Merge remote-tracking branch 'origin/main' into cxymds/perf-decommission-entry-budget

# Conflicts:
#	crates/ecstore/src/core/pools.rs

crates/common/src/metrics.rs

@@ -901,10 +901,6 @@ pub struct Metrics {
     scanner_cycle_max_duration_millis: AtomicU64,
     scanner_cycle_max_objects: AtomicU64,
     scanner_cycle_max_directories: AtomicU64,
-    scanner_cycle_timeout_total: AtomicU64,
-    scanner_cycle_recovery_required_total: AtomicU64,
-    scanner_cycle_last_progress_age_seconds: AtomicU64,
-    scanner_leader_lease_without_progress: AtomicBool,
     scanner_bitrot_cycle_enabled: AtomicBool,
     scanner_bitrot_cycle_millis: AtomicU64,
     scanner_checkpoint: Mutex<Option<ScannerCheckpointReport>>,
@@ -1374,14 +1370,6 @@ pub struct ScannerMetricsReport {
     #[serde(default)]
     pub cycle_max_directories: u64,
     #[serde(default)]
-    pub cycle_timeout_total: u64,
-    #[serde(default)]
-    pub cycle_recovery_required_total: u64,
-    #[serde(default)]
-    pub cycle_last_progress_age: u64,
-    #[serde(default)]
-    pub leader_lease_without_progress: bool,
-    #[serde(default)]
     pub bitrot_cycle_enabled: bool,
     #[serde(default)]
     pub bitrot_cycle_seconds: f64,
@@ -1442,9 +1430,6 @@ const OTEL_SCANNER_BUCKETS_SCANNED: &str = "rustfs_scanner_buckets_scanned_total
 const OTEL_SCANNER_CYCLES: &str = "rustfs_scanner_cycles_total";
 const OTEL_SCANNER_CYCLE_DURATION_SECONDS: &str = "rustfs_scanner_cycle_duration_seconds";
 const OTEL_SCANNER_BUCKET_DRIVE_DURATION_SECONDS: &str = "rustfs_scanner_bucket_drive_duration_seconds";
-const OTEL_SCANNER_CYCLE_TIMEOUT_TOTAL: &str = "rustfs_scanner_cycle_timeout_total";
-const OTEL_SCANNER_CYCLE_LAST_PROGRESS_AGE: &str = "rustfs_scanner_cycle_last_progress_age";
-const OTEL_SCANNER_LEADER_LEASE_WITHOUT_PROGRESS: &str = "rustfs_scanner_leader_lease_without_progress";
 
 fn scan_cycle_result_label(result: u8) -> &'static str {
     match result {
@@ -1928,10 +1913,6 @@ impl Metrics {
             scanner_cycle_max_duration_millis: AtomicU64::new(0),
             scanner_cycle_max_objects: AtomicU64::new(0),
             scanner_cycle_max_directories: AtomicU64::new(0),
-            scanner_cycle_timeout_total: AtomicU64::new(0),
-            scanner_cycle_recovery_required_total: AtomicU64::new(0),
-            scanner_cycle_last_progress_age_seconds: AtomicU64::new(0),
-            scanner_leader_lease_without_progress: AtomicBool::new(false),
             scanner_bitrot_cycle_enabled: AtomicBool::new(false),
             scanner_bitrot_cycle_millis: AtomicU64::new(0),
             scanner_checkpoint: Mutex::new(None),
@@ -2431,29 +2412,12 @@ impl Metrics {
             .store(cycle_max_objects.unwrap_or_default(), Ordering::Relaxed);
         self.scanner_cycle_max_directories
             .store(cycle_max_directories.unwrap_or_default(), Ordering::Relaxed);
-        self.scanner_leader_lease_without_progress.store(false, Ordering::Relaxed);
-        self.scanner_cycle_last_progress_age_seconds.store(0, Ordering::Relaxed);
-        metrics::gauge!(OTEL_SCANNER_LEADER_LEASE_WITHOUT_PROGRESS).set(0.0);
-        metrics::gauge!(OTEL_SCANNER_CYCLE_LAST_PROGRESS_AGE).set(0.0);
         self.scanner_bitrot_cycle_enabled
             .store(bitrot_cycle.is_some(), Ordering::Relaxed);
         self.scanner_bitrot_cycle_millis
             .store(bitrot_cycle.map(duration_millis_saturated).unwrap_or_default(), Ordering::Relaxed);
     }
 
