test(ilm): cover durable transition progress recovery (#5286)

Co-authored-by: heihutu <heihutu@gmail.com>

crates/ecstore/src/bucket/lifecycle/bucket_lifecycle_ops.rs

@@ -4605,11 +4605,11 @@ mod tests {
         lifecycle_rule_has_date_expiration, lifecycle_version_purge_state_from_completed_targets,
         manual_transition_duration_elapsed, manual_transition_has_more_after_limit, manual_transition_version_marker,
         mark_delete_opts_skip_decommissioned_on_remote_success, merge_stale_multipart_candidate,
-        persist_manual_transition_page_checkpoint, recover_manual_transition_job, recover_manual_transition_jobs,
-        recover_manual_transition_jobs_once, replication_state_for_delete, resolve_tier_free_version_recovery_enabled,
-        resolve_transition_queue_capacity, resolve_transition_queue_send_timeout, resolve_transition_worker_count,
-        resolve_transition_workers_absolute_max, run_tier_free_version_recovery_loop, select_restore_s3_location,
-        set_lifecycle_observability_observer, set_recovered_free_version_enqueue_observer,
+        persist_manual_transition_job_progress, persist_manual_transition_page_checkpoint, recover_manual_transition_job,
+        recover_manual_transition_jobs, recover_manual_transition_jobs_once, replication_state_for_delete,
+        resolve_tier_free_version_recovery_enabled, resolve_transition_queue_capacity, resolve_transition_queue_send_timeout,
+        resolve_transition_worker_count, resolve_transition_workers_absolute_max, run_tier_free_version_recovery_loop,
+        select_restore_s3_location, set_lifecycle_observability_observer, set_recovered_free_version_enqueue_observer,
         should_defer_date_expiry_for_recent_config_update, should_reuse_lifecycle_delete_replication_state,
         transitioned_cleanup_tuple, transitioned_object_delete_opts, wait_for_tier_free_version_recovery,
     };
@@ -8394,6 +8394,86 @@ mod tests {
         assert!(!after_terminal_cancel.cancel_requested);
     }
 
+    #[tokio::test]
+    #[serial]
+    async fn manual_transition_progress_checkpoint_survives_reload_and_recovery() {
+        let (_paths, ecstore) = setup_test_env().await;
+        let job_id = Uuid::new_v4();
+        let bucket = "manual-progress-checkpoint-bucket";
+        let continuation_token =
+            encode_manual_transition_continuation_token(Some("logs/page-end".to_string()), Some("null".to_string()))
+                .expect("resume token should encode");
+        let options = ManualTransitionRunOptions {
+            prefix: "logs/".to_string(),
+            ..Default::default()
+        };
+        let mut record = ManualTransitionJobRecord::new(job_id, bucket, &options, "owner-a");
+        record.lease_expires_at_unix_nanos = 0;
+        save_manual_transition_job_record(ecstore.clone(), &record)
+            .await
+            .expect("running job record should save");
+        save_manual_transition_scope_admission_if_absent(ecstore.clone(), &ManualTransitionScopeAdmission::from_job(&record))
+            .await
+            .expect("running scope admission should save");
+
+        let checkpointed = persist_manual_transition_job_progress(
+            ecstore.clone(),
+            job_id,
+            &ManualTransitionRunReport {
+                bucket: bucket.to_string(),
+                prefix: "logs/".to_string(),
+                scanned: 2,
+                eligible: 2,
+                enqueued: 2,
+                continuation_token: Some(continuation_token.clone()),
+                truncated_by_limit: true,
+                ..Default::default()
+            },
+            ManualTransitionQueueSnapshot::default(),
+        )
+        .await
+        .expect("running progress checkpoint should persist");
+
+        assert_eq!(checkpointed.report.continuation_token.as_deref(), Some(continuation_token.as_str()));
+        assert_eq!(checkpointed.report.enqueued, 2);
+
+        let reloaded = load_manual_transition_job_record(ecstore.clone(), job_id)
+            .await
+            .expect("checkpointed job should reload");
+        assert_eq!(reloaded.report.continuation_token.as_deref(), Some(continuation_token.as_str()));
+        assert_eq!(reloaded.report.scanned, 2);
+        assert_eq!(reloaded.report.enqueued, 2);
+
+        let mut expired = reloaded.clone();
+        expired.lease_expires_at_unix_nanos = 0;
+        save_manual_transition_job_record(ecstore.clone(), &expired)
+            .await
+            .expect("checkpointed job owner loss should persist");
+
+        let outcome = recover_manual_transition_job(ecstore.clone(), job_id, ManualTransitionQueueSnapshot::default())
+            .await
+            .expect("recovery should fail closed when checkpointed page has pending worker results");
+        assert_eq!(outcome, ManualTransitionJobRecoveryOutcome::Unknown);
+        let recovered = load_manual_transition_job_record(ecstore.clone(), job_id)
+            .await
+            .expect("unknown checkpointed job should reload");
+        assert_eq!(recovered.state, ManualTransitionJobState::Unknown);
+        assert!(
+            recovered
+                .error
+                .as_deref()
+                .is_some_and(|error| error.contains("page/task journal")),
+            "unknown checkpoint recovery must retain page/task journal context: {recovered:#?}"
+        );
+        assert!(
+            matches!(
+                load_manual_transition_scope_admission(ecstore, &record.scope_key).await,
+                Err(Error::ConfigNotFound)
+            ),
+            "unknown checkpoint recovery must release scope admission"
+        );
+    }
+
     #[tokio::test]
     #[serial]
     async fn manual_transition_job_lost_worker_results_mark_unknown_and_release_admission() {