Port "connection timeout to action server" changes to 3.6.x - [ENG 689] (#12965)

varunshankar · tmbo · web-flow · commit 609287ec3cda · 2023-11-30T13:26:52.000+01:00
* Merge pull request #106 from RasaHQ/ENG-680-DEFAULT_KEEP_ALIVE_TIMEOUT Fix connection to action server - [ENG 680] --------- Co-authored-by: Tom Bocklisch <tom@rasa.com>
diff --git a/rasa/core/agent.py b/rasa/core/agent.py
@@ -112,53 +112,59 @@ async def _pull_model_and_fingerprint(
 
     logger.debug(f"Requesting model from server {model_server.url}...")
 
-    try:
-        params = model_server.combine_parameters()
-        async with model_server.session.request(
-            "GET",
-            model_server.url,
-            timeout=DEFAULT_REQUEST_TIMEOUT,
-            headers=headers,
-            params=params,
-        ) as resp:
-            if resp.status in [204, 304]:
-                logger.debug(
-                    "Model server returned {} status code, "
-                    "indicating that no new model is available. "
-                    "Current fingerprint: {}"
-                    "".format(resp.status, fingerprint)
-                )
-                return None
-            elif resp.status == 404:
-                logger.debug(
-                    "Model server could not find a model at the requested "
-                    "endpoint '{}'. It's possible that no model has been "
-                    "trained, or that the requested tag hasn't been "
-                    "assigned.".format(model_server.url)
-                )
-                return None
-            elif resp.status != 200:
-                logger.debug(
-                    "Tried to fetch model from server, but server response "
-                    "status code is {}. We'll retry later..."
-                    "".format(resp.status)
+    async with model_server.session() as session:
+        try:
+            params = model_server.combine_parameters()
+            async with session.request(
+                "GET",
+                model_server.url,
+                timeout=DEFAULT_REQUEST_TIMEOUT,
+                headers=headers,
+                params=params,
+            ) as resp:
+
+                if resp.status in [204, 304]:
+                    logger.debug(
+                        "Model server returned {} status code, "
+                        "indicating that no new model is available. "
+                        "Current fingerprint: {}"
+                        "".format(resp.status, fingerprint)
+                    )
+                    return None
+                elif resp.status == 404:
+                    logger.debug(
+                        "Model server could not find a model at the requested "
+                        "endpoint '{}'. It's possible that no model has been "
+                        "trained, or that the requested tag hasn't been "
+                        "assigned.".format(model_server.url)
+                    )
+                    return None
+                elif resp.status != 200:
+                    logger.debug(
+                        "Tried to fetch model from server, but server response "
+                        "status code is {}. We'll retry later..."
+                        "".format(resp.status)
+                    )
+                    return None
+
+                model_path = Path(model_directory) / resp.headers.get(
+                    "filename", "model.tar.gz"
                 )
-                return None
-            model_path = Path(model_directory) / resp.headers.get(
-                "filename", "model.tar.gz"
+                with open(model_path, "wb") as file:
+                    file.write(await resp.read())
+
+                logger.debug("Saved model to '{}'".format(os.path.abspath(model_path)))
+
+                # return the new fingerprint
+                return resp.headers.get("ETag")
+
+        except aiohttp.ClientError as e:
+            logger.debug(
+                "Tried to fetch model from server, but "
+                "couldn't reach server. We'll retry later... "
+                "Error: {}.".format(e)
             )
-            with open(model_path, "wb") as file:
-                file.write(await resp.read())
-            logger.debug("Saved model to '{}'".format(os.path.abspath(model_path)))
-            # return the new fingerprint
-            return resp.headers.get("ETag")
-    except aiohttp.ClientError as e:
-        logger.debug(
-            "Tried to fetch model from server, but "
-            "couldn't reach server. We'll retry later... "
-            "Error: {}.".format(e)
-        )
-        return None
+            return None
 
 
 async def _run_model_pulling_worker(model_server: EndpointConfig, agent: Agent) -> None:
diff --git a/rasa/core/constants.py b/rasa/core/constants.py
@@ -24,6 +24,8 @@
 
 DEFAULT_LOCK_LIFETIME = 60  # in seconds
 
+DEFAULT_KEEP_ALIVE_TIMEOUT = 120  # in seconds
+
 BEARER_TOKEN_PREFIX = "Bearer "
 
