Files
rustfs/crates/scanner/src
Zhengchao An 5d820df79c chore(obs): rename ReplicationStats to ReplicationMetricsSnapshot (#6344)
* chore(obs): ReplicationStats -> ReplicationMetricsSnapshot, BucketReplicationStats -> BucketReplicationMetricsSnapshot

Rename in-obs-crate ReplicationStats and BucketReplicationStats to
ReplicationMetricsSnapshot and BucketReplicationMetricsSnapshot respectively.
No serde impact (these types are Prometheus metric collectors, not serialized).
No external consumers found outside the obs crate.

* cleanup: remove #[serial] annotations from e2e_test, scanner, lifecycle, and object-capacity crates

Remove no-op #[serial] attributes (nextest ignores serial_test) and the
serial_test dependency from four crates. All tests already use temp_env
for env-var isolation, making #[serial] purely redundant.

Crates cleaned:
- e2e_test (37 annotations, 9 imports, removed serial_test dep)
- rustfs-scanner (115 annotations across 7 files, removed serial_test dep)
- rustfs-lifecycle (46 annotations, removed serial_test dep)
- rustfs-object-capacity (38 annotations, removed serial_test dep)

Also converted scanner/tests/lifecycle_integration_test.rs
with_forced_immediate_enqueue_timeout helper from unsafe raw
env::set_var/remove_var to temp_env::async_with_vars for proper
isolation, and added async_closure feature to scanner's temp-env dep.

* fix(lifecycle): restore #[serial] on 2 tests that read env vars without temp_env

eval_inner_expires_latest_object_after_days_due and
eval_inner_does_not_panic_on_many_equal_due_events call eval_inner()
which reads ENV_ILM_PROCESS_TIME via std::env::var(). Without #[serial]
they race with other tests that set these vars via temp_env.

* style: cargo fmt
2026-08-22 00:30:30 +00:00
..