-    pub fn record_scanner_cycle_timeout(&self, recovery_required: bool, progress_age: Duration) {
-        self.scanner_cycle_timeout_total.fetch_add(1, Ordering::Relaxed);
-        if recovery_required {
-            self.scanner_cycle_recovery_required_total.fetch_add(1, Ordering::Relaxed);
-        }
-        self.scanner_cycle_last_progress_age_seconds
-            .store(progress_age.as_secs(), Ordering::Relaxed);
-        self.scanner_leader_lease_without_progress.store(true, Ordering::Relaxed);
-        metrics::counter!(OTEL_SCANNER_CYCLE_TIMEOUT_TOTAL).increment(1);
-        metrics::gauge!(OTEL_SCANNER_CYCLE_LAST_PROGRESS_AGE).set(progress_age.as_secs_f64());
-        metrics::gauge!(OTEL_SCANNER_LEADER_LEASE_WITHOUT_PROGRESS).set(1.0);
-    }
-
     pub fn record_scanner_set_scan_state(&self, concurrency_limit: Option<usize>, queued: Option<usize>, active: Option<usize>) {
         if let Some(concurrency_limit) = concurrency_limit {
             self.scanner_set_scan_concurrency_limit
@@ -3301,10 +3265,6 @@ impl Metrics {
         m.cycle_max_duration_seconds = self.scanner_cycle_max_duration_millis.load(Ordering::Relaxed) as f64 / 1000.0;
         m.cycle_max_objects = self.scanner_cycle_max_objects.load(Ordering::Relaxed);
         m.cycle_max_directories = self.scanner_cycle_max_directories.load(Ordering::Relaxed);
-        m.cycle_timeout_total = self.scanner_cycle_timeout_total.load(Ordering::Relaxed);
-        m.cycle_recovery_required_total = self.scanner_cycle_recovery_required_total.load(Ordering::Relaxed);
-        m.cycle_last_progress_age = self.scanner_cycle_last_progress_age_seconds.load(Ordering::Relaxed);
-        m.leader_lease_without_progress = self.scanner_leader_lease_without_progress.load(Ordering::Relaxed);
         m.bitrot_cycle_enabled = self.scanner_bitrot_cycle_enabled.load(Ordering::Relaxed);
         m.bitrot_cycle_seconds = self.scanner_bitrot_cycle_millis.load(Ordering::Relaxed) as f64 / 1000.0;
         m.scan_checkpoint = match self.scanner_checkpoint.lock() {
@@ -4966,20 +4926,4 @@ mod tests {
         assert!(!report.bitrot_cycle_enabled);
         assert_eq!(report.bitrot_cycle_seconds, 0.0);
     }
-
-    #[tokio::test]
-    async fn scanner_cycle_timeout_metrics_reset_for_a_new_cycle() {
-        let metrics = Metrics::new();
-        metrics.record_scanner_cycle_timeout(true, Duration::from_secs(17));
-        let timed_out = metrics.report().await;
-        assert_eq!(timed_out.cycle_timeout_total, 1);
-        assert_eq!(timed_out.cycle_last_progress_age, 17);
-        assert!(timed_out.leader_lease_without_progress);
-
-        metrics.record_scanner_cycle_config(Duration::from_secs(60), None, Some(Duration::from_secs(1)), None, None);
-        let current = metrics.report().await;
-        assert_eq!(current.cycle_timeout_total, 1);
-        assert_eq!(current.cycle_last_progress_age, 0);
-        assert!(!current.leader_lease_without_progress);
-    }
 }

crates/config/README.md

@@ -84,12 +84,6 @@ Current guidance:
 - `RUSTFS_SCANNER_CYCLE_MAX_OBJECTS` (canonical)
 - `RUSTFS_SCANNER_CYCLE_MAX_DIRECTORIES` (canonical)
 
-Scanner cycle budget controls:
-
-- When `RUSTFS_SCANNER_CYCLE_MAX_DURATION_SECS` is unset, the finite default is 1800 seconds (30 minutes), matching the scanner benchmark guidance.
-- An explicit `0` preserves the compatibility behavior of an unbounded runtime budget. Object and directory budgets likewise remain unbounded when explicitly set to `0`.
-- A timed-out cycle cancels cooperative scanner work, then fences its leader epoch before releasing the lease. An uncooperative I/O operation is dropped after the bounded shutdown window; its cursor is not claimed to be durable and the scanner reports `recovery-required` when the worker cannot stop cooperatively, the cycle state was not confirmed durable, or epoch fencing cannot be persisted.
-
 ## Mmap read environment aliases
 
 - `RUSTFS_OBJECT_MMAP_READ_ENABLE` (canonical)

crates/config/src/constants/scanner.rs

@@ -143,12 +143,9 @@ pub const ENV_SCANNER_MAX_WAIT_SECS: &str = "RUSTFS_SCANNER_MAX_WAIT_SECS";
 /// Default scanner speed preset.
 pub const DEFAULT_SCANNER_SPEED: &str = "default";
 
-/// Default scanner cycle runtime budget when no override is configured.
-///
-/// An explicit `0` remains the compatibility escape hatch for an unbounded
-/// cycle. Keeping the unset default finite prevents a stalled scanner I/O
-/// operation from holding the leader lease forever.
-pub const DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS: u64 = 30 * 60;
+/// Default scanner cycle runtime budget.
+/// `0` keeps the existing unbounded per-cycle behavior.
+pub const DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS: u64 = 0;
 
 /// Default scanner per-cycle object budget.
 /// `0` keeps the existing unbounded per-cycle behavior.

