llama-stack-mirror/.github/workflows/backward-compat.yml
Ashwin Bharambe e809d21357
Some checks failed
SqlStore Integration Tests / test-postgres (3.12) (push) Failing after 1s
Integration Auth Tests / test-matrix (oauth2_token) (push) Failing after 1s
SqlStore Integration Tests / test-postgres (3.13) (push) Failing after 0s
Python Package Build Test / build (3.12) (push) Failing after 1s
Test External Providers Installed via Module / test-external-providers-from-module (venv) (push) Has been skipped
Python Package Build Test / build (3.13) (push) Failing after 1s
Integration Tests (Replay) / Integration Tests (, , , client=, ) (push) Failing after 4s
Unit Tests / unit-tests (3.12) (push) Failing after 4s
Unit Tests / unit-tests (3.13) (push) Failing after 4s
Test External API and Providers / test-external (venv) (push) Failing after 42s
Vector IO Integration Tests / test-matrix (push) Failing after 45s
API Conformance Tests / check-schema-compatibility (push) Successful in 54s
UI Tests / ui-tests (22) (push) Successful in 52s
Pre-commit / pre-commit (push) Successful in 3m28s
feat: add backward compatibility tests for run.yaml (#3952)
This adds automated backward compatibility testing for `run.yaml` files.
As we evolve `StackRunConfig`, changes can inadvertently break existing
user configurations. This workflow catches those breaks before merge.

We test old run.yaml files (from main and the latest release) against
the PR's new code. If configs that worked before now fail, the PR is
blocked unless explicitly acknowledged as a breaking change.

**Two test layers:**
- Schema validation: Quick pytest checks that configs parse without
errors
- Integration tests: Full test suite execution to catch runtime semantic
issues (cross-field validations, provider initialization, etc.)

**What we test against:**
- main branch: Breaking changes here block the PR (this is the gate)
- Latest release: Informational only - shows if we've drifted from what
users have

If tests fail, the PR author must acknowledge the breaking change by
adding `!:` to the PR title (e.g., `feat!: change xyz`) or including
`BREAKING CHANGE:` in a commit message. Once acknowledged, the check
passes with a warning.

These jobs are run:
1. `check-main-compatibility` - Schema validation of all distribution
run.yaml files from main
2. `test-integration-main` - Full integration test suite using main's
ci-tests run.yaml
3. `test-integration-release` - Integration tests with latest release
config (informational)
4. `check-schema-release-compatibility` - Schema checks against release
(informational)

The integration tests catch issues that schema validation alone would
miss, like assertion failures in
`StackRunConfig.validate_server_stores()` or provider-specific runtime
logic.

Resolves #3311
Related to #3237
2025-10-28 21:51:56 -07:00

574 lines
22 KiB
YAML

name: Backward Compatibility Check
run-name: Check backward compatibility for run.yaml configs
on:
pull_request:
branches: [main]
paths:
- 'src/llama_stack/core/datatypes.py'
- 'src/llama_stack/providers/datatypes.py'
- 'src/llama_stack/distributions/**/run.yaml'
- 'tests/backward_compat/**'
- '.github/workflows/backward-compat.yml'
concurrency:
group: ${{ github.workflow }}-${{ github.ref }}
cancel-in-progress: true
jobs:
check-main-compatibility:
name: Check Compatibility with main
runs-on: ubuntu-latest
steps:
- name: Checkout PR branch
uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
with:
fetch-depth: 0 # Need full history to access main branch
- name: Set up Python
uses: actions/setup-python@e797f83bcb11b83ae66e0230d6156d7c80228e7c # v6.0.0
with:
python-version: '3.12'
- name: Install uv
uses: astral-sh/setup-uv@6b9c6063abd6010835644d4c2e1bef4cf5cd0fca # v6.0.1
with:
enable-cache: true
- name: Install dependencies
run: |
uv sync --group dev
- name: Extract run.yaml files from main branch