 # The lowest priority is intended to be used by machine learning policies.
diff --git a/rasa/core/run.py b/rasa/core/run.py
@@ -1,9 +1,19 @@
 import asyncio
 import logging
 import uuid
+import platform
 import os
 from functools import partial
-from typing import Any, List, Optional, TYPE_CHECKING, Text, Union, Dict
+from typing import (
+    Any,
+    Callable,
+    List,
+    Optional,
+    Text,
+    Tuple,
+    Union,
+    Dict,
+)
 
 import rasa.core.utils
 from rasa.plugin import plugin_manager
@@ -23,8 +33,6 @@
 from sanic import Sanic
 from asyncio import AbstractEventLoop
 
-if TYPE_CHECKING:
-    from aiohttp import ClientSession
 
 logger = logging.getLogger()  # get the root logger
 
@@ -80,6 +88,14 @@ def _create_app_without_api(cors: Optional[Union[Text, List[Text]]] = None) -> S
     return app
 
 
+def _is_apple_silicon_system() -> bool:
+    # check if the system is MacOS
+    if platform.system().lower() != "darwin":
+        return False
+    # check for arm architecture, indicating apple silicon
+    return platform.machine().startswith("arm") or os.uname().machine.startswith("arm")
+
+
 def configure_app(
     input_channels: Optional[List["InputChannel"]] = None,
     cors: Optional[Union[Text, List[Text], None]] = None,
@@ -99,6 +115,9 @@ def configure_app(
     syslog_port: Optional[int] = None,
     syslog_protocol: Optional[Text] = None,
     request_timeout: Optional[int] = None,
+    server_listeners: Optional[List[Tuple[Callable, Text]]] = None,
+    use_uvloop: Optional[bool] = True,
+    keep_alive_timeout: int = constants.DEFAULT_KEEP_ALIVE_TIMEOUT,
 ) -> Sanic:
     """Run the agent."""
     rasa.core.utils.configure_file_logging(
@@ -118,6 +137,14 @@ def configure_app(
     else:
         app = _create_app_without_api(cors)
 
+    app.config.KEEP_ALIVE_TIMEOUT = keep_alive_timeout
+    if _is_apple_silicon_system() or not use_uvloop:
+        app.config.USE_UVLOOP = False
+        # some library still sets the loop to uvloop, even if disabled for sanic
+        # using uvloop leads to breakingio errors, see
+        # https://rasahq.atlassian.net/browse/ENG-667
+        asyncio.set_event_loop_policy(None)
+
     if input_channels:
         channels.channel.register(input_channels, app, route=route)
     else:
@@ -150,6 +177,10 @@ async def run_cmdline_io(running_app: Sanic) -> None:
 
         app.add_task(run_cmdline_io)
 
+    if server_listeners:
+        for (listener, event) in server_listeners:
+            app.register_listener(listener, event)
+
     return app
 
 
@@ -179,6 +210,7 @@ def serve_application(
     syslog_port: Optional[int] = None,
     syslog_protocol: Optional[Text] = None,
     request_timeout: Optional[int] = None,
+    server_listeners: Optional[List[Tuple[Callable, Text]]] = None,
 ) -> None:
     """Run the API entrypoint."""
     if not channel and not credentials:
@@ -204,6 +236,7 @@ def serve_application(
         syslog_port=syslog_port,
         syslog_protocol=syslog_protocol,
         request_timeout=request_timeout,
+        server_listeners=server_listeners,
     )
 
     ssl_context = server.create_ssl_context(
@@ -217,7 +250,7 @@ def serve_application(
         partial(load_agent_on_start, model_path, endpoints, remote_storage),
         "before_server_start",
     )
-    app.register_listener(create_connection_pools, "after_server_start")
+
     app.register_listener(close_resources, "after_server_stop")
 
     number_of_workers = rasa.core.utils.number_of_sanic_workers(
@@ -279,44 +312,3 @@ async def close_resources(app: Sanic, _: AbstractEventLoop) -> None:
     event_broker = current_agent.tracker_store.event_broker
     if event_broker:
         await event_broker.close()
-
-    action_endpoint = current_agent.action_endpoint
-    if action_endpoint:
-        await action_endpoint.session.close()
-
-    model_server = current_agent.model_server
-    if model_server:
-        await model_server.session.close()
-
-
-async def create_connection_pools(app: Sanic, _: AbstractEventLoop) -> None:
-    """Create connection pools for the agent's action server and model server."""
-    current_agent = getattr(app.ctx, "agent", None)
-    if not current_agent:
-        logger.debug("No agent found after server start.")
-        return None
-
-    create_action_endpoint_connection_pool(current_agent)
-    create_model_server_connection_pool(current_agent)
-
-    return None
-
-
-def create_action_endpoint_connection_pool(agent: Agent) -> Optional["ClientSession"]:
-    """Create a connection pool for the action endpoint."""
-    action_endpoint = agent.action_endpoint
-    if not action_endpoint:
-        logger.debug("No action endpoint found after server start.")
-        return None
-
-    return action_endpoint.session
-
-
-def create_model_server_connection_pool(agent: Agent) -> Optional["ClientSession"]:
-    """Create a connection pool for the model server."""
-    model_server = agent.model_server
-    if not model_server:
-        logger.debug("No model server endpoint found after server start.")
-        return None
-
-    return model_server.session
diff --git a/rasa/utils/endpoints.py b/rasa/utils/endpoints.py
@@ -1,8 +1,6 @@
 import ssl
-from functools import cached_property
 
 import aiohttp
-import logging
 import os
 from aiohttp.client_exceptions import ContentTypeError
 from sanic.request import Request
@@ -11,10 +9,11 @@
 from rasa.shared.exceptions import FileNotFoundException
 import rasa.shared.utils.io
 import rasa.utils.io
+import structlog
 from rasa.core.constants import DEFAULT_REQUEST_TIMEOUT
 
 
-logger = logging.getLogger(__name__)
+structlogger = structlog.get_logger()
 
 
 def read_endpoint_config(
@@ -32,9 +31,13 @@ def read_endpoint_config(
 
         return EndpointConfig.from_dict(content[endpoint_type])
     except FileNotFoundError:
-        logger.error(
-            "Failed to read endpoint configuration "
-            "from {}. No such file.".format(os.path.abspath(filename))
+        structlogger.error(
+            "endpoint.read.failed_no_such_file",
+            filename=os.path.abspath(filename),
+            event_info=(
+                "Failed to read endpoint configuration file - "
+                "the file was not found."
+            ),
         )
         return None
 
@@ -56,9 +59,13 @@ def concat_url(base: Text, subpath: Optional[Text]) -> Text:
     """
     if not subpath:
         if base.endswith("/"):
-            logger.debug(
-                f"The URL '{base}' has a trailing slash. Please make sure the "
-                f"target server supports trailing slashes for this endpoint."
+            structlogger.debug(
+                "endpoint.concat_url.trailing_slash",
+                url=base,
+                event_info=(
+                    "The URL has a trailing slash. Please make sure the "
+                    "target server supports trailing slashes for this endpoint."
+                ),
             )
         return base
 
@@ -95,7 +102,6 @@ def __init__(
         self.cafile = cafile
         self.kwargs = kwargs
 
-    @cached_property
     def session(self) -> aiohttp.ClientSession:
         """Creates and returns a configured aiohttp client session."""
         # create authentication parameters
@@ -164,23 +170,26 @@ async def request(
                     f"'{os.path.abspath(self.cafile)}' does not exist."
                 ) from e
 
-        async with self.session.request(
-            method,
-            url,
-            headers=headers,
-            params=self.combine_parameters(kwargs),
-            compress=compress,
-            ssl=sslcontext,
-            **kwargs,
-        ) as response:
-            if response.status >= 400:
-                raise ClientResponseError(
-                    response.status, response.reason, await response.content.read()
-                )
-            try:
-                return await response.json()
-            except ContentTypeError:
-                return None
+        async with self.session() as session:
+            async with session.request(
+                method,
+                url,
+                headers=headers,
+                params=self.combine_parameters(kwargs),
+                compress=compress,
+                ssl=sslcontext,
+                **kwargs,
+            ) as response:
+                if response.status >= 400:
+                    raise ClientResponseError(
+                        response.status,
+                        response.reason,
+                        await response.content.read(),
+                    )
+                try:
+                    return await response.json()
+                except ContentTypeError:
+                    return None
 
     @classmethod
     def from_dict(cls, data: Dict[Text, Any]) -> "EndpointConfig":
@@ -263,7 +272,7 @@ def float_arg(
     try:
         return float(str(arg))
     except (ValueError, TypeError):
-        logger.warning(f"Failed to convert '{arg}' to float.")
+        structlogger.warning("endpoint.float_arg.convert_failed", arg=arg, key=key)
         return default
 
 
@@ -291,5 +300,6 @@ def int_arg(
     try:
         return int(str(arg))
     except (ValueError, TypeError):
-        logger.warning(f"Failed to convert '{arg}' to int.")
+
+        structlogger.warning("endpoint.int_arg.convert_failed", arg=arg, key=key)
         return default
diff --git a/tests/core/test_run.py b/tests/core/test_run.py
diff --git a/tests/utils/test_endpoints.py b/tests/utils/test_endpoints.py