crates/ecstore/src/runtime/instance.rs

@@ -160,6 +160,10 @@ pub struct InstanceContext {
     /// workers (scanner/heal/tier/lifecycle) without touching another instance.
     /// Replaces the process-global cancel-token static.
     background_cancel_token: OnceLock<CancellationToken>,
+    /// Serializes decommission data-movement operations with cancellation and
+    /// a subsequent restart. Readers are held across one object side effect;
+    /// the transition path takes the writer after cancelling the routine.
+    decommission_operation_gate: Arc<RwLock<()>>,
     /// Resolves object-encryption material at the application boundary.
     object_encryption_resolver: OnceLock<Arc<dyn ObjectEncryptionResolver>>,
     tier_delete_journal_recovery_stores: std::sync::Mutex<HashSet<Uuid>>,
@@ -200,6 +204,7 @@ impl InstanceContext {
             local_disk_set_drives: Arc::new(RwLock::new(Vec::new())),
             bucket_metadata_sys: std::sync::Mutex::new(None),
             background_cancel_token: OnceLock::new(),
+            decommission_operation_gate: Arc::new(RwLock::new(())),
             object_encryption_resolver: OnceLock::new(),
             tier_delete_journal_recovery_stores: std::sync::Mutex::new(HashSet::new()),
             transition_transaction_recovery_stores: std::sync::Mutex::new(HashSet::new()),
@@ -218,6 +223,10 @@ impl InstanceContext {
         self.lock_manager.clone()
     }
 
+    pub(crate) fn decommission_operation_gate(&self) -> Arc<RwLock<()>> {
+        Arc::clone(&self.decommission_operation_gate)
+    }
+
     /// Install the application-owned object-encryption resolver once.
     pub fn set_object_encryption_resolver(
         &self,

crates/ecstore/src/services/metrics_realtime.rs

@@ -256,10 +256,6 @@ fn to_madmin_scanner_metrics(metrics: rustfs_common::metrics::ScannerMetricsRepo
         cycle_max_duration_seconds: metrics.cycle_max_duration_seconds,
         cycle_max_objects: metrics.cycle_max_objects,
         cycle_max_directories: metrics.cycle_max_directories,
-        cycle_timeout_total: metrics.cycle_timeout_total,
-        cycle_recovery_required_total: metrics.cycle_recovery_required_total,
-        cycle_last_progress_age: metrics.cycle_last_progress_age,
-        leader_lease_without_progress: metrics.leader_lease_without_progress,
         bitrot_cycle_enabled: metrics.bitrot_cycle_enabled,
         bitrot_cycle_seconds: metrics.bitrot_cycle_seconds,
         scan_checkpoint: metrics.scan_checkpoint.map(|checkpoint| MadminScannerCheckpointReport {
@@ -615,10 +611,6 @@ mod test {
             current_started: chrono_to_jiff_timestamp(current_started),
             last_cycle_partial_source: "usage".to_string(),
             last_cycle_partial_source_code: 1,
-            cycle_timeout_total: 3,
-            cycle_recovery_required_total: 2,
-            cycle_last_progress_age: 17,
-            leader_lease_without_progress: true,
             partial_cycles_by_source: vec![rustfs_common::metrics::ScannerSourceCycleSnapshot {
                 source: "usage".to_string(),
                 cycles: 2,
@@ -630,10 +622,6 @@ mod test {
         assert_eq!(scanner.current_started, chrono_to_jiff_timestamp(current_started));
         assert_eq!(scanner.last_cycle_partial_source, "usage");
         assert_eq!(scanner.last_cycle_partial_source_code, 1);
-        assert_eq!(scanner.cycle_timeout_total, 3);
-        assert_eq!(scanner.cycle_recovery_required_total, 2);
-        assert_eq!(scanner.cycle_last_progress_age, 17);
-        assert!(scanner.leader_lease_without_progress);
         let usage = scanner
             .partial_cycles_by_source
             .iter()

crates/madmin/src/metrics.rs

@@ -689,14 +689,6 @@ pub struct ScannerMetrics {
     pub cycle_max_objects: u64,
     #[serde(rename = "cycle_max_directories", default)]
     pub cycle_max_directories: u64,
-    #[serde(rename = "cycle_timeout_total", default)]
-    pub cycle_timeout_total: u64,
-    #[serde(rename = "cycle_recovery_required_total", default)]
-    pub cycle_recovery_required_total: u64,
-    #[serde(rename = "cycle_last_progress_age", default)]
-    pub cycle_last_progress_age: u64,
-    #[serde(rename = "leader_lease_without_progress", default)]
-    pub leader_lease_without_progress: bool,
     #[serde(rename = "bitrot_cycle_enabled", default)]
     pub bitrot_cycle_enabled: bool,
     #[serde(rename = "bitrot_cycle_seconds", default)]
@@ -772,8 +764,6 @@ impl ScannerMetrics {
             self.cycle_max_duration_seconds = other.cycle_max_duration_seconds;
             self.cycle_max_objects = other.cycle_max_objects;
             self.cycle_max_directories = other.cycle_max_directories;
-            self.cycle_last_progress_age = other.cycle_last_progress_age;
-            self.leader_lease_without_progress = other.leader_lease_without_progress;
             self.bitrot_cycle_enabled = other.bitrot_cycle_enabled;
             self.bitrot_cycle_seconds = other.bitrot_cycle_seconds;
         }
@@ -867,12 +857,6 @@ impl ScannerMetrics {
             .saturating_add(other.last_cycle_replication_checks);
         self.last_cycle_usage_saves = self.last_cycle_usage_saves.saturating_add(other.last_cycle_usage_saves);
         self.failed_cycles = self.failed_cycles.saturating_add(other.failed_cycles);
-        self.cycle_timeout_total = self.cycle_timeout_total.saturating_add(other.cycle_timeout_total);
-        self.cycle_recovery_required_total = self
-            .cycle_recovery_required_total
-            .saturating_add(other.cycle_recovery_required_total);
-        self.cycle_last_progress_age = self.cycle_last_progress_age.max(other.cycle_last_progress_age);
-        self.leader_lease_without_progress |= other.leader_lease_without_progress;
         self.superseded_cycles = self.superseded_cycles.saturating_add(other.superseded_cycles);
         self.partial_cycles_unknown = self.partial_cycles_unknown.saturating_add(other.partial_cycles_unknown);
         self.partial_cycles_runtime = self.partial_cycles_runtime.saturating_add(other.partial_cycles_runtime);

crates/scanner/src/data_usage_define.rs

@@ -125,34 +125,6 @@ pub(crate) async fn read_config_with_revision<S: ScannerObjectIO>(
     }
 }
 
-/// Read only the object revision without materializing its body.
-pub(crate) async fn read_config_revision<S: ScannerObjectIO>(store: Arc<S>, path: &str) -> StorageResult<DataUsageCacheRevision> {
-    match store
-        .get_object_reader(
-            RUSTFS_META_BUCKET,
-            path,
-            None,
-            HeaderMap::new(),
-            &ObjectOptions {
-                no_lock: true,
-                ..Default::default()
-            },
-        )
-        .await
-    {
-        Ok(reader) => reader
-            .object_info
-            .etag
-            .filter(|etag| !etag.is_empty())
-            .map(DataUsageCacheRevision::Etag)
-            .ok_or_else(|| StorageError::other(format!("scanner config object {path} has no ETag"))),
-        Err(Error::FileNotFound | Error::VolumeNotFound | Error::ObjectNotFound(_, _) | Error::BucketNotFound(_)) => {
-            Ok(DataUsageCacheRevision::Missing)
-        }
-        Err(err) => Err(err),
-    }
-}
-
 #[derive(Clone, Debug)]
 pub(crate) struct DataUsageCacheRevisions {
     main: DataUsageCacheRevision,
@@ -174,11 +146,6 @@ pub static LEGACY_DATA_USAGE_OBJ_NAME_PATH: LazyLock<String> =
 pub static DATA_USAGE_BLOOM_NAME_PATH: LazyLock<String> =
     LazyLock::new(|| format!("{BUCKET_META_PREFIX}{SLASH_SEPARATOR}{DATA_USAGE_BLOOM_NAME}"));
 
-/// Durable companion object for a cycle-state object which cannot be decoded.
-/// The primary object is deliberately never replaced or deleted by recovery.
-pub static DATA_USAGE_BLOOM_RECOVERY_PATH: LazyLock<String> =
-    LazyLock::new(|| format!("{}.recovery-required.json", DATA_USAGE_BLOOM_NAME_PATH.as_str()));
-
 pub static BACKGROUND_HEAL_INFO_PATH: LazyLock<String> =
     LazyLock::new(|| format!("{BUCKET_META_PREFIX}{SLASH_SEPARATOR}.background-heal.json"));
 

crates/scanner/src/data_usage_define/persistence.rs

@@ -74,7 +74,7 @@ impl DataUsageCache {
         let loaded = Self::load_cache(store.clone(), name).await?;
         let backup = match loaded.backup_revision {
             Some(revision) => Some(revision),
-            None => match read_config_revision(store, &backup_path).await {
+            None => match Self::revision_for_path(store, &backup_path).await {
                 Ok(revision) => Some(revision),
                 Err(err) => {
                     counter!(METRIC_CACHE_BACKUP_REVISION_FAILURE_TOTAL).increment(1);
@@ -336,6 +336,33 @@ impl DataUsageCache {
         }
     }
 
+    async fn revision_for_path<S: ScannerObjectIO>(store: Arc<S>, path: &str) -> StorageResult<DataUsageCacheRevision> {
+        match store
+            .get_object_reader(
+                RUSTFS_META_BUCKET,
+                path,
+                None,
+                HeaderMap::new(),
+                &ObjectOptions {
+                    no_lock: true,
+                    ..Default::default()
+                },
+            )
+            .await
+        {
+            Ok(reader) => reader
+                .object_info
+                .etag
+                .filter(|etag| !etag.is_empty())
+                .map(DataUsageCacheRevision::Etag)
+                .ok_or_else(|| StorageError::other(format!("scanner cache object {path} has no ETag"))),
+            Err(Error::FileNotFound | Error::VolumeNotFound | Error::ObjectNotFound(_, _) | Error::BucketNotFound(_)) => {
+                Ok(DataUsageCacheRevision::Missing)
+            }
+            Err(err) => Err(err),
+        }
+    }
+
     pub(super) fn cache_save_timeout() -> Duration {
         crate::runtime_config::scanner_cache_save_timeout()
     }

