, so it may
surface real aarch64 failures. That is the point, but it is the change
most likely to need iteration.
- ThinLTO applies to published macOS and Windows binaries, typically
within a few percent of fat LTO. Linux release builds are untouched.

## Follow-ups

- `python.yml` `pydantic1x` (37 min) and `Doctest` (33 min) each rebuild
the extension from source via `pip install -e .` with no Rust cache;
they should consume the wheel the `linux` job already builds. Worth
~$235/mo and ~70 min of compute per run. Separate PR.
- The three `ubuntu-2404-8x-x64` npm-publish builds (~$420/mo at the old
cadence) are the remaining large-runner spend;
`aarch64-unknown-linux-gnu` could run natively on free
`ubuntu-24.04-arm`. Worth doing after this lands so the ThinLTO change
can be validated first.
- The wheel composite actions declare `python-minor-version` as required
but never use it, and every caller omits it (actionlint warns).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-07-28 14:38:21 -07:00

421 lines
17 KiB
YAML

name: NPM Publish
env:
MACOSX_DEPLOYMENT_TARGET: '10.13'
CARGO_INCREMENTAL: '0'
permissions:
contents: write
id-token: write
on:
push:
tags:
- "v*"
# The cross-compiled targets (musl especially) break from toolchain and
# dependency changes that nothing else in CI catches, and discovering that
# mid-release is expensive. A nightly run keeps that signal while dropping
# the full 8-target release matrix from all ~90 pushes to main each month.
# `report-failure` files an issue when a nightly breaks.
schedule:
- cron: "0 8 * * *"
workflow_dispatch:
pull_request:
# This should trigger a dry run (we skip the final publish step)
paths:
- .github/workflows/npm-publish.yml
- Cargo.toml # Change in dependency frequently breaks builds
- Cargo.lock
concurrency:
group: ${{ github.workflow }}-${{ github.ref }}
cancel-in-progress: true
jobs:
build-lancedb:
strategy:
fail-fast: false
matrix:
settings:
- target: aarch64-apple-darwin
host: macos-latest
features: fp16kernels
pre_build: |-
brew install protobuf
# Fat LTO (the workspace default in .cargo/config.toml) is
# single-threaded and is the peak-memory step of the build. On
# this runner it accounted for ~111 of the job's ~113 minutes,
# making it the critical path of the entire publish pipeline.
# ThinLTO parallelizes it across the runner's cores, for a few
# percent of runtime performance.
export CARGO_PROFILE_RELEASE_LTO=thin
export CARGO_PROFILE_RELEASE_CODEGEN_UNITS=16
- target: x86_64-pc-windows-msvc
host: windows-2025
features: ","
pre_build: |-
choco install --no-progress protoc ninja nasm
tail -n 1000 /c/ProgramData/chocolatey/logs/chocolatey.log
# There is an issue where choco doesn't add nasm to the path
export PATH="$PATH:/c/Program Files/NASM"
nasm -v
# See the ThinLTO note on aarch64-apple-darwin above. Keeping
# peak memory down is also what lets this run on the standard
# 4-core runner: the 8-core larger runner was only needed to
# stop fat LTO from OOMing rustc-LLVM.
export CARGO_PROFILE_RELEASE_LTO=thin
export CARGO_PROFILE_RELEASE_CODEGEN_UNITS=16
- target: aarch64-pc-windows-msvc
host: windows-2025
features: ","
