cascade/.github/workflows/build-cli.yaml at main · stablekernel/cascade · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
# Build cascade CLI
# Called by orchestrate workflow - no direct triggers
# Distribution: Use `go install github.com/stablekernel/cascade/cmd/cascade@latest`
name: Build CLI

on:
  workflow_call:
    inputs:
      sha:
        description: 'SHA to build'
        type: string
        required: false
    outputs:
      result:
        description: 'Build result (success/failure)'
        value: ${{ jobs.e2e-tests.result }}

permissions:
  contents: read

jobs:
  build:
    name: Build
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3

      - uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
        with:
          go-version: '1.23'
          cache: true

      - name: Build binary
        run: go build -o cascade ./cmd/cascade

      - name: Verify CLI works
        run: |
          ./cascade version
          ./cascade --help
          ./cascade parse-config --help
          ./cascade detect-changes --help
          ./cascade generate-changelog --help
          ./cascade generate-workflow --help

  e2e-tests:
    name: E2E Tests
    runs-on: ubuntu-latest
    needs: [build]
    # The act + gitea scenarios run serially (-parallel 1, see the test step),
    # which trades wall-clock for reliability, so the job needs more than the
    # default headroom. 70m covers the 60m go test budget plus checkout, Go
    # install, and container teardown.
    timeout-minutes: 70
    steps:
      - uses: actions/checkout@df4cb1c069e1874edd31b4311f1884172cec0e10 # v6.0.3

      - uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
        with:
          # This job runs the e2e module (go 1.24); install the Go it declares
          # so `go test` doesn't hard-fail under GOTOOLCHAIN=local with
          # "go.mod requires go >= 1.24.0". The root build job above stays on
          # the root go.mod's 1.23.
          go-version-file: e2e/go.mod
          cache: true
          cache-dependency-path: e2e/go.sum

      - name: Widen Docker network address pool
        # Each scenario creates its own docker network, and the daemon's
        # default address pool is small. Even with synchronous per-scenario
        # network teardown, brief cleanup lag under serial load can leave the
        # pool short and fail a later scenario at setup with "all predefined
        # address pools have been fully subnetted". Carving /24 subnets out of
        # a 10.99.0.0/16 base yields 256 networks of headroom, well above the
        # handful in flight at once, so cleanup lag can never exhaust the pool.
        run: |
          sudo mkdir -p /etc/docker
          echo '{"default-address-pools":[{"base":"10.99.0.0/16","size":24}]}' | sudo tee /etc/docker/daemon.json
          sudo systemctl restart docker
          # Wait for the daemon to come back before the tests start.
          timeout 60 sh -c 'until docker info >/dev/null 2>&1; do sleep 1; done'

      - name: Run E2E tests
        working-directory: e2e
        # Run scenarios serially (-parallel 1). The 4-core / ~7.9GB runner is
        # OOM-killed at 4 (silent FAIL, see #104), and even at 2 the concurrent
        # act + gitea container load throttles gitea (405 "try again later") and
        # destabilises act runs, producing intermittent failures unrelated to
        # the product. Serial execution removes that contention; the longer
        # 60m timeout covers the resulting slower wall-clock.
        run: go test -v -parallel 1 -timeout 60m ./...

      # On a retry-exhausted scenario the harness writes the last attempt's raw
      # act stdout/stderr to e2e/_artifacts/<scenario>-attempt<N>.log so the
      # stack-trace origin survives the CI log retention window. Upload it on
      # every run (always()) so the evidence is recoverable whether the job
      # failed or a flake was absorbed. The directory may not exist when no
      # scenario exhausted its retries; if-no-files-found: ignore keeps that a
      # clean no-op rather than a warning.
      - name: Upload e2e crash evidence
        if: always()
        uses: actions/upload-artifact@043fb46d1a93c77aae656e7c1c64a875d1fc6a0a # v7.0.1
        with:
          name: e2e-crash-evidence
          path: e2e/_artifacts/
          if-no-files-found: ignore
          retention-days: 14