crates/scanner/src/lib.rs

@@ -75,10 +75,7 @@ pub use remote_scanner::{
 };
 pub use runtime_config::{apply_scanner_runtime_config, scanner_runtime_config_status, validate_scanner_runtime_config};
 pub use rustfs_common::last_minute;
-pub use scanner::{
-    ScannerCycleRecoveryMarker, ScannerCycleRecoveryStatus, ScannerCycleScheduleStatus, init_data_scanner,
-    reset_scanner_cycle_recovery, scanner_cycle_recovery_status, scanner_cycle_schedule_status, scanner_topology_digest,
-};
+pub use scanner::{ScannerCycleScheduleStatus, init_data_scanner, scanner_cycle_schedule_status, scanner_topology_digest};
 pub use scanner_io::{
     ScannerDirtyUsageAckError, ScannerDirtyUsageState, acknowledge_dirty_usage_generation, clear_dirty_usage_bucket,
     record_dirty_usage_bucket, record_scanner_maintenance_change, scanner_activity_epoch, scanner_dirty_usage_state,

crates/scanner/src/runtime_config.rs

@@ -125,10 +125,7 @@ impl Default for ScannerRuntimeConfig {
             cycle_interval_source: ScannerRuntimeConfigSource::Default,
             bitrot_cycle: Some(Duration::from_secs(DEFAULT_HEAL_BITROT_CYCLE_SECS)),
             bitrot_cycle_source: ScannerRuntimeConfigSource::Default,
-            cycle_budget: ScannerCycleBudgetConfig {
-                max_duration: Some(Duration::from_secs(DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS)),
-                ..Default::default()
-            },
+            cycle_budget: ScannerCycleBudgetConfig::default(),
             cycle_max_duration_source: ScannerRuntimeConfigSource::Default,
             cycle_max_objects_source: ScannerRuntimeConfigSource::Default,
             cycle_max_directories_source: ScannerRuntimeConfigSource::Default,
@@ -377,10 +374,7 @@ fn validate_persisted_scanner_runtime_config(config: &ServerConfig) -> Result<()
     }
     validate_optional_config_u64(scanner_kvs, SCANNER_START_DELAY, "")?;
     validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE, "")?;
-    if let Some(value) = config_value(scanner_kvs, SCANNER_CYCLE_MAX_DURATION, DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS) {
-        let secs = parse_config_u64(SCANNER_CYCLE_MAX_DURATION, value)?;
-        cycle_duration_from_secs(SCANNER_CYCLE_MAX_DURATION, secs)?;
-    }
+    validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE_MAX_DURATION, DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS)?;
     validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE_MAX_OBJECTS, DEFAULT_SCANNER_CYCLE_MAX_OBJECTS)?;
     validate_optional_config_u64(scanner_kvs, SCANNER_CYCLE_MAX_DIRECTORIES, DEFAULT_SCANNER_CYCLE_MAX_DIRECTORIES)?;
     if let Some(value) = config_value(heal_kvs, HEAL_BITROT_CYCLE, DEFAULT_HEAL_BITROT_CYCLE_SECS) {
@@ -442,46 +436,19 @@ fn lookup_max_wait(
     Ok((speed.max_sleep(), speed_source))
 }
 
-fn lookup_cycle_duration(kvs: Option<&KVS>) -> Result<(Option<Duration>, ScannerRuntimeConfigSource), ScannerRuntimeConfigError> {
-    match rustfs_utils::get_env_parse_outcome::<u64>(ENV_SCANNER_CYCLE_MAX_DURATION_SECS) {
-        rustfs_utils::EnvParseOutcome::Parsed(secs) => {
-            return cycle_duration_from_secs(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, secs)
-                .map(|duration| (duration, ScannerRuntimeConfigSource::Env));
-        }
-        rustfs_utils::EnvParseOutcome::Invalid => {
-            // Do not include the raw environment value in the typed error:
-            // deployments occasionally put sensitive material in inherited
-            // environment snapshots. The key still identifies the control.
-            return Err(invalid_value(
-                ENV_SCANNER_CYCLE_MAX_DURATION_SECS,
-                "<invalid>",
-                "expected unsigned integer seconds",
-            ));
-        }
-        rustfs_utils::EnvParseOutcome::Absent => {}
+fn lookup_optional_seconds(
+    kvs: Option<&KVS>,
+    key: &'static str,
+    env_key: &'static str,
+    default: u64,
+) -> Result<(Option<Duration>, ScannerRuntimeConfigSource), ScannerRuntimeConfigError> {
+    if let Some(secs) = rustfs_utils::get_env_opt_u64(env_key) {
+        return Ok((Some(Duration::from_secs(secs)), ScannerRuntimeConfigSource::Env));
     }
-
-    if let Some(value) = config_value(kvs, SCANNER_CYCLE_MAX_DURATION, DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS) {
-        let secs = parse_config_u64(SCANNER_CYCLE_MAX_DURATION, value)?;
-        return cycle_duration_from_secs(SCANNER_CYCLE_MAX_DURATION, secs)
-            .map(|duration| (duration, ScannerRuntimeConfigSource::Config));
+    if let Some(value) = config_value(kvs, key, default) {
+        return parse_config_u64(key, value).map(|secs| (Some(Duration::from_secs(secs)), ScannerRuntimeConfigSource::Config));
     }
-
-    Ok((
-        Some(Duration::from_secs(DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS)),
-        ScannerRuntimeConfigSource::Default,
-    ))
-}
-
-fn cycle_duration_from_secs(key: &'static str, secs: u64) -> Result<Option<Duration>, ScannerRuntimeConfigError> {
-    if secs == 0 {
-        return Ok(None);
-    }
-    let duration = Duration::from_secs(secs);
-    if std::time::Instant::now().checked_add(duration).is_none() {
-        return Err(invalid_value(key, "<overflow>", "duration exceeds the timer range"));
-    }
-    Ok(Some(duration))
+    Ok((None, ScannerRuntimeConfigSource::Default))
 }
 
 fn lookup_start_delay(kvs: Option<&KVS>) -> Result<(Option<Duration>, ScannerRuntimeConfigSource), ScannerRuntimeConfigError> {
@@ -586,7 +553,12 @@ pub(crate) fn lookup_scanner_runtime_config(
         (speed.cycle_interval(), speed_source)
     };
 