id: extract_configs
run: |
# Get list of run.yaml paths from main
git fetch origin main
CONFIG_PATHS=$(git ls-tree -r --name-only origin/main | grep "src/llama_stack/distributions/.*/run.yaml$" || true)
if [ -z "$CONFIG_PATHS" ]; then
echo "No run.yaml files found in main branch"
exit 1
fi
# Extract all configs to a temp directory
mkdir -p /tmp/main_configs
echo "Extracting configs from main branch:"
while IFS= read -r config_path; do
if [ -z "$config_path" ]; then
continue
fi
# Extract filename for storage
filename=$(basename $(dirname "$config_path"))
echo " - $filename (from $config_path)"
git show origin/main:"$config_path" > "/tmp/main_configs/${filename}.yaml"
done <<< "$CONFIG_PATHS"
echo ""
echo "Extracted $(ls /tmp/main_configs/*.yaml | wc -l) config files"
- name: Test all configs from main
id: test_configs
continue-on-error: true
run: |
# Run pytest once with all configs parameterized
if COMPAT_TEST_CONFIGS_DIR=/tmp/main_configs uv run pytest tests/backward_compat/test_run_config.py -v; then
echo "failed=false" >> $GITHUB_OUTPUT
else
echo "failed=true" >> $GITHUB_OUTPUT
exit 1
fi
- name: Check for breaking change acknowledgment
id: check_ack
if: steps.test_configs.outputs.failed == 'true'
run: |
echo "Breaking changes detected. Checking for acknowledgment..."
# Check PR title for '!:' marker (conventional commits)
PR_TITLE="${{ github.event.pull_request.title }}"
if [[ "$PR_TITLE" =~ ^[a-z]+\!: ]]; then
echo "✓ Breaking change acknowledged in PR title"
echo "acknowledged=true" >> $GITHUB_OUTPUT
exit 0
fi
# Check commit messages for BREAKING CHANGE:
if git log origin/main..HEAD --format=%B | grep -q "BREAKING CHANGE:"; then
echo "✓ Breaking change acknowledged in commit message"
echo "acknowledged=true" >> $GITHUB_OUTPUT
exit 0
fi
echo "✗ Breaking change NOT acknowledged"
echo "acknowledged=false" >> $GITHUB_OUTPUT
env:
GH_TOKEN: ${{ github.token }}
- name: Evaluate results
if: always()
run: |
FAILED="${{ steps.test_configs.outputs.failed }}"
ACKNOWLEDGED="${{ steps.check_ack.outputs.acknowledged }}"
if [[ "$FAILED" == "true" ]]; then
if [[ "$ACKNOWLEDGED" == "true" ]]; then
echo ""
echo "⚠️ WARNING: Breaking changes detected but acknowledged"
echo ""
echo "This PR introduces backward-incompatible changes to run.yaml."
echo "The changes have been properly acknowledged."
echo ""
exit 0 # Pass the check
else
echo ""
echo "❌ ERROR: Breaking changes detected without acknowledgment"
echo ""
echo "This PR introduces backward-incompatible changes to run.yaml"
echo "that will break existing user configurations."
echo ""
echo "To acknowledge this breaking change, do ONE of:"
echo " 1. Add '!:' to your PR title (e.g., 'feat!: change xyz')"
echo " 2. Add the 'breaking-change' label to this PR"
echo " 3. Include 'BREAKING CHANGE:' in a commit message"
echo ""
exit 1 # Fail the check
fi
fi
test-integration-main:
name: Run Integration Tests with main Config
runs-on: ubuntu-latest
steps:
- name: Checkout PR branch
uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
with:
fetch-depth: 0
- name: Extract ci-tests run.yaml from main
run: |
git fetch origin main
git show origin/main:src/llama_stack/distributions/ci-tests/run.yaml > /tmp/main-ci-tests-run.yaml
echo "Extracted ci-tests run.yaml from main branch"
- name: Setup test environment
uses: ./.github/actions/setup-test-environment
with:
python-version: '3.12'
client-version: 'latest'
setup: 'ollama'
suite: 'base'
inference-mode: 'replay'
- name: Run integration tests with main config
id: test_integration
continue-on-error: true
uses: ./.github/actions/run-and-record-tests
with:
stack-config: /tmp/main-ci-tests-run.yaml
setup: 'ollama'
inference-mode: 'replay'
suite: 'base'
- name: Check for breaking change acknowledgment
id: check_ack
if: steps.test_integration.outcome == 'failure'
run: |
echo "Integration tests failed. Checking for acknowledgment..."
