From 73a2dc929358f013e121a023af1878bc225baeb8 Mon Sep 17 00:00:00 2001
From: mateo-berri <277851410+mateo-berri@users.noreply.github.com>
Date: Sat, 19 Sep 2026 19:57:49 -0700
Subject: [PATCH] fix(interactions): fail a cross-replica delete when its
 pre-delete fetch fails so the creating poll keeps the bill

---
 litellm/interactions/background_cost_polling.py   | 11 ++++++-----
 .../interactions/test_background_cost_polling.py  | 15 ++++++++-------
 2 files changed, 14 insertions(+), 12 deletions(-)

diff --git a/litellm/interactions/background_cost_polling.py b/litellm/interactions/background_cost_polling.py
index ce0774d6b8c..a28e66d52d8 100644
--- a/litellm/interactions/background_cost_polling.py
+++ b/litellm/interactions/background_cost_polling.py
@@ -593,14 +593,15 @@ async def maybe_settle_background_interaction_before_delete(
         api_base=api_base if isinstance(api_base, str) else None,
         store=settlement_store,
     )
-    response: Final = await _fetch_before_delete(context, fetch_interaction)
-    if response is None:
+    try:
+        response: Final = await fetch_interaction(context)
+    except Exception:
         verbose_logger.debug(
-            "Leaving background interaction %s to the poll that created it: this process could not fetch it with "
-            "the delete's credentials, so the delete is about to fail the same way",
+            "Failing the delete of background interaction %s: this process could not fetch it with the delete's "
+            "credentials, so the poll that created it keeps the bill",
             interaction_id,
         )
-        return None
+        raise
     return await _settle_before_delete(context, response)
 
 
diff --git a/tests/test_litellm/interactions/test_background_cost_polling.py b/tests/test_litellm/interactions/test_background_cost_polling.py
index df80acfd7e6..22c697e04bc 100644
--- a/tests/test_litellm/interactions/test_background_cost_polling.py
+++ b/tests/test_litellm/interactions/test_background_cost_polling.py
@@ -696,23 +696,24 @@ async def test_delete_on_another_replica_releases_the_create_reservation():
 
 
 @pytest.mark.asyncio
-async def test_delete_on_another_replica_leaves_an_unfetchable_create_to_its_own_poll():
+async def test_delete_on_another_replica_fails_when_it_cannot_fetch_and_leaves_the_bill_to_the_creating_poll():
     """
     The settling replica fetches with the delete's credentials, never the
     create's, so a fetch it cannot make (a key only the deployment carries)
-    says nothing about the interaction: the delete is about to fail the same
-    way, and the poll on the creating replica still owns the bill.
+    says nothing about the interaction. Deleting anyway would strand the bill
+    behind a deleted interaction, so the delete fails with the fetch's error
+    and the poll on the creating replica still owns the bill.
     """
     store = InMemoryBackgroundSettlementStore()
     logging_obj = _logging_obj(litellm_params={"metadata": _create_metadata()})
     await _create_on_a_replica_that_then_dies(logging_obj, store)
     fetch, _ = _fetch_sequence(RuntimeError("Google API key is required"))
 
-    outcome = await maybe_settle_background_interaction_before_delete(
-        interaction_id="interactions/bg-abc", delete_kwargs={}, fetch_interaction=fetch, store=store
-    )
+    with pytest.raises(RuntimeError, match="Google API key is required"):
+        await maybe_settle_background_interaction_before_delete(
+            interaction_id="interactions/bg-abc", delete_kwargs={}, fetch_interaction=fetch, store=store
+        )
 
-    assert outcome is None
     assert await store.is_claimed("interactions/bg-abc") is False
     poll_fetch, _ = _fetch_sequence(_response("completed", with_usage=True))
     await asyncio.wait_for(_register_poll(logging_obj, poll_fetch=poll_fetch, store=store), timeout=5)