From f182e2f5438097ed344ac16f4eecc77570420c0d Mon Sep 17 00:00:00 2001
From: roxanne-o <roxanne@groundlight.ai>
Date: Tue, 24 Feb 2026 23:52:52 +0000
Subject: [PATCH] Fix RemoteDisconnected errors from stale pooled connections

When the SDK reuses an HTTP connection that has been idle for ~60s, the
server-side infrastructure (ALB / nginx) may have already closed it.
urllib3 then raises RemoteDisconnected which was not retried or caught.

- Add a default urllib3 Retry policy (3 retries with backoff) so stale
  connections are transparently re-established at the transport layer.
- Catch MaxRetryError and ProtocolError in rest.py so exhausted retries
  surface as ApiException instead of raw urllib3 errors.
---
 generated/groundlight_openapi_client/rest.py |  6 ++++++
 src/groundlight/client.py                    | 20 +++++++++++++++++---
 2 files changed, 23 insertions(+), 3 deletions(-)

diff --git a/generated/groundlight_openapi_client/rest.py b/generated/groundlight_openapi_client/rest.py
index 16d8ca86..cd8d4b2c 100644
--- a/generated/groundlight_openapi_client/rest.py
+++ b/generated/groundlight_openapi_client/rest.py
@@ -218,6 +218,12 @@ def request(
         except urllib3.exceptions.SSLError as e:
             msg = "{0}\n{1}".format(type(e).__name__, str(e))
             raise ApiException(status=0, reason=msg)
+        except urllib3.exceptions.MaxRetryError as e:
+            msg = "{0}\n{1}".format(type(e).__name__, str(e))
+            raise ApiException(status=0, reason=msg)
+        except urllib3.exceptions.ProtocolError as e:
+            msg = "{0}\n{1}".format(type(e).__name__, str(e))
+            raise ApiException(status=0, reason=msg)
 
         if _preload_content:
             r = RESTResponse(r)
diff --git a/src/groundlight/client.py b/src/groundlight/client.py
index 669d3a7c..345f21e5 100644
--- a/src/groundlight/client.py
+++ b/src/groundlight/client.py
@@ -59,6 +59,19 @@
 # It used to take >8 min to timeout to a bad IP address
 DEFAULT_REQUEST_TIMEOUT = 10  # seconds
 
+# Default urllib3 transport-level retry policy.  Handles stale/dead pooled
+# connections (RemoteDisconnected) that occur when the server (ALB / nginx)
+# closes an idle keep-alive connection before the client reuses it.
+DEFAULT_HTTP_TRANSPORT_RETRY = Retry(
+    total=3,
+    connect=3,
+    read=3,
+    redirect=3,
+    backoff_factor=0.2,
+    allowed_methods=None,  # retry all HTTP methods including POST
+    raise_on_status=False,  # let the SDK's own status-code handling run
+)
+
 
 class GroundlightClientError(Exception):
     pass
@@ -156,9 +169,10 @@ def __init__(
         # Specify the endpoint
         self.endpoint = sanitize_endpoint_url(endpoint)
         self.configuration = Configuration(host=self.endpoint)
-        if http_transport_retries is not None:
-            # Once we upgrade openapitools to ^7.7.0, retries can be passed into the constructor of Configuration above.
-            self.configuration.retries = http_transport_retries
+        # Once we upgrade openapitools to ^7.7.0, retries can be passed into the constructor of Configuration above.
+        self.configuration.retries = (
+            http_transport_retries if http_transport_retries is not None else DEFAULT_HTTP_TRANSPORT_RETRY
+        )
 
         if not api_token:
             try: