- Fix body receiver timeout handling to prevent infinite loops
- Add WebSocket data forwarding via callbacks instead of StreamReader
- Fix HTTP/2 stream race condition where DATA frames arrive before first read
- Update WebSocketProtocol constructor (removed reader parameter)
Add double-check after clearing _data_event to prevent deadlock when
data arrives between clear() and wait(). The race condition occurred
when:
1. Task A checks buffer, needs more data
2. Task A clears _data_event
3. Task B (feed_data) sets event
4. Task A awaits on cleared event - deadlock
The fix re-checks the buffer after clear() to catch data that arrived
in the race window.
Also adds tests for edge cases: race condition simulation, EOF during
wait, fragmented message reassembly, and control frames during
fragmentation.
- LimitRequestLine now accepts optional max_size parameter
- Use default max limits when limit_request_line or limit_request_field_size is 0
- Add tests validating default max enforcement (8190 bytes)
- Handle alternate exceptions from fast parser in test_invalid_requests
Require gunicorn_h1c >= 0.4.1 for fast parser mode. Add new exception
types and limit parameters to PythonProtocol for parity with C parser.
Update tests to parametrize across both parser implementations.
Add test suite that exercises both PythonProtocol and H1CProtocol
implementations with identical test cases using pytest parametrization.
Tests cover request line parsing, headers, body handling (Content-Length
and chunked), connection handling, parser reset, and callback behavior.
Add PythonProtocol class that mirrors H1CProtocol callback interface:
- Callbacks: on_message_begin, on_url, on_header, on_headers_complete,
on_body, on_message_complete
- Properties: method, path, http_version, headers, content_length,
is_chunked, should_keep_alive
- Methods: feed(data), reset()
- Supports Content-Length and chunked transfer encoding
Add CallbackRequest adapter for building requests from parser state.
Works with both H1CProtocol (C extension) and PythonProtocol.
Add unit tests for PythonProtocol and CallbackRequest.
- Replace datetime.now() with time.monotonic() for request timing
- Add access_log_enabled property to skip log work when disabled
- Rewrite BodyReceiver with Future-based waiting (no create_task)
- Remove StreamReader for HTTP/1.1, use direct bytearray buffering
- Add BufferReader wrapper for FastAsyncRequest compatibility
- Use pre-cached chunk prefixes in _send_body()
- Convert async methods to sync where no await needed
- Batch response writes (headers + body in single write)
Performance: 4,200 -> 69,500 req/s
Wire HttpParser to ASGI hot path, replacing AsyncRequest.parse() with
direct buffer-based parsing. Add FastAsyncRequest wrapper for body
reading. Replace per-request Queue/Task with BodyReceiver for on-demand
body reading. Keep headers as bytes end-to-end to avoid conversion
overhead. Add backpressure control and keepalive timer. Cache response
status lines and Date header.
Benchmark shows 3x improvement: ~875K req/s for simple GET (was ~340K).
- Integrate gunicorn_h1c fast parser into WSGI Request class
- Add _check_fast_parser() and _parse_fast() methods
- Tests use Python parser for consistent validation behavior
- Update config description to reflect all worker types
- Use os.register_at_fork() to properly handle fork() with asyncio
- Start control server after initial workers spawn, not before
- Change default socket path to /run/gunicorn.ctl (like BIRD)
- Add integration tests for sync, gthread, and gevent workers
Fixes#3509
Add a timeout when waiting for initial request data in worker threads.
If no data arrives within 5 seconds, the connection is deferred back
to the main poller instead of blocking the thread indefinitely.
This fixes a regression from v24 where connections were submitted
directly to the thread pool after accept(). In v23, connections were
registered with the poller first and only submitted when data arrived.
The new hybrid approach maintains the performance benefits for fast
clients (immediate processing) while protecting against slow-client
scenarios that can exhaust the thread pool and cause health check
timeouts.
Changes:
- Add _DEFER sentinel and DEFAULT_WORKER_DATA_TIMEOUT constant
- Add TConn.wait_for_data() method using selectors
- Add TConn.data_ready flag to track data availability
- Add pending_conns deque for deferred connections
- Add on_pending_socket_readable() callback
- Add murder_pending() to clean up timed-out pending connections
- Modify handle() to wait for data with timeout before processing
- Modify finish_request() to handle _DEFER and register with poller
Fixes#3518
- Read chunk size lines and trailers in 64-byte blocks instead of 1 byte
at a time, pushing back excess data to the unreader buffer
- Reuse BytesIO buffers with truncate/seek instead of creating new
objects to reduce GC pressure in AsyncUnreader
- Use bytearray.find() directly instead of converting to bytes first
in header parsing loop
- Use index-based iteration for header parsing instead of list.pop(0)
which is O(n) per pop vs O(1) for index access
Add tests for the optimized parsing code paths.
- Use preexec_fn=os.setsid to create new process group
- Send signals to process group with os.killpg() instead of single process
- Add explicit timeout and graceful-timeout to gunicorn command
- Fixes test failures on PyPy 3.10 where signals weren't propagating properly
Replace signal-based dirty add/remove with protocol messages:
- Add MSG_TYPE_MANAGE to dirty protocol for worker management
- Add MANAGE_OP_ADD and MANAGE_OP_REMOVE operation codes
- Add handle_manage_request() in DirtyArbiter
- Update handlers to send messages instead of SIGTTIN/SIGTTOU signals
New workers only load apps that haven't reached their worker limits.
When all apps are at their limits, returns reason in response.
Only increment num_workers when a worker is actually spawned.
Add MSG_TYPE_STATUS to dirty protocol to allow querying the dirty
arbiter for its workers. The control socket now connects to the
dirty arbiter socket to retrieve worker information.
Add a control socket server and CLI client for runtime management
of Gunicorn instances, similar to birdc for BIRD routing daemon.
Features:
- Control socket server running in arbiter process (asyncio/threaded)
- gunicornc CLI with interactive and single-command modes
- JSON protocol with length-prefixed framing
- Commands: show workers/stats/config/listeners/dirty, worker add/remove/kill,
dirty add/remove, reload, reopen, shutdown
- Stats tracking (uptime, workers spawned/killed, reloads)
- Configurable socket path and permissions
New config options:
- control_socket: Unix socket path (default: gunicorn.ctl)
- control_socket_mode: Socket permissions (default: 0o600)
- --no-control-socket: Disable control socket
Add support for SIGTTIN and SIGTTOU signals to the dirty arbiter,
allowing dynamic scaling of dirty workers at runtime without restarting
gunicorn.
Changes:
- Add TTIN/TTOU to DirtyArbiter.SIGNALS
- Add num_workers instance variable for dynamic count
- Add _get_minimum_workers() to enforce app worker constraints
- Add signal handlers for TTIN (increase) and TTOU (decrease)
- Update manage_workers() to use dynamic count
- Add documentation for dynamic scaling
- Add unit tests for signal handling
- Add Docker integration tests
The minimum worker constraint ensures TTOU cannot reduce workers below
what apps require (e.g., if an app has workers=3, minimum is 3).
Closes#3489
* feat(dirty): add stash - global shared state between workers
Add a simple key-value store (stash) that allows dirty workers to share
state through the arbiter. Tables are stored directly in arbiter memory
for fast access and simplicity.
Features:
- Auto-create tables on first access
- Dict-like interface via stash.table()
- Pattern matching for keys (glob patterns)
- Module-level API: stash.put(), stash.get(), stash.delete(), etc.
Usage:
from gunicorn.dirty import stash
stash.put("sessions", "user:1", {"name": "Alice"})
user = stash.get("sessions", "user:1")
# Or dict-like
sessions = stash.table("sessions")
sessions["user:1"] = {"name": "Alice"}
New files:
- gunicorn/dirty/stash.py - Client API and StashTable class
- Protocol additions for MSG_TYPE_STASH and STASH_OP_* codes
Note: Tables are ephemeral - lost if arbiter restarts.
* test(dirty): add tests for stash protocol and encoding
Test coverage for:
- Stash message creation and encoding
- Protocol constants (MSG_TYPE_STASH, STASH_OP_*)
- Error classes (StashError, StashTableNotFoundError, StashKeyNotFoundError)
- StashTable dict-like interface
- Edge cases: unicode, complex values, special patterns
* example(dirty): add stash usage example and integration tests
- Add SessionApp to dirty_app.py demonstrating stash usage
- Add /session/* endpoints to wsgi_app.py
- Add test_stash_integration.py with Docker tests
- Update docker-compose.yml with stash-test service
- Fix: Set GUNICORN_DIRTY_SOCKET in dirty arbiter for worker access
* docs(dirty): add stash documentation
JSON serializes all dict keys as strings, so for compatibility the TLV
encoder should do the same. This fixes an error when tasks return dicts
with integer keys (e.g., aggregation results grouped by numeric ID).
Update client and streaming tests to work with the binary protocol:
- Update MockStreamWriter/MockStreamReader to use BinaryProtocol
- Replace string request IDs with integers
- Update test assertions to decode binary protocol messages
- Use HEADER_SIZE and decode_header/decode_message instead of old API
Update arbiter tests to work with the binary protocol:
- Update MockStreamWriter to decode binary messages
- Import binary protocol constants from module level
Update worker tests to work with the binary protocol:
- Use integer request IDs instead of strings
- Update MockStreamWriter to decode binary messages
- Import binary protocol constants from module level
Replace JSON-based protocol with binary format using 16-byte header:
- Magic bytes (GD), version, message type, payload length, request ID
- TLV-encoded payloads for efficient binary data transfer
- No base64 encoding needed for binary data
- Backwards compatible API (DirtyProtocol alias, dict-based interface)
Header format inspired by OpenBSD msgctl/msgsnd.
