perf: cache service info responses + JSON file load

bento_service_registry/bento_services_json.py

@@ -1,10 +1,11 @@
 import aiofiles
 import orjson
+from async_lru import alru_cache
 from fastapi import Depends
 
 from typing import Annotated
 
-from .config import ConfigDependency
+from .config import Config, ConfigDependency
 from .types import BentoService
 
 
@@ -22,7 +23,9 @@
 BentoServicesByKind = dict[str, BentoService]
 
 
-async def get_bento_services_by_compose_id(config: ConfigDependency) -> BentoServicesByComposeID:
+# cache bento_services.json contents for the lifetime of the service:
+@alru_cache()
+async def _get_bento_services_by_compose_id(config: Config) -> BentoServicesByComposeID:
     async with aiofiles.open(config.bento_services, "rb") as fh:
         bento_services_data: BentoServicesByComposeID = orjson.loads(await fh.read())
 
@@ -43,6 +46,10 @@ async def get_bento_services_by_compose_id(config: ConfigDependency) -> BentoSer
     }
 
 
+async def get_bento_services_by_compose_id(config: ConfigDependency) -> BentoServicesByComposeID:
+    return await _get_bento_services_by_compose_id(config)
+
+
 BentoServicesByComposeIDDependency = Annotated[BentoServicesByComposeID, Depends(get_bento_services_by_compose_id)]
 
 

bento_service_registry/http_session.py

@@ -12,7 +12,7 @@
 
 
 async def get_http_session(config: ConfigDependency):
-    connector = aiohttp.TCPConnector(verify_ssl=config.bento_validate_ssl)
+    connector = aiohttp.TCPConnector(verify_ssl=config.bento_validate_ssl, force_close=True)
     session = aiohttp.ClientSession(
         connector=connector,
         timeout=aiohttp.ClientTimeout(total=config.contact_timeout),

bento_service_registry/service_info.py

@@ -1,10 +1,12 @@
+from async_lru import alru_cache
 from bento_lib.service_info.helpers import build_service_info_from_pydantic_config
 from bento_lib.service_info.types import GA4GHServiceInfo
 from bento_service_registry import __version__
 from fastapi import Depends
+from logging import Logger
 from typing import Annotated
 
-from .config import ConfigDependency
+from .config import Config, ConfigDependency
 from .constants import BENTO_SERVICE_KIND, SERVICE_TYPE
 from .logger import LoggerDependency
 
@@ -15,7 +17,8 @@
 ]
 
 
-async def get_service_info(config: ConfigDependency, logger: LoggerDependency) -> GA4GHServiceInfo:
+@alru_cache()
+async def _get_service_info(config: Config, logger: Logger) -> GA4GHServiceInfo:
     return await build_service_info_from_pydantic_config(
         config,
         logger,
@@ -25,4 +28,8 @@ async def get_service_info(config: ConfigDependency, logger: LoggerDependency) -
     )
 
 
+async def get_service_info(config: ConfigDependency, logger: LoggerDependency) -> GA4GHServiceInfo:
+    return await _get_service_info(config, logger)
+
+
 ServiceInfoDependency = Annotated[GA4GHServiceInfo, Depends(get_service_info)]

bento_service_registry/services.py

@@ -28,10 +28,14 @@
 ]
 
 
+CACHE_TTL = 30
+
+
 class ServiceManager:
     def __init__(self, logger: logging.Logger):
         self._co: Awaitable[list[dict | None]] | None = None
         self._logger = logger
+        self._cache: dict[str, tuple[datetime, GA4GHServiceInfo]] = {}
 
     async def get_service(
         self,
@@ -51,6 +55,15 @@ async def get_service(
         service_info_url: str = urljoin(f"{s_url}/", "service-info")
 
         dt = datetime.now()
+
+        if service_info_url in self._cache:
+            entry_dt, entry = self._cache[service_info_url]
+            if (entry_age := (dt - entry_dt).total_seconds()) > CACHE_TTL:
+                del self._cache[service_info_url]
+            else:
+                self._logger.debug(f"Found {service_info_url} in cache (age={entry_age:.1f}s)")
+                return entry
+
         self._logger.info(f"Contacting {service_info_url}{' with bearer token' if authz_header else ''}")
 
         service_resp: dict | None = None
@@ -70,7 +83,9 @@ async def get_service(
 
                 try:
                     service_resp = {**(await r.json()), "url": s_url}
-                    self._logger.debug(f"{service_info_url}: Took {(datetime.now() - dt).total_seconds():.1f}s")
+                    res_dt = datetime.now()
+                    self._cache[service_info_url] = (res_dt, service_resp)
+                    self._logger.debug(f"{service_info_url}: Took {(res_dt - dt).total_seconds():.1f}s")
                 except (JSONDecodeError, aiohttp.ContentTypeError, TypeError) as e:
                     # JSONDecodeError can happen if the JSON is invalid
                     # ContentTypeError can happen if the Content-Type is not application/json
@@ -95,10 +110,10 @@ async def get_services(
         service_info: GA4GHServiceInfo,
     ) -> tuple[dict, ...]:
         if not self._co:
-            self._co = asyncio.gather(*[
+            self._co = asyncio.gather(*(
                 self.get_service(authz_header, http_session, service_info, s)
                 for s in bento_services_by_kind.values()
-            ])
+            ))
 
         service_list: list[dict | None] = await self._co
         self._co = None