-    let (cycle_max_duration, cycle_max_duration_source) = lookup_cycle_duration(scanner_kvs)?;
+    let (cycle_max_duration, cycle_max_duration_source) = lookup_optional_seconds(
+        scanner_kvs,
+        SCANNER_CYCLE_MAX_DURATION,
+        ENV_SCANNER_CYCLE_MAX_DURATION_SECS,
+        DEFAULT_SCANNER_CYCLE_MAX_DURATION_SECS,
+    )?;
     let (cycle_max_objects, cycle_max_objects_source) = lookup_count_budget(
         scanner_kvs,
         SCANNER_CYCLE_MAX_OBJECTS,
@@ -891,10 +863,10 @@ mod tests {
     use rustfs_config::server_config::{Config as ServerConfig, KVS};
     use rustfs_config::{
         DEFAULT_DELIMITER, DEFAULT_HEAL_BITROT_CYCLE_SECS, ENV_SCANNER_BITROT_CYCLE_SECS, ENV_SCANNER_CACHE_SAVE_TIMEOUT_SECS,
-        ENV_SCANNER_CYCLE, ENV_SCANNER_CYCLE_MAX_DURATION_SECS, ENV_SCANNER_CYCLE_MAX_OBJECTS, ENV_SCANNER_DELAY,
-        ENV_SCANNER_MAX_WAIT_SECS, ENV_SCANNER_SPEED, HEAL_BITROT_CYCLE, HEAL_SUB_SYS, SCANNER_BITROT_CYCLE,
-        SCANNER_CACHE_SAVE_TIMEOUT, SCANNER_CYCLE, SCANNER_CYCLE_MAX_DIRECTORIES, SCANNER_CYCLE_MAX_DURATION,
-        SCANNER_CYCLE_MAX_OBJECTS, SCANNER_DELAY, SCANNER_IDLE_MODE, SCANNER_SPEED, SCANNER_SUB_SYS, ScannerSpeed,
+        ENV_SCANNER_CYCLE, ENV_SCANNER_CYCLE_MAX_OBJECTS, ENV_SCANNER_DELAY, ENV_SCANNER_MAX_WAIT_SECS, ENV_SCANNER_SPEED,
+        HEAL_BITROT_CYCLE, HEAL_SUB_SYS, SCANNER_BITROT_CYCLE, SCANNER_CACHE_SAVE_TIMEOUT, SCANNER_CYCLE,
+        SCANNER_CYCLE_MAX_DIRECTORIES, SCANNER_CYCLE_MAX_DURATION, SCANNER_CYCLE_MAX_OBJECTS, SCANNER_DELAY, SCANNER_IDLE_MODE,
+        SCANNER_SPEED, SCANNER_SUB_SYS, ScannerSpeed,
     };
     use std::collections::HashMap;
     use std::time::Duration;
@@ -969,50 +941,6 @@ mod tests {
         });
     }
 
-    #[test]
-    fn scanner_unset_budget_uses_safe_default_but_explicit_zero_is_unbounded() {
-        let config = server_config_with_scanner(&[]);
-        with_var_unset(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, || {
-            let resolved = lookup_scanner_runtime_config(Some(&config)).expect("scanner runtime config");
-            assert_eq!(resolved.cycle_budget.max_duration, Some(Duration::from_secs(1800)));
-            assert_eq!(resolved.cycle_max_duration_source, ScannerRuntimeConfigSource::Default);
-        });
-
-        let config = server_config_with_scanner(&[(SCANNER_CYCLE_MAX_DURATION, "0")]);
-        with_var_unset(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, || {
-            let resolved = lookup_scanner_runtime_config(Some(&config)).expect("scanner runtime config");
-            assert_eq!(resolved.cycle_budget.max_duration, None);
-            assert_eq!(resolved.cycle_max_duration_source, ScannerRuntimeConfigSource::Config);
-        });
-    }
-
-    #[test]
-    fn cycle_budget_invalid_or_overflow_config_is_rejected() {
-        with_var(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, Some("invalid"), || {
-            let error = lookup_scanner_runtime_config(None).expect_err("invalid duration env must be rejected");
-            assert!(error.to_string().contains(ENV_SCANNER_CYCLE_MAX_DURATION_SECS));
-            assert!(error.to_string().contains("<invalid>"));
-            assert!(!error.to_string().contains(": invalid ("));
-        });
-        with_var(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, Some("18446744073709551616"), || {
-            assert!(lookup_scanner_runtime_config(None).is_err());
-        });
-        with_var(ENV_SCANNER_CYCLE_MAX_DURATION_SECS, Some("18446744073709551615"), || {
-            assert!(lookup_scanner_runtime_config(None).is_err());
-        });
-        let config = server_config_with_scanner(&[(SCANNER_CYCLE_MAX_DURATION, "not-a-duration")]);
-        assert!(lookup_scanner_runtime_config(Some(&config)).is_err());
-    }
-
-    #[test]
-    fn scanner_runtime_config_validation_rejects_overflow_persisted_duration() {
-        let config = server_config_with_scanner(&[(SCANNER_CYCLE_MAX_DURATION, "18446744073709551615")]);
-
-        let error = validate_scanner_runtime_config(&config)
-            .expect_err("persisted duration that exceeds the timer range must be rejected");
-        assert!(error.to_string().contains(SCANNER_CYCLE_MAX_DURATION));
-    }
-
     #[test]
     fn scanner_runtime_config_normalizes_persisted_default_speed() {
         let config = server_config_with_scanner(&[(SCANNER_SPEED, "default")]);

