6517401f01
Every path-gated workflow begins with one small job -- `does this change touch
it`, or `which workspaces this change touches` for the standards build -- that
all of its real jobs declare in `needs`. All 55 of those gate jobs ran on
`linux-arm64`.
That is the one label the fleet cannot spare. Sampled while 267 runs were
queued, every waiting run had the same shape:
[waiting] linux-arm64 does this change touch it
[blocked] linux member · dataflow device, simulated
[blocked] macos member · dataflow device absent
237 runs, each holding a gate in the longest queue, and every real job blocked
behind it. Not only `linux` jobs: `macos`, `linux-hw`, `linux-arm64-hw` and
`macos-m3-ultra` jobs were blocked too. The measured effect was `linux-arm64`
263 deep against `linux` **zero**, with jetson-hub's 20 cores taking no job for
a full hour -- starved not for want of work but because the work could not be
unblocked. A gate waiting on the resource it is rationing is head-of-line
blocking, and it makes every other runner in the fleet idle on its behalf.
The gates move to `linux`. Nothing else changes: the diff is 55 `runs-on` lines
and no others, and every touched file still parses.
Why `linux` and not elsewhere. The gate is not free -- it compares
`BASE...HEAD`, which needs the merge base, so `fetch-depth: 0` is a full clone
rather than a ten-second script. `macos` is served by hosts on a Mac Studio
sitting at load 185; `linux` is a separate 20-core box that was measurably
empty. This moves a serialisation point to the least contended label; it does
not remove it, and the gate is still a single point of failure for its
workflow. The structural fix is an early-exit first step inside each job, which
costs one checkout per leg and is a larger change.
Follow-up, deliberately not bundled: jetson-hub runs at capacity 2 with 15 GB of
RAM, so 55 clones still serialise two at a time. Raising capacity helps gates
and hurts the heavy emulated-architecture jobs that share the label, so it wants
measuring on its own.