- Bump version to 25.0.2
- Update copyright year to 2026 in LICENSE and NOTICE
- Add license headers to all Python source files
- Add changelog entry for 25.0.2
- Fix nginx config to use keepalive with upstream (was sending
Connection: close which caused premature connection closure)
- Add _safe_write() to handle socket errors (EPIPE, ECONNRESET,
ENOTCONN) gracefully when client disconnects
- Fix ASGI scope server/client to always be 2-tuples for IPv6
compatibility (IPv6 sockets return 4-tuples)
- Add write_eof() before close() to ensure buffered data is flushed
- Bind to [::] for dual-stack IPv4/IPv6 support in test containers
Closes#3484
When a client disconnects during an ASGI request, the worker now:
1. Sends http.disconnect message to the app's receive queue
2. Allows a configurable grace period for cleanup (default: 3 seconds)
3. Only cancels the task after the grace period expires
This follows the ASGI HTTP Connection Scope spec which defines
http.disconnect as the message apps should receive when clients
disconnect: https://asgi.readthedocs.io/en/latest/specs/www.html#disconnect-receive-event
The grace period prevents CancelledError from propagating to async
database operations, allowing SQLAlchemy and other async DB libraries
to properly reset their connection pools.
New config option: --asgi-disconnect-grace-period (default: 3 seconds)
Closes#3482
The dirty module (which uses asyncio and concurrent.futures) was being
imported at gunicorn startup via gunicorn.arbiter. This caused
concurrent.futures to be imported before user code could call
gevent.monkey.patch_all(), breaking gevent's monkey-patching.
Changes:
- gunicorn/arbiter.py: Import DirtyArbiter and set_dirty_socket_path
lazily inside spawn_dirty_arbiter() instead of at module level
- gunicorn/dirty/worker.py: Import ThreadPoolExecutor lazily inside
run() method instead of at module level
- Add tests/workers/test_gevent_import_order.py with 5 tests verifying:
- concurrent.futures is NOT imported when gunicorn.arbiter loads
- gevent patching works correctly with gunicorn
- Reproduces the exact scenario from the bug report gist
This ensures gevent's monkey.patch_all() can run before concurrent.futures
is imported, allowing proper patching of threading primitives.
- Fix path expectation in test_scope_path_preserved (router strips /http prefix)
- Fix lifespan state check to use scope_state instead of module_state
- Add tolerance for partial failures in proxy concurrent test
- Add retry logic with proper assertions in HTTPS proxy FastAPI test
- Add get_app_workers_attribute() to read workers class attribute
- Update _parse_app_specs() to check class attribute when no config override
- Add Docker-based e2e tests for per-app worker allocation
- Add test apps: HeavyModelApp (workers=2), LightweightApp
- Add unit tests for get_app_workers_attribute function
- Add integration tests for class attribute detection
Allow dirty apps to specify how many workers should load them, enabling
significant memory savings for heavy applications like ML models.
- Add `workers` class attribute to DirtyApp (None = all workers)
- Add `parse_dirty_app_spec()` to parse "module:Class:N" format
- Add `DirtyNoWorkersAvailableError` for app-specific error handling
- Update DirtyArbiter with per-app worker tracking and routing
- Maintain backward compatibility when no dirty_apps configured
Example: 8 workers x 10GB model = 80GB RAM needed
With workers=2: 2 x 10GB = 20GB RAM (75% savings)
Configuration formats:
- Class attribute: `workers = 2` on DirtyApp subclass
- Config format: `module:class:N` (e.g., `myapp.ml:HugeModel:2`)
- Add explicit do_handshake() in base_async.py before ALPN check
when do_handshake_on_connect is False
- Mark eventlet worker as deprecated (removal in 26.0)
- Add HTTP/2 gevent example with Docker and tests
- Update documentation to reflect eventlet deprecation
- Remove eventlet websocket example (gevent version exists)
The ALPN fix ensures HTTP/2 works correctly with gevent and eventlet
workers when do_handshake_on_connect config is False (the default).
Without explicit handshake, selected_alpn_protocol() returns None.
- Send GOAWAY with correct error codes for protocol violations
- Handle StreamClosedError and FlowControlError gracefully
- Return False instead of raising for missing/closed streams
- Handle flow control window overflow per RFC 7540
- Fix reader race condition and add h2 exception handling
- Wait for WINDOW_UPDATE when flow control window is zero/negative
- Use h2 exception's error_code for INITIAL_WINDOW_SIZE violations
Implement HTTP 103 Early Hints as modern replacement for HTTP/2 Server Push.
This allows servers to send resource hints before the final response,
enabling browsers to preload assets in parallel.
WSGI support:
- Add wsgi.early_hints callback to environ dict
- Apps can call environ['wsgi.early_hints'](headers) to send 103 responses
- Silently ignored for HTTP/1.0 clients (don't support 1xx responses)
ASGI support:
- Handle http.response.informational message type
- Apps can await send({"type": "http.response.informational", "status": 103, ...})
HTTP/2 support:
- Add send_informational() method to HTTP2ServerConnection
- Add async send_informational() method to AsyncHTTP2Connection
- Wire up early hints in gthread worker for HTTP/2 requests
Includes unit tests and Docker integration tests for all protocols.