pre_build: |-
choco install --no-progress protoc
rustup target add aarch64-pc-windows-msvc
# See the ThinLTO note on aarch64-apple-darwin above.
export CARGO_PROFILE_RELEASE_LTO=thin
export CARGO_PROFILE_RELEASE_CODEGEN_UNITS=16
- target: x86_64-unknown-linux-gnu
host: ubuntu-latest
features: fp16kernels
# https://github.com/napi-rs/napi-rs/blob/main/debian.Dockerfile
docker: ghcr.io/napi-rs/napi-rs/nodejs-rust:lts-debian
pre_build: |-
set -e &&
apt-get update &&
apt-get install -y protobuf-compiler pkg-config &&
# The base image (manylinux2014-cross) sets TARGET_CC to the old
# GCC 4.8 cross-compiler. aws-lc-sys checks TARGET_CC before CC,
# so it picks up GCC even though the napi-rs image sets CC=clang.
# Override to use the image's clang-18 which supports -fuse-ld=lld.
export TARGET_CC=clang TARGET_CXX=clang++
- target: x86_64-unknown-linux-musl
# This one seems to need some extra memory
host: ubuntu-2404-8x-x64
features: fp16kernels
pre_build: |-
set -e &&
sudo apt-get update &&
sudo apt-get install -y protobuf-compiler pkg-config &&
rustup target add x86_64-unknown-linux-musl &&
export EXTRA_ARGS="-x"
- target: aarch64-unknown-linux-gnu
host: ubuntu-2404-8x-x64
# https://github.com/napi-rs/napi-rs/blob/main/debian-aarch64.Dockerfile
docker: ghcr.io/napi-rs/napi-rs/nodejs-rust:lts-debian-aarch64
features: "fp16kernels"
pre_build: |-
set -e &&
apt-get update &&
apt-get install -y protobuf-compiler pkg-config &&
export TARGET_CC=clang TARGET_CXX=clang++ &&
# The manylinux2014 sysroot has glibc 2.17 headers which lack
# AT_HWCAP2 (added in Linux 3.17). Define it for aws-lc-sys.
export CFLAGS="$CFLAGS -DAT_HWCAP2=26" &&
rustup target add aarch64-unknown-linux-gnu
- target: aarch64-unknown-linux-musl
host: ubuntu-2404-8x-x64
features: ","
pre_build: |-
set -e &&
sudo apt-get update &&
sudo apt-get install -y protobuf-compiler &&
rustup target add aarch64-unknown-linux-musl &&
export EXTRA_ARGS="-x"
name: build - ${{ matrix.settings.target }}
runs-on: ${{ matrix.settings.host }}
defaults:
run:
working-directory: nodejs
steps:
- uses: actions/checkout@v6
- name: Setup pnpm
uses: pnpm/action-setup@v6
with:
version: 11.1.1
- name: Setup node
uses: actions/setup-node@v6
with:
# pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
# in October.
node-version: 24
cache: pnpm
cache-dependency-path: nodejs/pnpm-lock.yaml
- name: Install
uses: dtolnay/rust-toolchain@stable
if: ${{ !matrix.settings.docker }}
with:
toolchain: stable
targets: ${{ matrix.settings.target }}
# These builds were entirely uncached: the old key was static, so
# `actions/cache` (which only writes on a miss) could never refresh it,
# and the multi-GB whole-`target/` copy it tried to store never fit the
# repo's cache budget, so no entry was ever saved. rust-cache prunes
# `target/` to dependency artifacts and keys on Cargo.lock plus the rustc
# version, which both fixes the key and keeps entries a sane size.
#
# This caches dependency *compilation* only. The LTO link of the cdylib
# re-runs regardless, since the local crate changes every time, so the
# win is larger on the non-LTO jobs than here.
- name: Cache cargo (native builds)
uses: Swatinem/rust-cache@v2
if: ${{ !matrix.settings.docker }}
with:
# The release profile and per-target dirs differ from what the test
# workflows cache, so these need to be separate entries.
key: release-${{ matrix.settings.target }}
# Only the nightly run on main writes, so tag and PR runs restore a
# warm entry without every dependabot PR writing its own (which would
# be unreadable elsewhere anyway, since GitHub scopes caches to the
# creating ref). The nightly cadence also keeps entries inside
# GitHub's 7-day eviction window, which a tag-only trigger would not.
save-if: ${{ github.ref == 'refs/heads/main' }}
# Docker builds can use rust-cache too. `target/` already lives on the
# host because the whole workspace is bind-mounted into the container, and
# rust-cache's prune and save run host-side, so they can manage it -- which
# is what keeps the entry to dependency artifacts rather than a multi-GB
# copy of everything.
#
# Two differences from the native builds. The container's CARGO_HOME is
# bind-mounted from `.cargo-cache` rather than the host's ~/.cargo, so that
# has to be cached explicitly. And the key is derived from the *host* rustc
# version, which is not the compiler that produced these artifacts; that is
# safe because cargo fingerprints the real compiler and rebuilds on a
# mismatch, it just means a base-image toolchain bump costs one cold build
# instead of invalidating the key.
- name: Cache cargo (docker builds)
uses: Swatinem/rust-cache@v2
if: ${{ matrix.settings.docker }}
with:
key: docker-${{ matrix.settings.target }}
cache-directories: .cargo-cache
save-if: ${{ github.ref == 'refs/heads/main' }}
- name: Install dependencies
run: pnpm install --frozen-lockfile
- name: Install Zig
uses: mlugg/setup-zig@v2
if: ${{ contains(matrix.settings.target, 'musl') }}
with:
version: 0.14.1
- name: Install cargo-zigbuild
uses: taiki-e/install-action@v2
if: ${{ contains(matrix.settings.target, 'musl') }}
with:
tool: cargo-zigbuild
- name: Build in docker
uses: addnab/docker-run-action@v3
if: ${{ matrix.settings.docker }}
with:
image: ${{ matrix.settings.docker }}
# All three mounts must live under `.cargo-cache`, which is what the
# cache step above saves. Previously the registry mounts pointed at
# `.cargo/...`, a path nothing cached, so the container re-downloaded
# the whole crate registry on every run.
options: "--user 0:0 -v ${{ github.workspace }}/.cargo-cache/git/db:/usr/local/cargo/git/db \
-v ${{ github.workspace }}/.cargo-cache/registry/cache:/usr/local/cargo/registry/cache \
-v ${{ github.workspace }}/.cargo-cache/registry/index:/usr/local/cargo/registry/index \
-v ${{ github.workspace }}:/build -w /build/nodejs"
run: |
set -e
${{ matrix.settings.pre_build }}
npx napi build --platform --release \
--features ${{ matrix.settings.features }} \
--target ${{ matrix.settings.target }} \
--dts ../lancedb/native.d.ts \
--js ../lancedb/native.js \
--strip \
--output-dir dist/
# The container runs as root (`--user 0:0`), so everything it wrote to the
# mounted cache dirs is root-owned. rust-cache's post step runs as the
# runner user and has to both read these and delete from them while
# pruning, so hand them back before it runs.
- name: Take ownership of docker build output
if: ${{ matrix.settings.docker }}
run: |
sudo chown -R "$(id -u):$(id -g)" \
"${{ github.workspace }}/.cargo-cache" \
"${{ github.workspace }}/target"
- name: Build
run: |
${{ matrix.settings.pre_build }}
npx napi build --platform --release \
--features ${{ matrix.settings.features }} \
--target ${{ matrix.settings.target }} \
--dts ../lancedb/native.d.ts \
--js ../lancedb/native.js \
--strip \
$EXTRA_ARGS \
--output-dir dist/
if: ${{ !matrix.settings.docker }}
shell: bash
# The standard Windows runners have ~14 GB free, and a release `target/`
# for this workspace is a large fraction of that. Report the remaining
# headroom so a build that only just fits is visible before a dependency
# bump turns it into a failed release. `always()` so the numbers are
# still there when the build is what ran out of space.
- name: Report disk headroom
if: always()
run: df -h
shell: bash
- name: Upload artifact
uses: actions/upload-artifact@v7
with:
name: lancedb-${{ matrix.settings.target }}
path: nodejs/dist/*.node
if-no-files-found: error
# The generic files are the same in all distros so we just pick
# one to do the upload.
- name: Make generic artifacts
if: ${{ matrix.settings.target == 'aarch64-apple-darwin' }}
run: pnpm tsc
- name: Upload Generic Artifacts
if: ${{ matrix.settings.target == 'aarch64-apple-darwin' }}
uses: actions/upload-artifact@v7
with:
name: nodejs-dist
path: |
nodejs/dist/*
!nodejs/dist/*.node
test-lancedb:
name: "Test: ${{ matrix.settings.target }} - node@${{ matrix.node }}"
needs:
- build-lancedb
strategy:
fail-fast: false
matrix:
settings:
# TODO: Get tests passing on Windows (failing from test tmpdir issue)
# - host: windows-latest
# target: x86_64-pc-windows-msvc
- host: macos-latest
target: aarch64-apple-darwin
- target: x86_64-unknown-linux-gnu
host: ubuntu-latest
- target: aarch64-unknown-linux-gnu
host: ubuntu-2404-8x-arm64
node:
- '20'
runs-on: ${{ matrix.settings.host }}
defaults:
run:
shell: bash
working-directory: nodejs
steps:
- uses: actions/checkout@v6
- name: Setup pnpm
uses: pnpm/action-setup@v6
with:
version: 11.1.1
- name: Setup Node.js 24 for install
uses: actions/setup-node@v6
with:
# pnpm 11 requires Node >= 22.13; use 24 since 22 hits EOL
# in October.
node-version: 24
cache: pnpm
cache-dependency-path: nodejs/pnpm-lock.yaml
- name: Install dependencies
run: pnpm install --frozen-lockfile
- name: Setup Node.js ${{ matrix.node }} for test
uses: actions/setup-node@v6
with:
node-version: ${{ matrix.node }}
- name: Download artifacts
uses: actions/download-artifact@v8
with:
name: lancedb-${{ matrix.settings.target }}
path: nodejs/dist/
# For testing purposes:
# run-id: 13982782871
# github-token: ${{ secrets.GITHUB_TOKEN }} # token with actions:read permissions on target repo
- uses: actions/download-artifact@v8
with:
name: nodejs-dist
path: nodejs/dist
# For testing purposes:
# github-token: ${{ secrets.GITHUB_TOKEN }} # token with actions:read permissions on target repo
# run-id: 13982782871
- name: List packages
run: ls -R dist
- name: Move built files
run: cp dist/native.d.ts dist/native.js dist/*.node lancedb/
- name: Test bindings
# Invoke jest directly because pnpm 11 itself requires Node 22+
# while the matrix tests on older Node versions.
run: npx jest --verbose
publish:
name: Publish
runs-on: ubuntu-latest
defaults:
run:
shell: bash
working-directory: nodejs
needs:
- test-lancedb
steps:
- uses: actions/checkout@v6
- name: Setup pnpm
uses: pnpm/action-setup@v6
with:
version: 11.1.1
- name: Setup node
uses: actions/setup-node@v6
with:
node-version: 24
cache: pnpm
cache-dependency-path: nodejs/pnpm-lock.yaml
registry-url: "https://registry.npmjs.org"
- name: Install dependencies
run: pnpm install --frozen-lockfile
- uses: actions/download-artifact@v8
with:
name: nodejs-dist
path: nodejs/dist
# For testing purposes:
# run-id: 13982782871
# github-token: ${{ secrets.GITHUB_TOKEN }} # token with actions:read permissions on target repo
- uses: actions/download-artifact@v8
name: Download arch-specific binaries
with:
pattern: lancedb-*
path: nodejs/nodejs-artifacts
merge-multiple: true
# For testing purposes:
# run-id: 13982782871
# github-token: ${{ secrets.GITHUB_TOKEN }} # token with actions:read permissions on target repo
- name: Display structure of downloaded files
run: find dist && find nodejs-artifacts
- name: Move artifacts
run: pnpm exec napi artifacts -d nodejs-artifacts
- name: List packages
run: find npm
- name: Publish
env:
DRY_RUN: ${{ !startsWith(github.ref, 'refs/tags/v') }}
run: |
npm config set provenance true
ARGS="--access public"
if [[ $DRY_RUN == "true" ]]; then
ARGS="$ARGS --dry-run"
fi
VERSION=$(node -p "require('./package.json').version")
if [[ $VERSION == *-* ]]; then
ARGS="$ARGS --tag preview"
fi
npm publish $ARGS
report-failure:
name: Report Workflow Failure
runs-on: ubuntu-latest
needs: [build-lancedb, test-lancedb, publish]
# Nightly runs are the only thing watching the cross-compiled targets now,
# so they have to report failures too or the signal is silently lost.
if: always() && failure() && (startsWith(github.ref, 'refs/tags/v') || github.event_name == 'schedule')
permissions:
contents: read
issues: write
steps:
- uses: actions/checkout@v6
- uses: ./.github/actions/create-failure-issue
with:
job-results: ${{ toJSON(needs) }}
workflow-name: ${{ github.workflow }}