Co-Authored-By: Claude Opus 5 <[email protected]>
190 lines
8.2 KiB
YAML
190 lines
8.2 KiB
YAML
name: WAI Studio
|
|
|
|
# WAI Studio on the web (wai/wai-studio): one front end, its core in WebAssembly, served by
|
|
# `wai dev` at /_wai/ and hosted by the WAI site.
|
|
#
|
|
# - catalogue · linux: Studio's actions hold to the operation catalogue (scripts/actions.mjs
|
|
# --check: an action with no operation, an operation placed for the browser with no action,
|
|
# a mismatched name), the unit tests that need no core build (the generation and every rule
|
|
# of the drift check broken in a copy, the palette, the local core's calls), and the core's
|
|
# own tests and clippy (core/, wai-studio-core, on the http-conformance vectors);
|
|
# - browser · macos: e2e/run.sh builds wai, the relay, the room service, the core, the room
|
|
# client's packages, the quantum engine, Quantum Studio and Studio; runs every unit test,
|
|
# the core's calls among them, and the drift check; then e2e/studio.e2e.mjs against a
|
|
# `wai dev` stack in Chromium and Firefox: each panel on both hosts, Inspect and Verify on
|
|
# conformance vectors with the results their cases expect, a room joined as a viewer, the
|
|
# palette, axe (WCAG 2.2 A and AA), no console error or failed request;
|
|
# - webkit · linux: the same in WebKit, in the browser image of the Playwright release
|
|
# package-lock.json pins (by digest). The macOS runner is started from SSH, and a process of
|
|
# an SSH session there cannot open WebKit's helper processes: WebKit never launches.
|
|
#
|
|
# `wai dev`'s side (/_wai/ on its origin, nothing listening beyond loopback) is tested in the
|
|
# WAI conformance workflow's quickstart step (tests/cli_studio.rs and the binary's own tests).
|
|
#
|
|
# Push filters match the push's delta, so every path Studio builds on is listed.
|
|
|
|
concurrency:
|
|
group: ${{ github.workflow }}-${{ github.ref }}
|
|
cancel-in-progress: true
|
|
|
|
on:
|
|
push:
|
|
branches: [main]
|
|
paths:
|
|
- 'wai/wai-studio/**'
|
|
- 'wai/wai-rs/**'
|
|
- 'wai/catalogue/**'
|
|
- 'wai/quantum-studio/**'
|
|
- 'wai/wai-web/**'
|
|
- 'wai/wai-quantum/**'
|
|
- 'wai/wai-ops/**'
|
|
- 'wai/wai-live-web/**'
|
|
- 'wai/quickstart/build-web.sh'
|
|
- 'wai/http-conformance/**'
|
|
- 'wai/pack-conformance/stills/**'
|
|
- 'joule-code/joule-code-rs/crates/joule-moq-web/**'
|
|
- 'joule-code/joule-code-rs/crates/joule-moq-relay/**'
|
|
- 'joule-code/joule-code-rs/crates/joule-moq-room/**'
|
|
- 'rust-toolchain.toml'
|
|
- '.github/workflows/wai-studio.yml'
|
|
pull_request:
|
|
paths:
|
|
- 'wai/wai-studio/**'
|
|
- 'wai/wai-rs/**'
|
|
- 'wai/catalogue/**'
|
|
- 'wai/quantum-studio/**'
|
|
- 'wai/wai-web/**'
|
|
- 'wai/wai-quantum/**'
|
|
- 'wai/wai-ops/**'
|
|
- 'wai/wai-live-web/**'
|
|
- 'wai/quickstart/build-web.sh'
|
|
- 'wai/http-conformance/**'
|
|
- 'wai/pack-conformance/stills/**'
|
|
- 'joule-code/joule-code-rs/crates/joule-moq-web/**'
|
|
- 'joule-code/joule-code-rs/crates/joule-moq-relay/**'
|
|
- 'joule-code/joule-code-rs/crates/joule-moq-room/**'
|
|
- 'rust-toolchain.toml'
|
|
- '.github/workflows/wai-studio.yml'
|
|
|
|
jobs:
|
|
# The path filters gate the trigger; this job gates the work from the pull request's own
|
|
# diff (scripts/pr-touches-paths.py, reading the list above), as wai-pack.yml explains.
|
|
relevant:
|
|
name: does this change touch it
|
|
runs-on: linux
|
|
outputs:
|
|
run: ${{ steps.gate.outputs.run }}
|
|
steps:
|
|
- uses: actions/checkout@v5
|
|
with:
|
|
fetch-depth: 0
|
|
- id: gate
|
|
run: |
|
|
set -euo pipefail
|
|
r=$(python3 scripts/pr-touches-paths.py .github/workflows/wai-studio.yml \
|
|
"${{ github.event.pull_request.base.sha }}" \
|
|
"${{ github.event.pull_request.head.sha }}")
|
|
echo "this change touches WAI Studio: $r"
|
|
echo "run=$r" >> "$GITHUB_OUTPUT"
|
|
|
|
catalogue:
|
|
name: the catalogue's actions, the unit tests and the core · linux
|
|
needs: relevant
|
|
if: needs.relevant.outputs.run == 'true'
|
|
runs-on: linux
|
|
timeout-minutes: 60
|
|
steps:
|
|
- uses: actions/checkout@v5
|
|
- name: Install Rust 1.98.0
|
|
shell: bash
|
|
run: |
|
|
export PATH="$HOME/.cargo/bin:$PATH"
|
|
command -v rustup >/dev/null 2>&1 || \
|
|
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --profile minimal --default-toolchain 1.98.0
|
|
rustup toolchain install 1.98.0 --profile minimal --component clippy
|
|
rustup default 1.98.0
|
|
echo "$HOME/.cargo/bin" >> "$GITHUB_PATH"
|
|
- uses: actions/setup-node@v4
|
|
with:
|
|
node-version: '22'