# Check PR title for '!:' marker (conventional commits)
PR_TITLE="${{ github.event.pull_request.title }}"
if [[ "$PR_TITLE" =~ ^[a-z]+\!: ]]; then
echo "✓ Breaking change acknowledged in PR title"
echo "acknowledged=true" >> $GITHUB_OUTPUT
exit 0
fi
# Check commit messages for BREAKING CHANGE:
if git log origin/main..HEAD --format=%B | grep -q "BREAKING CHANGE:"; then
echo "✓ Breaking change acknowledged in commit message"
echo "acknowledged=true" >> $GITHUB_OUTPUT
exit 0
fi
echo "✗ Breaking change NOT acknowledged"
echo "acknowledged=false" >> $GITHUB_OUTPUT
env:
GH_TOKEN: ${{ github.token }}
- name: Evaluate integration test results
if: always()
run: |
TEST_FAILED="${{ steps.test_integration.outcome == 'failure' }}"
ACKNOWLEDGED="${{ steps.check_ack.outputs.acknowledged }}"
if [[ "$TEST_FAILED" == "true" ]]; then
if [[ "$ACKNOWLEDGED" == "true" ]]; then
echo ""
echo "⚠️ WARNING: Integration tests failed with main config but acknowledged"
echo ""
exit 0 # Pass the check
else
echo ""
echo "❌ ERROR: Integration tests failed with main config without acknowledgment"
echo ""
echo "To acknowledge this breaking change, do ONE of:"
echo " 1. Add '!:' to your PR title (e.g., 'feat!: change xyz')"
echo " 2. Include 'BREAKING CHANGE:' in a commit message"
echo ""
exit 1 # Fail the check
fi
fi
test-integration-release:
name: Run Integration Tests with Latest Release (Informational)
runs-on: ubuntu-latest
steps:
- name: Checkout PR branch
uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
with:
fetch-depth: 0
- name: Get latest release
id: get_release
run: |
# Get the latest release from GitHub
LATEST_TAG=$(gh release list --limit 1 --json tagName --jq '.[0].tagName' 2>/dev/null || echo "")
if [ -z "$LATEST_TAG" ]; then
echo "No releases found, skipping release compatibility check"
echo "has_release=false" >> $GITHUB_OUTPUT
exit 0
fi
echo "Latest release: $LATEST_TAG"
echo "has_release=true" >> $GITHUB_OUTPUT
echo "tag=$LATEST_TAG" >> $GITHUB_OUTPUT
env:
GH_TOKEN: ${{ github.token }}
- name: Extract ci-tests run.yaml from release
if: steps.get_release.outputs.has_release == 'true'
id: extract_config
run: |
RELEASE_TAG="${{ steps.get_release.outputs.tag }}"
# Try with src/ prefix first (newer releases), then without (older releases)
if git show "$RELEASE_TAG:src/llama_stack/distributions/ci-tests/run.yaml" > /tmp/release-ci-tests-run.yaml 2>/dev/null; then
echo "Extracted ci-tests run.yaml from release $RELEASE_TAG (src/ path)"
echo "has_config=true" >> $GITHUB_OUTPUT
elif git show "$RELEASE_TAG:llama_stack/distributions/ci-tests/run.yaml" > /tmp/release-ci-tests-run.yaml 2>/dev/null; then
echo "Extracted ci-tests run.yaml from release $RELEASE_TAG (old path)"
echo "has_config=true" >> $GITHUB_OUTPUT
else
echo "::warning::ci-tests/run.yaml not found in release $RELEASE_TAG"
echo "has_config=false" >> $GITHUB_OUTPUT
fi
- name: Setup test environment
if: steps.get_release.outputs.has_release == 'true' && steps.extract_config.outputs.has_config == 'true'
uses: ./.github/actions/setup-test-environment
with:
python-version: '3.12'
client-version: 'latest'
setup: 'ollama'