crates/scanner/src/scanner.rs

@@ -20,7 +20,7 @@ use std::sync::{Arc, LazyLock, RwLock};
 
 use crate::data_usage_define::{
     BACKGROUND_HEAL_INFO_PATH, DATA_USAGE_BLOOM_NAME_PATH, DATA_USAGE_OBJ_NAME_PATH, DATA_USAGE_OBSERVED_OBJ_NAME_PATH,
-    DataUsageCache, DataUsageCacheRevision, LEGACY_DATA_USAGE_OBJ_NAME_PATH, read_config_revision, read_config_with_revision,
+    DataUsageCache, DataUsageCacheRevision, LEGACY_DATA_USAGE_OBJ_NAME_PATH, read_config_with_revision,
 };
 use crate::runtime_config::{
     ScannerRuntimeConfig, ScannerRuntimeConfigSource, refresh_scanner_runtime_config_from_global, scanner_bitrot_cycle,
@@ -52,10 +52,11 @@ use rustfs_config::{
 };
 use rustfs_config::{ENV_SCANNER_CYCLE, ENV_SCANNER_SPEED, ENV_SCANNER_START_DELAY_SECS};
 use rustfs_data_usage::observed_data_usage_is_newer;
-use rustfs_lock::NamespaceLockGuard;
 use serde::{Deserialize, Serialize};
 use sha2::{Digest as _, Sha256};
-use tokio::sync::{Notify, mpsc};
+#[cfg(test)]
+use tokio::sync::Notify;
+use tokio::sync::mpsc;
 use tokio::time::{Duration, Instant};
 use tokio_util::sync::CancellationToken;
 use tokio_util::task::AbortOnDropHandle;
@@ -103,13 +104,6 @@ const CLEAN_IDLE_BACKOFF_FACTOR: u32 = 2;
 /// unavailable peer cannot drive a tight retry loop.
 const SCANNER_RETRY_BASE_INTERVAL: Duration = Duration::from_secs(5);
 const SCANNER_RETRY_MAX_INTERVAL: Duration = Duration::from_secs(30 * 60);
-/// A transient backend outage remains self-healing after the short retry
-/// budget is exhausted, but the probe is intentionally sparse until storage
-/// recovers or an operator reset wakes the scanner.
-const SCANNER_CYCLE_RECOVERY_PAUSED_INTERVAL: Duration = Duration::from_secs(5 * 60);
-/// Permanent recovery states still get a sparse status probe so a reset that
-/// races the wait registration cannot leave the scanner asleep forever.
-const SCANNER_CYCLE_RECOVERY_BLOCKED_PROBE_INTERVAL: Duration = Duration::from_secs(5 * 60);
 const SCANNER_LEADER_LOCK_POLL_INTERVAL: Duration = Duration::from_secs(1);
 #[cfg(not(test))]
 const SCANNER_LOCK_LOSS_SHUTDOWN_TIMEOUT: Duration = Duration::from_secs(30);
@@ -131,12 +125,6 @@ type ScannerCycleStatePersistTestHook = (u64, Arc<Notify>);
 static SCANNER_CYCLE_STATE_PERSIST_TEST_HOOK: LazyLock<StdMutex<Option<ScannerCycleStatePersistTestHook>>> =
     LazyLock::new(|| StdMutex::new(None));
 
-static SCANNER_CYCLE_RECOVERY_WAKE: LazyLock<Notify> = LazyLock::new(Notify::new);
-
-pub(super) fn notify_scanner_cycle_recovery_wake() {
-    SCANNER_CYCLE_RECOVERY_WAKE.notify_one();
-}
-
 #[cfg(test)]
 struct ScannerCycleStatePersistTestHookGuard;
 
@@ -588,21 +576,19 @@ pub async fn init_data_scanner(ctx: CancellationToken, storeapi: Arc<ECStore>) {
             tokio::time::sleep(sleep_time).await;
         }
 
-        let mut transient_backoff = ScannerRetryBackoff::default();
-        let mut recovery_retry_count = 0_u32;
         loop {
             if ctx_clone.is_cancelled() {
                 break;
             }
 
