The resharding donor's abort()/interrupt() don't call _promises.setError(), so external waiters on the donor's promise futures like RecoverRefreshThread on the critical section futures stay blocked until the run() execution chain reaches _finishReshardingOperation()/
_runMandatoryCleanup() and sets the error there. We should refactor the code to eagerly set errors in abort()/interrupt() to have same parity as the recipient service.