Add web deployment support and fix event loop handling#76
Open
LucasOnTheHub wants to merge 14 commits intomassive-com:masterfrom
Open
Add web deployment support and fix event loop handling#76LucasOnTheHub wants to merge 14 commits intomassive-com:masterfrom
LucasOnTheHub wants to merge 14 commits intomassive-com:masterfrom
Conversation
- Add --host/--port CLI args and MCP_HOST/MCP_PORT env vars for HTTP transports - Add configure_server() to set FastMCP host/port settings - Update Dockerfile to default to streamable-http transport and expose port 8000 - Add docker-compose.web.yml for easy HTTP deployment https://claude.ai/code/session_01SqUoq1cMV7cZArVK5wRWgr
Add streamable-http web deployment support for Claude web connector
…tmcp-YGe80 Revert "Add streamable-http web deployment support for Claude web con…
- Align pyproject.toml version to 1.0.3 (was 0.8.3, manifest.json already 1.0.3) - Replace generic description with proper project description - Add missing license, keywords, and project URLs to pyproject.toml - Update README install URLs from v0.8.3 to v1.0.3 - Fix Dockerfile base image to Python 3.12 (matches requires-python) - Remove dead os.environ fallback in index.py (env cleared at startup) - Remove unused os import from index.py https://claude.ai/code/session_01QhkJLFSZANGg1bcgK1aEQW
…pdates-c5NoD Sync versions, metadata, and fix stale references
The FastMCP cloud hosting expects `app.py:mcp` as the entrypoint. This file loads env vars, configures credentials, and exposes the FastMCP server instance as `mcp`. https://claude.ai/code/session_01L6iEMsSTcMkAhB5hftm51q
…or-BRBit Add app.py entrypoint for FastMCP web deployment
When run() is called inside an environment that already has an active asyncio event loop (Prefect Horizon, AWS Lambda, Jupyter), the anyio.run() call inside FastMCP.run() raises RuntimeError. Detect the running loop and spin up a dedicated thread with its own event loop to run the async server, avoiding the conflict. https://claude.ai/code/session_01FYaDEf5J1UWrbg3KL8DRas
…rror-xSCef Fix "Already running asyncio in this thread" error in Prefect/Lambda
The existing asyncio.get_running_loop() check can miss event loops in certain environments (AWS Lambda, Prefect, etc.) where the loop exists but isn't detectable from a synchronous call site. When mass_mcp.run() then tries anyio.run()/asyncio.run(), it fails with a RuntimeError. Add a try/except fallback around mass_mcp.run() that catches the asyncio conflict error and retries in a dedicated thread with its own event loop, matching the behavior of the existing proactive detection path. https://claude.ai/code/session_01Ky8mRZVDKbTyZ16YRnQn2W
…rror-MROKc Fix "Already running asyncio in this thread" error with retry fallback
… hosting - Replace os.environ.clear() with selective secret removal: the total wipe broke PATH, HOME, SSL_CERT_FILE and other vars needed by hosted runtimes (ASGI servers, Prefect Horizon, FastMCP Cloud) for SSL, subprocess spawning, and HTTP transport. - Replace threading.Lock in _get_index() with asyncio.Lock to avoid blocking the event loop in concurrent web request contexts. - Split httpx client cleanup: expose async shutdown_http_client() for ASGI lifespan handlers, keep sync _close_http_client_sync() as atexit fallback for CLI/stdio mode only. - Improve app.py: add security (remove secret env vars after capture), import shutdown_http_client for lifespan use, add logging. https://claude.ai/code/session_01FrTHwD3YLpyRocAauKvATX
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR adds support for web/ASGI deployments (FastMCP Cloud, Prefect Horizon) while fixing event loop conflicts in nested async contexts. It introduces graceful async shutdown, replaces blocking locks with async locks, and improves environment variable handling for hosted runtimes.
Key Changes
app.pyentrypoint for FastMCP cloud/streamable-http hosting with proper environment variable injection and secret cleanupthreading.Lockwithasyncio.Lockin_get_index()to prevent blocking the event loop in ASGI contextsshutdown_http_client()async function for ASGI lifespan handlers_close_http_client_sync()as atexit fallback for CLI usage_get_http_client()run()function to detect already-running event loops and spawn a dedicated thread with its own event loop, preventingRuntimeError: Already running asyncio in this threados.environ.clear()to a targeted keep-list approach, preserving system variables needed by hosted runtimes (PATH, HOME, SSL_CERT_FILE, etc.) while removing only secretsbuild_index()no longer readsMASSIVE_LLMS_TXT_URLfrom environment; it must be passed explicitly viaconfigure_credentials()Implementation Details
_get_index()checks if index is already built before acquiring the async lockrun()handles frameworks (AWS Lambda, Prefect, Jupyter) where event loops exist but aren't detected byasyncio.get_running_loop()app.pyto prevent exposure in hosted environmentshttps://claude.ai/code/session_01FrTHwD3YLpyRocAauKvATX