langgraph-api 0.3.0__py3-none-any.whl → 0.3.4__py3-none-any.whl

langgraph_api/__init__.py

@@ -1 +1 @@
-__version__ = "0.3.0"
+__version__ = "0.3.4"

langgraph_api/api/meta.py

@@ -54,10 +54,16 @@ async def meta_metrics(request: ApiRequest):
         metadata.PROJECT_ID, metadata.HOST_REVISION_ID, metrics_format
     )
 
+    pg_redis_stats = pool_stats(
+        project_id=metadata.PROJECT_ID,
+        revision_id=metadata.HOST_REVISION_ID,
+        format=metrics_format,
+    )
+
     if metrics_format == "json":
         async with connect() as conn:
             resp = {
-                **pool_stats(),
+                **pg_redis_stats,
                 "queue": await Runs.stats(conn),
                 **http_metrics,
             }
@@ -93,6 +99,7 @@ async def meta_metrics(request: ApiRequest):
                 )
 
             metrics.extend(http_metrics)
+            metrics.extend(pg_redis_stats)
 
         metrics_response = "\n".join(metrics)
         return PlainTextResponse(metrics_response)

langgraph_api/config.py

@@ -180,6 +180,7 @@ REDIS_CLUSTER = env("REDIS_CLUSTER", cast=bool, default=False)
 REDIS_MAX_CONNECTIONS = env("REDIS_MAX_CONNECTIONS", cast=int, default=2000)
 REDIS_CONNECT_TIMEOUT = env("REDIS_CONNECT_TIMEOUT", cast=float, default=10.0)
 REDIS_MAX_IDLE_TIME = env("REDIS_MAX_IDLE_TIME", cast=float, default=120.0)
+REDIS_STREAM_TIMEOUT = env("REDIS_STREAM_TIMEOUT", cast=float, default=30.0)
 REDIS_KEY_PREFIX = env("REDIS_KEY_PREFIX", cast=str, default="")
 RUN_STATS_CACHE_SECONDS = env("RUN_STATS_CACHE_SECONDS", cast=int, default=60)
 
@@ -374,6 +375,7 @@ API_VARIANT = env("LANGSMITH_LANGGRAPH_API_VARIANT", cast=str, default="")
 # UI
 UI_USE_BUNDLER = env("LANGGRAPH_UI_BUNDLER", cast=bool, default=False)
 IS_QUEUE_ENTRYPOINT = False
+IS_EXECUTOR_ENTRYPOINT = False
 ref_sha = None
 if not os.getenv("LANGCHAIN_REVISION_ID") and (
     ref_sha := os.getenv("LANGSMITH_LANGGRAPH_GIT_REF_SHA")

langgraph_api/executor_entrypoint.py

@@ -20,4 +20,7 @@ if __name__ == "__main__":
         uvloop.install()
     except ImportError:
         pass
+    from langgraph_api import config
+
+    config.IS_EXECUTOR_ENTRYPOINT = True
     asyncio.run(main(grpc_port=args.grpc_port, entrypoint_name="python-executor"))

langgraph_api/queue_entrypoint.py

@@ -11,10 +11,8 @@ if not (
 
 import asyncio
 import contextlib
-import http.server
 import json
 import logging.config
-import os
 import pathlib
 import signal
 from contextlib import asynccontextmanager
@@ -22,6 +20,7 @@ from typing import cast
 
 import structlog
 
+from langgraph_runtime.database import pool_stats
 from langgraph_runtime.lifespan import lifespan
 from langgraph_runtime.metrics import get_metrics
 
@@ -29,69 +28,68 @@ logger = structlog.stdlib.get_logger(__name__)
 
