In each of the 3 resharding state machines, we reset the metrics object they store early (before the state machines themselves are destroyed). If a currentOp request comes in after we've reset the metrics object, but before the state machine itself is destroyed, we'll attempt to access a non-existent obj.
- is duplicated by
-
SERVER-88620 [sharding_auto_bootstrap] Investigate why $currentOp causes a segfault on specific resharding js tests
- Closed
- related to
-
SERVER-67370 Resharding metrics future callback can extend lifetime of ReshardingRecipient services
- Closed