suite: 'base'
inference-mode: 'replay'
- name: Run integration tests with release config (PR branch)
id: test_release_pr
if: steps.get_release.outputs.has_release == 'true' && steps.extract_config.outputs.has_config == 'true'
continue-on-error: true
uses: ./.github/actions/run-and-record-tests
with:
stack-config: /tmp/release-ci-tests-run.yaml
setup: 'ollama'
inference-mode: 'replay'
suite: 'base'
- name: Checkout main branch to test baseline
if: steps.get_release.outputs.has_release == 'true' && steps.extract_config.outputs.has_config == 'true'
run: |
git checkout origin/main
- name: Setup test environment for main
if: steps.get_release.outputs.has_release == 'true' && steps.extract_config.outputs.has_config == 'true'
uses: ./.github/actions/setup-test-environment
with:
python-version: '3.12'
client-version: 'latest'
setup: 'ollama'
suite: 'base'
inference-mode: 'replay'
- name: Run integration tests with release config (main branch)
id: test_release_main
if: steps.get_release.outputs.has_release == 'true' && steps.extract_config.outputs.has_config == 'true'
continue-on-error: true
uses: ./.github/actions/run-and-record-tests
with:
stack-config: /tmp/release-ci-tests-run.yaml
setup: 'ollama'
inference-mode: 'replay'
suite: 'base'
- name: Report results and post PR comment
if: always() && steps.get_release.outputs.has_release == 'true' && steps.extract_config.outputs.has_config == 'true'
run: |
RELEASE_TAG="${{ steps.get_release.outputs.tag }}"
PR_OUTCOME="${{ steps.test_release_pr.outcome }}"
MAIN_OUTCOME="${{ steps.test_release_main.outcome }}"
if [[ "$PR_OUTCOME" == "failure" && "$MAIN_OUTCOME" == "success" ]]; then
# NEW breaking change - PR fails but main passes
echo "::error::🚨 This PR introduces a NEW breaking change!"
# Check if we already posted a comment (to avoid spam on every push)
EXISTING_COMMENT=$(gh pr view ${{ github.event.pull_request.number }} --json comments --jq '.comments[] | select(.body | contains("🚨 New Breaking Change Detected") and contains("Integration tests")) | .id' | head -1)
if [[ -z "$EXISTING_COMMENT" ]]; then
gh pr comment ${{ github.event.pull_request.number }} --body "## 🚨 New Breaking Change Detected
**Integration tests against release \`$RELEASE_TAG\` are now failing**
⚠️ This PR introduces a breaking change that affects compatibility with the latest release.
- Users on release \`$RELEASE_TAG\` may not be able to upgrade
- Existing configurations may break
The tests pass on \`main\` but fail with this PR's changes.
> **Note:** This is informational only and does not block merge.
> Consider whether this breaking change is acceptable for users."
else
echo "Comment already exists, skipping to avoid spam"
fi
cat >> $GITHUB_STEP_SUMMARY <<EOF
## 🚨 NEW Breaking Change Detected
**Integration tests against release \`$RELEASE_TAG\` FAILED**
⚠️ **This PR introduces a NEW breaking change**
- Tests **PASS** on main branch ✅
- Tests **FAIL** on PR branch ❌
- Users on release \`$RELEASE_TAG\` may not be able to upgrade
- Existing configurations may break
> **Note:** This is informational only and does not block merge.
