flink-agents/.github/workflows/nightly-e2e.yml at main · apache/flink-agents · 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
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.

name: Nightly E2E Tests

on:
  schedule:
    - cron: '0 0 * * *'
  workflow_dispatch:
    inputs:
      flink-version:
        description: 'Flink version to test (e.g., 2.3.0, 2.2.1)'
        required: false
        default: '2.3.0'
        type: string
  pull_request:
    branches: [main]
    paths:
      - 'examples/**'
      - 'python/flink_agents/examples/**'
      - 'tools/install.sh'
      - 'e2e-test/test-scripts/test_submit_examples_to_flink.sh'
      - '.github/workflows/nightly-e2e.yml'

jobs:
  submit-examples-to-flink:
    name: submit-examples [flink-${{ matrix.flink-version }}]
    runs-on: ubuntu-latest
    # Budget: download ~5m + build ~15m + cluster/submit ~5m + buffer.
    timeout-minutes: 60
    env:
      SKIP_SPOTLESS_CHECK: true
    strategy:
      fail-fast: false
      matrix:
        flink-version: ['2.3.0']
    steps:
      - uses: actions/checkout@v4
      - name: Install java
        uses: actions/setup-java@v4
        with:
          java-version: '11'
          distribution: 'temurin'
      - name: Install python
        uses: actions/setup-python@v4
        with:
          python-version: '3.11'
      - name: Install uv
        run: |
          curl -LsSf https://astral.sh/uv/0.11.0/install.sh | sh
          echo "$HOME/.local/bin" >> $GITHUB_PATH
      - name: Run submit-examples E2E test
        env:
          FLINK_VERSION: ${{ github.event.inputs.flink-version || matrix.flink-version }}
          # Use a compact model that reliably follows the examples' structured
          # output prompts. The script aliases it to the names examples hardcode.
          OLLAMA_CHAT_MODEL: 'qwen3.5:4b'
          OLLAMA_CONTEXT_LENGTH: '2048'
          OLLAMA_EMBED_MODEL: 'nomic-embed-text'
        run: bash e2e-test/test-scripts/test_submit_examples_to_flink.sh
      - name: Upload Flink logs on failure
        if: failure()
        uses: actions/upload-artifact@v4
        with:
          name: flink-logs-${{ matrix.flink-version }}
          path: flink-logs-*.tar.gz
          if-no-files-found: ignore

  checkpoint-recovery:
    runs-on: ubuntu-latest
    # Scheduled and manual runs only. The pull_request trigger above exists for the
    # examples job; extending it to this one would spend a second full
    # download-and-build on every PR matching those paths, and nothing in this
    # repository caches either. Accepted consequence: no pre-merge signal.
    if: github.event_name != 'pull_request'
    # Budget. Typical: Flink download ~5m, tools/build.sh ~15m, plus the wheel and
    # PyFlink install; the last two are unbounded. Worst case is dominated by the
    # download: install.sh runs `curl --max-time 900 --retry 3` with no
    # --retry-max-time and each retry restarts that clock, so four attempts take
    # ~60m of the 75 below on their own — FLINK_BASE_URL defaults to the Apache
    # CDN, which makes that tail unlikely. The script's own waits add up to roughly
    # 22m on top, if every one runs to its deadline (its *_TIMEOUT block).
    timeout-minutes: 75
    env:
      # Null on a scheduled run, so the literal is what runs then; keep it in step
      # with the workflow_dispatch input default above. Leaving the version to the
      # script's own default instead would still run, but the log-artifact name
      # below reads this variable and would lose its version suffix.
      FLINK_VERSION: ${{ github.event.inputs.flink-version || '2.3.0' }}
    steps:
      - uses: actions/checkout@v4
      - name: Install java
        uses: actions/setup-java@v4
        with:
          java-version: '11'
          distribution: 'temurin'
      - name: Install python
        uses: actions/setup-python@v4
        with:
          python-version: '3.11'
      - name: Install uv
        run: |
          curl -LsSf https://astral.sh/uv/0.11.0/install.sh | sh
          echo "$HOME/.local/bin" >> $GITHUB_PATH
      # No model server env: the payload agent answers from a deterministic mock
      # chat model. The script sets SKIP_SPOTLESS_CHECK around its own
      # tools/build.sh call, so no job-level env is needed for that either.
      - name: Run checkpoint-recovery E2E test
        run: bash e2e-test/test-scripts/test_checkpoint_recovery.sh
      - name: Upload Flink logs on failure
        if: failure()
        uses: actions/upload-artifact@v4
        with:
          name: flink-logs-checkpoint-recovery-${{ env.FLINK_VERSION }}
          path: flink-logs-*.tar.gz
          if-no-files-found: ignore