-
Type:
Task
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Query Integration
-
0
-
None
-
3
-
TBD
-
None
-
None
-
None
-
None
-
None
-
None
-
0
We've seen a number of tasserts related to cursor destruction paths while creating the memory tracker. This is because when the cursor is destroyed before finishing the query (e.g. via killCursor or timeouts), the memory tracker can potentially try to reset metrics on an invalid opCtx.
We should write an FSM test that stresses this logic issuing a memory-tracked query and killing/timing out the cursor at various states during the query.
- is related to
-
SERVER-104793 Transfer ownership of memory tracker on ClusterCursor termination
-
- Closed
-