> Consider whether this breaking change is acceptable for users.
EOF
elif [[ "$PR_OUTCOME" == "failure" ]]; then
# Existing breaking change - both PR and main fail
echo "::warning::Breaking change already exists in main branch"
cat >> $GITHUB_STEP_SUMMARY <<EOF
## ⚠️ Release Compatibility Test Failed (Existing Issue)
**Integration tests against release \`$RELEASE_TAG\` FAILED**
- Tests **FAIL** on main branch ❌
- Tests **FAIL** on PR branch ❌
- This breaking change already exists in main (not introduced by this PR)
> **Note:** This is informational only.
EOF
else
# Success - tests pass
cat >> $GITHUB_STEP_SUMMARY <<EOF
## ✅ Release Compatibility Test Passed
Integration tests against release \`$RELEASE_TAG\` passed successfully.
This PR maintains compatibility with the latest release.
EOF
fi
env:
GH_TOKEN: ${{ github.token }}
check-schema-release-compatibility:
name: Check Schema Compatibility with Latest Release (Informational)
runs-on: ubuntu-latest
steps:
- name: Checkout PR branch
uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # v5.0.0
with:
fetch-depth: 0
- name: Set up Python
uses: actions/setup-python@e797f83bcb11b83ae66e0230d6156d7c80228e7c # v6.0.0
with:
python-version: '3.12'
- name: Install uv
uses: astral-sh/setup-uv@6b9c6063abd6010835644d4c2e1bef4cf5cd0fca # v6.0.1
with:
enable-cache: true
- name: Install dependencies
run: |
uv sync --group dev
- name: Get latest release
id: get_release
run: |
# Get the latest release from GitHub
LATEST_TAG=$(gh release list --limit 1 --json tagName --jq '.[0].tagName' 2>/dev/null || echo "")
if [ -z "$LATEST_TAG" ]; then
echo "No releases found, skipping release compatibility check"
echo "has_release=false" >> $GITHUB_OUTPUT
exit 0
fi
echo "Latest release: $LATEST_TAG"
echo "has_release=true" >> $GITHUB_OUTPUT
echo "tag=$LATEST_TAG" >> $GITHUB_OUTPUT
env:
GH_TOKEN: ${{ github.token }}
- name: Extract configs from release
if: steps.get_release.outputs.has_release == 'true'
id: extract_release_configs
run: |
RELEASE_TAG="${{ steps.get_release.outputs.tag }}"
# Get run.yaml files from the release (try both src/ and old path)
CONFIG_PATHS=$(git ls-tree -r --name-only "$RELEASE_TAG" | grep "llama_stack/distributions/.*/run.yaml$" || true)
if [ -z "$CONFIG_PATHS" ]; then
echo "::warning::No run.yaml files found in release $RELEASE_TAG"
echo "has_configs=false" >> $GITHUB_OUTPUT
exit 0
fi
# Extract all configs to a temp directory
mkdir -p /tmp/release_configs
echo "Extracting configs from release $RELEASE_TAG:"
while IFS= read -r config_path; do
if [ -z "$config_path" ]; then
continue
fi
filename=$(basename $(dirname "$config_path"))
echo " - $filename (from $config_path)"
git show "$RELEASE_TAG:$config_path" > "/tmp/release_configs/${filename}.yaml" 2>/dev/null || true
done <<< "$CONFIG_PATHS"
echo ""
echo "Extracted $(ls /tmp/release_configs/*.yaml 2>/dev/null | wc -l) config files"
echo "has_configs=true" >> $GITHUB_OUTPUT
- name: Test against release configs (PR branch)
id: test_schema_pr
if: steps.get_release.outputs.has_release == 'true' && steps.extract_release_configs.outputs.has_configs == 'true'
continue-on-error: true
run: |
RELEASE_TAG="${{ steps.get_release.outputs.tag }}"
COMPAT_TEST_CONFIGS_DIR=/tmp/release_configs uv run pytest tests/backward_compat/test_run_config.py -v --tb=short
- name: Checkout main branch to test baseline
if: steps.get_release.outputs.has_release == 'true' && steps.extract_release_configs.outputs.has_configs == 'true'
