fix: Cap LimitedConcurrencyClient.max_concurrency at 10

CHANGELOG.md

@@ -15,6 +15,8 @@
 
 ### Fixes
   - ControlModels throw warning instead of error in case a not recommended model is selected.
+  - Cap `LimitedConcurrencyClient.max_concurrency` at 10 and set default to 10.
+
 ### Deprecations
 ...
 

docker-compose.yaml

@@ -37,4 +37,4 @@ services:
   trace_viewer:
     image: ghcr.io/aleph-alpha/trace-viewer-trace-viewer:main
     ports:
-     - 3000:3000
+     - 5173:3000

src/intelligence_layer/connectors/limited_concurrency_client.py

@@ -1,4 +1,5 @@
 import time
+import warnings
 from functools import lru_cache
 from os import getenv
 from threading import Semaphore
@@ -98,24 +99,32 @@ class LimitedConcurrencyClient:
     """An Aleph Alpha Client wrapper that limits the number of concurrent requests.
 
     This just delegates each call to the wrapped Aleph Alpha Client and ensures that
-    never more then a given number of concurrent calls are executed against the API.
+    never more than a given number of concurrent calls are executed against the API.
 
     Args:
         client: The wrapped `Client`.
         max_concurrency: the maximal number of requests that may run concurrently
-            against the API.
+            against the API. Defaults to 10, which is also the maximum.
         max_retry_time: the maximal time in seconds a complete is retried in case a `BusyError` is raised.
 
     """
 
     def __init__(
         self,
         client: AlephAlphaClientProtocol,
-        max_concurrency: int = 20,
+        max_concurrency: int = 10,
         max_retry_time: int = 24 * 60 * 60,  # one day in seconds
     ) -> None:
         self._client = client
-        self._concurrency_limit_semaphore = Semaphore(max_concurrency)
+
+        limit_for_max_concurrency = 10
+        capped_max_concurrency = min(limit_for_max_concurrency, max_concurrency)
+        if max_concurrency > capped_max_concurrency:
+            warnings.warn(
+                f"Selected a value greater than the maximum allowed number. max_concurrency will be reduced to {limit_for_max_concurrency}."
+            )
+        self._concurrency_limit_semaphore = Semaphore(capped_max_concurrency)
+
         self._max_retry_time = max_retry_time
 
     @classmethod

tests/core/test_text_highlight.py

@@ -309,4 +309,4 @@ def test_highlight_clamps_end_correctly(
     # then
     assert result.highlights[0].start == start
     assert result.highlights[0].end == end
-    assert abs(result.highlights[0].score - score) <= 0.01
+    assert abs(result.highlights[0].score - score) <= 0.02