On slow runs (e.g. config_fuzzer_stress_concurrency_sharded_replication) replication lag can exceed the default 60 seconds writeConcern timeout used for many sharding code paths, including persisting the range deletion task. jstests/concurrency/fsm_workload_helpers/chunks.js should be resilient to failed moveChunks due to writeConcern timeouts when running in a slow build.