 
 async def health_and_metrics_server():
+    import uvicorn
+    from starlette.applications import Starlette
+    from starlette.responses import JSONResponse, PlainTextResponse
+    from starlette.routing import Route
+
     port = int(os.getenv("PORT", "8080"))
-    ok = json.dumps({"status": "ok"}).encode()
-    ok_len = str(len(ok))
-
-    class HealthAndMetricsHandler(http.server.SimpleHTTPRequestHandler):
-        def log_message(self, format, *args):
-            # Skip logging for /ok and /metrics endpoints
-            if getattr(self, "path", None) in ["/ok", "/metrics"]:
-                return
-            # Log other requests normally
-            super().log_message(format, *args)
-
-        def do_GET(self):
-            path = getattr(self, "path", None)
-            if path == "/ok":
-                self.send_response(200)
-                self.send_header("Content-Type", "application/json")
-                self.send_header("Content-Length", ok_len)
-                self.end_headers()
-                self.wfile.write(ok)
-            elif path == "/metrics":
-                metrics = get_metrics()
-                worker_metrics = cast(dict[str, int], metrics["workers"])
-                workers_max = worker_metrics["max"]
-                workers_active = worker_metrics["active"]
-                workers_available = worker_metrics["available"]
-
-                project_id = os.getenv("LANGSMITH_HOST_PROJECT_ID")
-                revision_id = os.getenv("LANGSMITH_HOST_REVISION_ID")
-
-                metrics = [
-                    "# HELP lg_api_workers_max The maximum number of workers available.",
-                    "# TYPE lg_api_workers_max gauge",
-                    f'lg_api_workers_max{{project_id="{project_id}", revision_id="{revision_id}"}} {workers_max}',
-                    "# HELP lg_api_workers_active The number of currently active workers.",
-                    "# TYPE lg_api_workers_active gauge",
-                    f'lg_api_workers_active{{project_id="{project_id}", revision_id="{revision_id}"}} {workers_active}',
-                    "# HELP lg_api_workers_available The number of available (idle) workers.",
-                    "# TYPE lg_api_workers_available gauge",
-                    f'lg_api_workers_available{{project_id="{project_id}", revision_id="{revision_id}"}} {workers_available}',
-                ]
-
-                metrics_response = "\n".join(metrics).encode()
-                metrics_len = str(len(metrics_response))
-
-                self.send_response(200)
-                self.send_header(
-                    "Content-Type", "text/plain; version=0.0.4; charset=utf-8"
-                )
-                self.send_header("Content-Length", metrics_len)
-                self.end_headers()
-                self.wfile.write(metrics_response)
-            else:
-                self.send_error(http.HTTPStatus.NOT_FOUND)
-
-    with http.server.ThreadingHTTPServer(
-        ("0.0.0.0", port), HealthAndMetricsHandler
-    ) as httpd:
-        logger.info(f"Health and metrics server started at http://0.0.0.0:{port}")
-        try:
-            await asyncio.to_thread(httpd.serve_forever)
-        finally:
-            httpd.shutdown()
+
+    async def health_endpoint(request):
+        return JSONResponse({"status": "ok"})
+
+    async def metrics_endpoint(request):
+        metrics = get_metrics()
+        worker_metrics = cast(dict[str, int], metrics["workers"])
+        workers_max = worker_metrics["max"]
+        workers_active = worker_metrics["active"]
+        workers_available = worker_metrics["available"]
+
+        project_id = os.getenv("LANGSMITH_HOST_PROJECT_ID")
+        revision_id = os.getenv("LANGSMITH_HOST_REVISION_ID")
+
+        metrics_lines = [
+            "# HELP lg_api_workers_max The maximum number of workers available.",
+            "# TYPE lg_api_workers_max gauge",
+            f'lg_api_workers_max{{project_id="{project_id}", revision_id="{revision_id}"}} {workers_max}',
+            "# HELP lg_api_workers_active The number of currently active workers.",
+            "# TYPE lg_api_workers_active gauge",
+            f'lg_api_workers_active{{project_id="{project_id}", revision_id="{revision_id}"}} {workers_active}',
+            "# HELP lg_api_workers_available The number of available (idle) workers.",
+            "# TYPE lg_api_workers_available gauge",
+            f'lg_api_workers_available{{project_id="{project_id}", revision_id="{revision_id}"}} {workers_available}',
+        ]
+
+        metrics_lines.extend(
+            pool_stats(
+                project_id=project_id,
+                revision_id=revision_id,
+            )
+        )
+
+        return PlainTextResponse(
+            "\n".join(metrics_lines),
+            media_type="text/plain; version=0.0.4; charset=utf-8",
+        )
+
+    app = Starlette(
+        routes=[
+            Route("/ok", health_endpoint),
+            Route("/metrics", metrics_endpoint),
+        ]
+    )
+
+    config = uvicorn.Config(
+        app,
+        host="0.0.0.0",
+        port=port,
+        log_level="error",
+        access_log=False,
+    )
+    server = uvicorn.Server(config)
+
+    logger.info(f"Health and metrics server started at http://0.0.0.0:{port}")
+    await server.serve()
 
 
 async def entrypoint(

langgraph_api/stream.py

@@ -2,7 +2,7 @@ import uuid
 from collections.abc import AsyncIterator, Callable
 from contextlib import AsyncExitStack, aclosing, asynccontextmanager
 from functools import lru_cache
-from typing import Any, cast
+from typing import Any, Literal, cast
 
 import langgraph.version
 import langsmith
@@ -423,7 +423,7 @@ async def consume(
     stream: AnyStream,
     run_id: str | uuid.UUID,
     resumable: bool = False,
-    stream_modes: set[StreamMode] | None = None,
+    stream_modes: set[StreamMode | Literal["metadata"]] | None = None,
 ) -> None:
     stream_modes = stream_modes or set()
     if "messages-tuple" in stream_modes:

{langgraph_api-0.3.0.dist-info → langgraph_api-0.3.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: langgraph-api
-Version: 0.3.0
+Version: 0.3.4
 Author-email: Nuno Campos <nuno@langchain.dev>, Will Fu-Hinthorn <will@langchain.dev>
 License: Elastic-2.0
 License-File: LICENSE

{langgraph_api-0.3.0.dist-info → langgraph_api-0.3.4.dist-info}/RECORD

@@ -1,12 +1,12 @@
-langgraph_api/__init__.py,sha256=VrXpHDu3erkzwl_WXrqINBm9xWkcyUy53IQOj042dOs,22
+langgraph_api/__init__.py,sha256=oYLGMpySamd16KLiaBTfRyrAS7_oyp-TOEHmzmeumwg,22
 langgraph_api/asgi_transport.py,sha256=XtiLOu4WWsd-xizagBLzT5xUkxc9ZG9YqwvETBPjBFE,5161
 langgraph_api/asyncio.py,sha256=mZ7G32JjrGxrlH4OMy7AKlBQo5bZt4Sm2rlrBcU-Vj8,9483
 langgraph_api/cli.py,sha256=-ruIeKi1imvS6GriOfRDZY-waV4SbWiJ0BEFAciPVYI,16330
 langgraph_api/command.py,sha256=3O9v3i0OPa96ARyJ_oJbLXkfO8rPgDhLCswgO9koTFA,768
-langgraph_api/config.py,sha256=9wXEcxvVfPQjiR7jugomwTJ3vs5d-YoYYfgLDjRc3EQ,12133
+langgraph_api/config.py,sha256=r9mmbyZlhBuJLpnTkaOLcNH6ufFNqm_2eCiuOmhqRl0,12241
 langgraph_api/cron_scheduler.py,sha256=25wYzEQrhPEivZrAPYOmzLPDOQa-aFogU37mTXc9TJk,2566
 langgraph_api/errors.py,sha256=zlnl3xXIwVG0oGNKKpXf1an9Rn_SBDHSyhe53hU6aLw,1858
-langgraph_api/executor_entrypoint.py,sha256=ClMyM9TB9oPisQzHqixA77Lnj_QGUg55MtQx-xku4o8,671
+langgraph_api/executor_entrypoint.py,sha256=CaX813ygtf9CpOaBkfkQXJAHjFtmlScCkrOvTDmu4Aw,750
 langgraph_api/feature_flags.py,sha256=GjwmNjfg0Jhs3OzR2VbK2WgrRy3o5l8ibIYiUtQkDPA,363
 langgraph_api/graph.py,sha256=HTjJNQadrdi1tzJYNJ_iPIR6-zqC4-hj6YTD6zGQHYA,25072
 langgraph_api/http.py,sha256=fyK-H-0UfNy_BzuVW3aWWGvhRavmGAVMkDwDArryJ_4,5659
@@ -14,7 +14,7 @@ langgraph_api/http_metrics.py,sha256=MU9ccXt7aBb0AJ2SWEjwtbtbJEWmeqSdx7-CI51e32o
 langgraph_api/logging.py,sha256=ZZ95dDdWDayIbH1bgwNfn0U3CQ8kDoAvDFBDACna4-A,5150
 langgraph_api/metadata.py,sha256=fVsbwxVitAj4LGVYpCcadYeIFANEaNtcx6LBxQLcTqg,6949
 langgraph_api/patch.py,sha256=iLwSd9ZWoVj6MxozMyGyMvWWbE9RIP5eZX1dpCBSlSU,1480
-langgraph_api/queue_entrypoint.py,sha256=KDLpQtBu3amZTbNHS-RGFLR0DphuVQN6kUZm3ZGLe9g,5991
+langgraph_api/queue_entrypoint.py,sha256=yFzVX3_YKTq4w1A5h5nRpVfiWuSOeJ9acHMPAcTIrKY,5282
 langgraph_api/route.py,sha256=EBhELuJ1He-ZYcAnR5YTImcIeDtWthDae5CHELBxPkM,5056
 langgraph_api/schema.py,sha256=6gabS4_1IeRWV5lyuDV-2i__8brXl89elAlmD5BmEII,8370
 langgraph_api/serde.py,sha256=3GvelKhySjlXaNqpg2GyUxU6-NEkvif7WlMF9if_EgU,6029
@@ -22,7 +22,7 @@ langgraph_api/server.py,sha256=uCAqPgCLJ6ckslLs0i_dacSR8mzuR0Y6PkkJYk0O3bE,7196
