ncbitaxon/.github/workflows/release.yml at master · obophenotype/ncbitaxon · 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
name: Release NCBITaxon

# Executes the full release workflow documented in README.md:
#   1. make clean all -B            (build ncbitaxon.{owl,obo,json}{,.gz} + diff)
#   2. cd subsets && make all -B    (build taxslim subset products)
#   3. Upload all artifacts as a draft GitHub release for review.
#
# This is intentionally manual-only (workflow_dispatch). NCBITaxon is huge and
# downloads the live NCBI taxdump, so we don't want it firing on every push.

on:
  workflow_dispatch:

permissions:
  contents: write  # required by softprops/action-gh-release to create the release

jobs:
  release:
    name: Build and publish NCBITaxon release
    # NOTE on runner sizing:
    #   * ubuntu-latest on public repos = 4 vCPU / 16 GB RAM / 14 GB SSD.
    #   * The historical Jenkins build uses ROBOT_JAVA_ARGS=-Xmx30G; odk.sh
    #     requests -Xmx20G. 16 GB is the bare minimum and will likely OOM on
    #     `robot convert` of the full ncbitaxon.ttl.
    #   * If this fails with OOMKilled or "Java heap space", switch to a
    #     larger runner label configured at the obophenotype org level
    #     (e.g. ubuntu-latest-8-cores, ubuntu-22.04-large) or a self-hosted
    #     runner with >=32 GB RAM and >=50 GB disk.
    runs-on: ubuntu-latest
    container:
      image: obolibrary/odkfull:v1.6
    timeout-minutes: 360

    steps:
      - name: Report runner resources
        run: |
          echo "=== Memory ==="
          free -h || true
          echo "=== Disk ==="
          df -h
          echo "=== CPU ==="
          nproc
          cat /proc/cpuinfo | grep -m1 'model name' || true

      - name: Checkout repository
        uses: actions/checkout@v4

      - name: Work around git safe.directory inside container
        run: git config --global --add safe.directory "$GITHUB_WORKSPACE"

      - name: Build ontology (make clean all -B)
        env:
          # Conservative heap: leave room for OS + container overhead in 16 GB.
          # Bump this (and the runner) if you move to a larger runner.
          ROBOT_JAVA_ARGS: '-Xmx12G'
          JAVA_OPTS: '-Xmx12G'
        run: make clean all -B

      - name: Build subsets (cd subsets && make all -B)
        env:
          ROBOT_JAVA_ARGS: '-Xmx12G'
          JAVA_OPTS: '-Xmx12G'
        run: cd subsets && make all -B

      - name: List built artifacts
        run: |
          ls -lh ncbitaxon.json.gz \
                 ncbitaxon.obo  ncbitaxon.obo.gz \
                 ncbitaxon.owl  ncbitaxon.owl.gz
          ls -lh subsets/taxslim-disjoint-over-in-taxon.owl \
                 subsets/taxslim.obo \
                 subsets/taxslim.owl \
                 subsets/taxslim.json
          df -h

      - name: Compute default release name
        id: relname
        # GitHub's REST API requires a tag_name even for drafts. We default
        # to today's date (v<YYYY-MM-DD>); the actual tag is chosen in the
        # GitHub UI when the draft is published ("Choose a tag" -> "Create
        # new tag on publish").
        run: echo "name=v$(date +%Y-%m-%d)" >> "$GITHUB_OUTPUT"

      - name: Create draft GitHub release
        uses: softprops/action-gh-release@v2
        with:
          tag_name: ${{ steps.relname.outputs.name }}
          name: ${{ steps.relname.outputs.name }}
          draft: true
          files: |
            ncbitaxon.json.gz
            ncbitaxon.obo
            ncbitaxon.obo.gz
            ncbitaxon.owl
            ncbitaxon.owl.gz
            subsets/taxslim-disjoint-over-in-taxon.owl
            subsets/taxslim.obo
            subsets/taxslim.owl
            subsets/taxslim.json