|
|
- name: Studio's actions hold to the catalogue
|
|
working-directory: wai/wai-studio
|
|
run: node scripts/actions.mjs --check
|
|
# The core's calls (test/core.test.mjs) need the core built for the web; the browser
|
|
# job runs them. These need nothing built.
|
|
- name: Unit tests
|
|
shell: bash
|
|
working-directory: wai/wai-studio
|
|
run: |
|
|
set -euo pipefail
|
|
node --test --test-reporter=tap test/actions.test.mjs test/palette.test.mjs test/local-core.test.mjs 2>&1 | tee "$RUNNER_TEMP/unit.log"
|
|
grep -qE '^# pass [1-9]' "$RUNNER_TEMP/unit.log" && grep -qE '^# fail 0$' "$RUNNER_TEMP/unit.log" || { echo "::error::no unit test ran, or one failed"; exit 1; }
|
|
- name: The core's tests and clippy
|
|
shell: bash
|
|
working-directory: wai/wai-studio/core
|
|
run: |
|
|
set -euo pipefail
|
|
cargo test --locked 2>&1 | tee "$RUNNER_TEMP/core.log"
|
|
grep -qE '^test result: ok\. [1-9][0-9]* passed; 0 failed' "$RUNNER_TEMP/core.log" || { echo "::error::no test ran"; exit 1; }
|
|
cargo clippy --locked --all-targets -- -D warnings
|
|
|
|
browser:
|
|
name: Studio in Chromium and Firefox · macos
|
|
needs: relevant
|
|
if: needs.relevant.outputs.run == 'true'
|
|
runs-on: macos
|
|
timeout-minutes: 120
|
|
steps:
|
|
- uses: actions/checkout@v5
|
|
# launchd's PATH omits cargo and node on the macOS host runner.
|
|
- name: Put cargo and node on PATH
|
|
shell: bash
|
|
run: |
|
|
echo "$HOME/.cargo/bin" >> "$GITHUB_PATH"
|
|
echo "/opt/homebrew/bin" >> "$GITHUB_PATH"
|
|
- name: wasm32 target and wasm-bindgen CLI 0.2.127
|
|
shell: bash
|
|
run: |
|
|
rustup target add wasm32-unknown-unknown
|
|
command -v wasm-bindgen >/dev/null && wasm-bindgen --version | grep -q ' 0\.2\.127$' || cargo install --locked wasm-bindgen-cli --version 0.2.127
|
|
- name: The browsers Playwright drives
|
|
shell: bash
|
|
working-directory: wai/wai-studio
|
|
run: |
|
|
set -euo pipefail
|
|
npm ci --ignore-scripts --no-audit --no-fund
|
|
npx playwright install chromium firefox
|
|
- name: Build everything, the unit tests, the drift check, and Studio in Chromium and Firefox
|
|
shell: bash
|
|
env:
|
|
STUDIO_OUT: ${{ runner.temp }}/wai-studio-e2e
|
|
run: nice -n 10 bash wai/wai-studio/e2e/run.sh --browsers chromium,firefox
|
|
|
|
webkit:
|
|
name: Studio in WebKit · linux
|
|
needs: relevant
|
|
if: needs.relevant.outputs.run == 'true'
|
|
runs-on: linux
|
|
timeout-minutes: 150
|
|
# Playwright 1.60.0's image (package-lock.json pins 1.60.0): Node, and WebKit with every
|
|
# library it needs.
|
|
container:
|
|
image: mcr.microsoft.com/playwright@sha256:9bd26ad900bb5e0f4dee75839e957a89ae89c2b7ab1e76050e559790e946b948
|
|
steps:
|
|
- uses: actions/checkout@v5
|
|
# The floor's zstd is C, compiled for wasm32 by clang and archived by llvm-ar.
|
|
- name: Rust 1.98.0 with the wasm32 target, clang and llvm, wasm-bindgen CLI 0.2.127
|
|
shell: bash
|
|
run: |
|
|
set -euo pipefail
|
|
apt-get update -qq
|
|
apt-get install -y -qq --no-install-recommends curl ca-certificates build-essential pkg-config clang llvm python3 git > /dev/null
|
|
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --profile minimal --default-toolchain 1.98.0
|
|
"$HOME/.cargo/bin/rustup" target add wasm32-unknown-unknown
|
|
"$HOME/.cargo/bin/cargo" install --locked wasm-bindgen-cli --version 0.2.127
|
|
echo "$HOME/.cargo/bin" >> "$GITHUB_PATH"
|
|
- name: Build everything, the unit tests, the drift check, and Studio in WebKit
|
|
shell: bash
|
|
env:
|
|
STUDIO_OUT: ${{ runner.temp }}/wai-studio-e2e
|
|
run: bash wai/wai-studio/e2e/run.sh --browsers webkit
|