 langgraph_api/sse.py,sha256=SLdtZmTdh5D8fbWrQjuY9HYLd2dg8Rmi6ZMmFMVc2iE,4204
 langgraph_api/state.py,sha256=5RTOShiFVnkx-o6t99_x63CGwXw_8Eb-dSTpYirP8ro,4683
 langgraph_api/store.py,sha256=NIoNZojs6NbtG3VLBPQEFNttvp7XPkHAfjbQ3gY7aLY,4701
-langgraph_api/stream.py,sha256=iEApgVxJU9v58J5oKSuNe_c7ThpWf0bprgIDHUs8IzA,18397
+langgraph_api/stream.py,sha256=iRF4Hu6A1n-KvDR4ki1BeOvOnOxGoV2NV8tiBHUWZOs,18428
 langgraph_api/thread_ttl.py,sha256=7H3gFlWcUiODPoaEzcwB0LR61uvcuyjD0ew_4BztB7k,1902
 langgraph_api/traceblock.py,sha256=Qq5CUdefnMDaRDnyvBSWGBClEj-f3oO7NbH6fedxOSE,630
 langgraph_api/utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -32,7 +32,7 @@ langgraph_api/worker.py,sha256=HVGyGVEYcXG-iKVgoBdFgANGxPjSs57JRl5OB4ra4nw,15267
 langgraph_api/api/__init__.py,sha256=WHy6oNLWtH1K7AxmmsU9RD-Vm6WP-Ov16xS8Ey9YCmQ,6090
 langgraph_api/api/assistants.py,sha256=5gVvU58Y1-EftBhCHGbEaOi_7cqGMKWhOt_GVfBC0Gg,16836
 langgraph_api/api/mcp.py,sha256=qe10ZRMN3f-Hli-9TI8nbQyWvMeBb72YB1PZVbyqBQw,14418
-langgraph_api/api/meta.py,sha256=w88TK1Wu4xOhgCfs04LBfL4pZkWhUW6QRwwAWdFby5A,4245
+langgraph_api/api/meta.py,sha256=dFD9ZgykbKARLdVSaJD9vO3CShvEyBmGpkjE8tqii0c,4448
 langgraph_api/api/openapi.py,sha256=If-z1ckXt-Yu5bwQytK1LWyX_T7G46UtLfixgEP8hwc,11959
 langgraph_api/api/runs.py,sha256=AiohGTFLjWCb-oTXoNDvPMod4v6RS_ivlieoiqDmtQM,21812
 langgraph_api/api/store.py,sha256=xGcPFx4v-VxlK6HRU9uCjzCQ0v66cvc3o_PB5_g7n0Q,5550
@@ -97,8 +97,8 @@ langgraph_runtime/store.py,sha256=7mowndlsIroGHv3NpTSOZDJR0lCuaYMBoTnTrewjslw,11
 LICENSE,sha256=ZPwVR73Biwm3sK6vR54djCrhaRiM4cAD2zvOQZV8Xis,3859
 logging.json,sha256=3RNjSADZmDq38eHePMm1CbP6qZ71AmpBtLwCmKU9Zgo,379
 openapi.json,sha256=h1LbSeGqr2Oor6vO8d3m67XJ1lHhVYVyt2ULvyhf_Ks,160215
-langgraph_api-0.3.0.dist-info/METADATA,sha256=d8jDrigzqnFIyEDbLFhtK_xmHD45s1LIACSzVkRtiMU,3890
-langgraph_api-0.3.0.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-langgraph_api-0.3.0.dist-info/entry_points.txt,sha256=hGedv8n7cgi41PypMfinwS_HfCwA7xJIfS0jAp8htV8,78
-langgraph_api-0.3.0.dist-info/licenses/LICENSE,sha256=ZPwVR73Biwm3sK6vR54djCrhaRiM4cAD2zvOQZV8Xis,3859
-langgraph_api-0.3.0.dist-info/RECORD,,
+langgraph_api-0.3.4.dist-info/METADATA,sha256=L5VNuwfF9vgK-Pq3KK1AOz1E2_45tosXJlbigcHVd0Q,3890
+langgraph_api-0.3.4.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+langgraph_api-0.3.4.dist-info/entry_points.txt,sha256=hGedv8n7cgi41PypMfinwS_HfCwA7xJIfS0jAp8htV8,78
+langgraph_api-0.3.4.dist-info/licenses/LICENSE,sha256=ZPwVR73Biwm3sK6vR54djCrhaRiM4cAD2zvOQZV8Xis,3859
+langgraph_api-0.3.4.dist-info/RECORD,,