Skip to content

ci(mcp): trigger cloud-mcp deploy after PyPI publish #1050

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Aaron ("AJ") Steers (aaronsteers) wants to merge 4 commits into
base: main
Choose a base branch
from
Open

ci(mcp): trigger cloud-mcp deploy after PyPI publish #1050

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
90e381d
ci(mcp): trigger cloud-mcp deploy after PyPI publish
devin-ai-integration[bot] Jun 27, 2026
2fabc84
ci: lock down deploy_cloud_mcp job permissions
devin-ai-integration[bot] Jun 27, 2026
8c3e7b7
ci: pin peter-evans/repository-dispatch to SHA
devin-ai-integration[bot] Jun 27, 2026
bb0a5fb
ci(mcp): deploy prereleases to cloud-mcp-preview, stable to production
devin-ai-integration[bot] Jun 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 36 additions & 0 deletions .github/workflows/pypi_publish.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -79,3 +79,39 @@ jobs:
- name: Publish to PyPI
if: github.event_name == 'release' || (github.event_name == 'workflow_dispatch' && github.event.inputs.publish == 'true')
uses: pypa/gh-action-pypi-publish@v1.13.0

deploy_cloud_mcp:
name: Deploy Cloud MCP
needs: publish_to_pypi
if: github.event_name == 'release'
runs-on: ubuntu-latest
permissions: {}
steps:
Comment thread
coderabbitai[bot] marked this conversation as resolved.
- name: Wait for PyPI availability
run: sleep 120

- name: Trigger cloud-mcp production deploy
uses: peter-evans/repository-dispatch@ff45666b9427631e3450c54a1bcbee4d9ff4d7c0 # v3
with:
token: ${{ secrets.GITHUB_CI_WORKFLOW_TRIGGER_PAT }}
repository: airbytehq/airbyte-ops-mcp
event-type: deploy-cloud-mcp
client-payload: '{"mcp-server": "cloud-mcp"}'
Comment on lines +90 to +99

@coderabbitai coderabbitai Bot Jun 29, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🩺 Stability & Availability | 🟠 Major | ⚡ Quick win

Could we gate the dispatch on actual PyPI availability instead of a fixed 120s sleep, wdyt?

This still races PyPI propagation. If the new airbyte version takes longer than two minutes to appear, the downstream build can fail or rebuild cloud-mcp against the previous package version. Polling for the exact published version before sending deploy-cloud-mcp would make both release and preview deploys deterministic.

Also applies to: 108-117

🤖 Prompt for AI Agents 
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In @.github/workflows/pypi_publish.yml around lines 90 - 99, Replace the fixed
sleep in the PyPI publish workflow with a check that waits until the newly
published airbyte version is actually available on PyPI before triggering the
repository dispatch. Update the logic around the “Wait for PyPI availability”
step and the subsequent `repository-dispatch` action so deployment only proceeds
once the exact version can be resolved, and apply the same gating to the preview
path referenced by the shared deploy steps.

@devin-ai-integration devin-ai-integration Bot Jun 29, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The 2-minute fixed wait was an explicit design choice by the maintainer. A polling loop would add complexity (need to extract the version, handle PyPI API rate limits, determine timeout/retry logic) for marginal benefit — PyPI propagation almost always completes well within 2 minutes, and even if it doesn't, the downstream Dockerfile.cloud-mcp runs uv pip install airbyte without a version pin, so it would pick up the latest available version regardless.

If we want to tighten this in the future, it could be a follow-up improvement.


deploy_cloud_mcp_preview:
name: Deploy Cloud MCP (Preview)
needs: publish_to_pypi
if: github.event_name == 'workflow_dispatch' && github.event.inputs.publish == 'true'
runs-on: ubuntu-latest
permissions: {}
steps:
- name: Wait for PyPI availability
run: sleep 120

- name: Trigger cloud-mcp preview deploy
uses: peter-evans/repository-dispatch@ff45666b9427631e3450c54a1bcbee4d9ff4d7c0 # v3
with:
token: ${{ secrets.GITHUB_CI_WORKFLOW_TRIGGER_PAT }}
repository: airbytehq/airbyte-ops-mcp
event-type: deploy-cloud-mcp
client-payload: '{"mcp-server": "cloud-mcp", "preview": "true"}'
Loading