run: |
git checkout origin/main
- name: Install dependencies for main
if: steps.get_release.outputs.has_release == 'true' && steps.extract_release_configs.outputs.has_configs == 'true'
run: |
uv sync --group dev
- name: Test against release configs (main branch)
id: test_schema_main
if: steps.get_release.outputs.has_release == 'true' && steps.extract_release_configs.outputs.has_configs == 'true'
continue-on-error: true
run: |
RELEASE_TAG="${{ steps.get_release.outputs.tag }}"
COMPAT_TEST_CONFIGS_DIR=/tmp/release_configs uv run pytest tests/backward_compat/test_run_config.py -v --tb=short
- name: Report results and post PR comment
if: always() && steps.get_release.outputs.has_release == 'true' && steps.extract_release_configs.outputs.has_configs == 'true'
run: |
RELEASE_TAG="${{ steps.get_release.outputs.tag }}"
PR_OUTCOME="${{ steps.test_schema_pr.outcome }}"
MAIN_OUTCOME="${{ steps.test_schema_main.outcome }}"
if [[ "$PR_OUTCOME" == "failure" && "$MAIN_OUTCOME" == "success" ]]; then
# NEW breaking change - PR fails but main passes
echo "::error::🚨 This PR introduces a NEW schema breaking change!"
# Check if we already posted a comment (to avoid spam on every push)
EXISTING_COMMENT=$(gh pr view ${{ github.event.pull_request.number }} --json comments --jq '.comments[] | select(.body | contains("🚨 New Schema Breaking Change Detected")) | .id' | head -1)
if [[ -z "$EXISTING_COMMENT" ]]; then
gh pr comment ${{ github.event.pull_request.number }} --body "## 🚨 New Schema Breaking Change Detected
**Schema validation against release \`$RELEASE_TAG\` is now failing**
⚠️ This PR introduces a schema breaking change that affects compatibility with the latest release.
- Users on release \`$RELEASE_TAG\` will not be able to upgrade
- Existing run.yaml configurations will fail validation
The tests pass on \`main\` but fail with this PR's changes.
> **Note:** This is informational only and does not block merge.
> Consider whether this breaking change is acceptable for users."
else
echo "Comment already exists, skipping to avoid spam"
fi
cat >> $GITHUB_STEP_SUMMARY <<EOF
## 🚨 NEW Schema Breaking Change Detected
**Schema validation against release \`$RELEASE_TAG\` FAILED**
⚠️ **This PR introduces a NEW schema breaking change**
- Tests **PASS** on main branch ✅
- Tests **FAIL** on PR branch ❌
- Users on release \`$RELEASE_TAG\` will not be able to upgrade
- Existing run.yaml configurations will fail validation
> **Note:** This is informational only and does not block merge.
> Consider whether this breaking change is acceptable for users.
EOF
elif [[ "$PR_OUTCOME" == "failure" ]]; then
# Existing breaking change - both PR and main fail
echo "::warning::Schema breaking change already exists in main branch"
cat >> $GITHUB_STEP_SUMMARY <<EOF
## ⚠️ Release Schema Compatibility Failed (Existing Issue)
**Schema validation against release \`$RELEASE_TAG\` FAILED**
- Tests **FAIL** on main branch ❌
- Tests **FAIL** on PR branch ❌
- This schema breaking change already exists in main (not introduced by this PR)
> **Note:** This is informational only.
EOF
else
# Success - tests pass
cat >> $GITHUB_STEP_SUMMARY <<EOF
## ✅ Release Schema Compatibility Passed
All run.yaml configs from release \`$RELEASE_TAG\` are compatible.
This PR maintains backward compatibility with the latest release.
EOF
fi
env:
GH_TOKEN: ${{ github.token }}