Run ci on bigger machine

[0xLucqs]

kdo

[Eikix]

Hey guys, what are the current bottlenecks? purely CPU cores?

I can suggest one easy path (requires £ and access to gh beta managed runners, like for sayajin-labs) -> GH managed runners

One hard path -> custom self hosted runners on aws, also very expensive and cumbersome to maintain

[0xLucqs]

Thanks we'll do that you regale as always 💯

[Eikix]

Application to GitHub managed runners is underway. We are on the waiting list, hopefully, we'll get an answer soon.

[AbdelStark]

Tried to setup a self hosted runner. The AWS machine is running and correctly configured. However when trying to run the workflow with this self hosted runner, the CI fails. First, I got issues with environment variables and specifically PATH getting overridden (seems to me by actions/checkout@v3 step). The result was that rustup command was not found and the step rustup show was failing. As a workaround, I tried adding source $HOME/.profile as a first step before rustup show. When doing so, this step works but then it fails on the cargo-llvm-cov.

Here is the latest attempt file I tried:

name: Check, Build & Tests

# Controls when the action will run.
on:
  # Triggers the workflow on push or pull request events but only for the master branch
  push:
    branches: [main]
  pull_request:
    branches: [main]

  # Allows you to run this workflow manually from the Actions tab
  workflow_dispatch:

jobs:
  lint:
    runs-on: [self-hosted, madara]
    steps:
      - uses: actions/checkout@v3
      - name: Setup rust toolchain
        run: |
          source $HOME/.profile
          rustup show
      - name: Set-Up
        run: |
          sudo apt-get update
          sudo apt-get install -y clang llvm libudev-dev protobuf-compiler
      - uses: Swatinem/rust-cache@v2
      - name: Format and clippy
        run: |
          cargo fmt --all -- --check
          cargo clippy --all -- -D warnings
          cargo clippy --tests -- -D warnings

  coverage:
    runs-on: [self-hosted, madara]
    steps:
      - uses: actions/checkout@v3
      - name: Setup rust toolchain
        run: |
          source $HOME/.profile
          rustup show
      - name: Set-Up
        run: |
          sudo apt-get update
          sudo apt-get install -y clang llvm libudev-dev protobuf-compiler
      - name: Install cargo-llvm-cov
        uses: taiki-e/install-action@cargo-llvm-cov
      - uses: Swatinem/rust-cache@v2
      - name: Coverage
        run: cargo llvm-cov --codecov --output-path codecov.json
      - name: Upload coverage to codecov.io
        uses: codecov/codecov-action@v3
        with:
          token: ${{ secrets.CODECOV_TOKEN }}
          files: codecov.json
          fail_ci_if_error: true

  integration-tests:
    runs-on: [self-hosted, madara]
    env:
      BINARY_PATH: ../target/debug/madara
    steps:
      - uses: actions/checkout@v3
      - uses: actions/setup-node@v3
        with:
          node-version: ">=20"
          cache: "npm"
          cache-dependency-path: ./tests/package-lock.json
      - name: Install
        run: |-
          cd tests
          npm install
      - name: Setup rust toolchain
        run: |
          source $HOME/.profile
          rustup show
      - name: Set-Up
        run: |
          sudo apt-get update
          sudo apt-get install -y clang llvm libudev-dev protobuf-compiler
      - uses: Swatinem/rust-cache@v2
      - run: cargo build --workspace
      - name: Run test
        run: |-
          cd tests
          npm run test

cc @drspacemn @LucasLvy

[drspacemn]

Okay so it looks like what happened:

• before the service was set up one runner was init manually
• so any changes i was making to the .path file weren't reflecting in the actual runner
• so ~/.cargo/bin wasn't in the path for any of the tooling

I killed that and restarted the service and our path looks good now.

One questions is. Do we want to run multiple runners on this machine so we can build tasks in parallel or are we going to have multiple runner nodes?