-            let run_result = run_data_scanner_with_maintenance_state(
+            if let Err(e) = run_data_scanner_with_maintenance_state(
                 ctx_clone.clone(),
                 storeapi_clone.clone(),
                 startup_features,
                 startup_maintenance_generation,
             )
-            .await;
-            if let Err(e) = &run_result {
+            .await
+            {
                 error!(
                     target: "rustfs::scanner",
                     event = EVENT_SCANNER_CYCLE_STATE,
@@ -613,52 +599,11 @@ pub async fn init_data_scanner(ctx: CancellationToken, storeapi: Arc<ECStore>) {
                     "Scanner runtime iteration failed"
                 );
             }
-            let recovery_status = scanner_cycle_recovery_status();
-            if recovery_status.retryable {
-                recovery_retry_count = recovery_retry_count.saturating_add(1);
-                let _ = record_scanner_cycle_recovery_retry(recovery_retry_count);
-            } else {
-                recovery_retry_count = 0;
-            }
-
-            let recovery_status = scanner_cycle_recovery_status();
-            if recovery_status.state == "paused" {
-                transient_backoff.record_retryable_cycle(false);
-                tokio::select! {
-                    _ = ctx_clone.cancelled() => break,
-                    _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
-                    _ = tokio::time::sleep(SCANNER_CYCLE_RECOVERY_PAUSED_INTERVAL) => {},
-                }
-                recovery_retry_count = 0;
-                continue;
-            }
-            if !recovery_status.retryable
-                && matches!(recovery_status.state.as_str(), "blocked" | "recovery-required" | "cleanup-pending")
-            {
-                transient_backoff.record_retryable_cycle(false);
-                tokio::select! {
-                    _ = ctx_clone.cancelled() => break,
-                    _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
-                    _ = tokio::time::sleep(SCANNER_CYCLE_RECOVERY_BLOCKED_PROBE_INTERVAL) => {},
-                }
-                continue;
-            }
-
-            let retry_delay = if recovery_status.retryable || run_result.is_err() {
-                transient_backoff.record_retryable_cycle(true);
-                transient_backoff
-                    .retry_interval(scanner_cycle_interval())
-                    .unwrap_or(SCANNER_RETRY_BASE_INTERVAL)
-            } else {
-                transient_backoff.record_retryable_cycle(false);
-                randomized_cycle_delay()
-            };
             // Backoff before retrying after lock contention or scanner-level failures.
             // Keep this cancellation-aware so shutdown is not delayed by backoff sleep.
             tokio::select! {
                 _ = ctx_clone.cancelled() => break,
-                _ = SCANNER_CYCLE_RECOVERY_WAKE.notified() => {},
-                _ = tokio::time::sleep(retry_delay) => {}
+                _ = tokio::time::sleep(randomized_cycle_delay()) => {}
             }
         }
     });
@@ -1038,116 +983,20 @@ fn data_usage_persist_timeout() -> Duration {
     DataUsageCache::persistence_timeout()
 }
 
-#[cfg(not(test))]
-const SCANNER_CYCLE_EPOCH_FENCE_TIMEOUT: Duration = Duration::from_secs(30);
-#[cfg(test)]
-const SCANNER_CYCLE_EPOCH_FENCE_TIMEOUT: Duration = Duration::from_millis(50);
-
-async fn fence_scanner_epoch_after_cycle_timeout<Store, LockLost>(
-    ctx: &CancellationToken,
-    storeapi: Arc<Store>,
-    cycle_info: &mut CurrentCycle,
-    cycle_revision: &mut DataUsageCacheRevision,
-    leader_epoch: &mut u64,
-    lock_lost: LockLost,
-) -> bool
-where
-    Store: ScannerObjectIO,
-    LockLost: Future<Output = ()>,
-{
-    let fence_ctx = ctx.child_token();
-    let claim = claim_scanner_leadership(&fence_ctx, storeapi, cycle_info, cycle_revision, leader_epoch);
-    tokio::pin!(claim);
-    tokio::pin!(lock_lost);
-    tokio::select! {
-        biased;
-        _ = &mut lock_lost => {
-            fence_ctx.cancel();
-            false
-        }
-        result = tokio::time::timeout(SCANNER_CYCLE_EPOCH_FENCE_TIMEOUT, &mut claim) => {
-            result.unwrap_or(false) && !fence_ctx.is_cancelled()
-        }
-    }
-}
-
-struct ScannerCycleDeadlineState<'a> {
-    cycle_info: &'a mut CurrentCycle,
-    cycle_revision: &'a mut DataUsageCacheRevision,
-    leader_epoch: &'a mut u64,
-    cycle_budget: &'a ScannerCycleBudget,
-}
-
-fn cycle_timeout_requires_recovery(worker_stopped: bool, cycle_state_persisted: bool, generation_fenced: bool) -> bool {
-    !worker_stopped || !cycle_state_persisted || !generation_fenced
-}
-
-async fn handle_scanner_cycle_deadline<Store>(
-    ctx: &CancellationToken,
-    storeapi: Arc<Store>,
-    state: ScannerCycleDeadlineState<'_>,
-    worker_stopped: bool,
-    guard: &mut NamespaceLockGuard,
-) where
-    Store: ScannerObjectIO,
-{
-    let fenced = fence_scanner_epoch_after_cycle_timeout(
-        ctx,
-        storeapi,
-        state.cycle_info,
-        state.cycle_revision,
-        state.leader_epoch,
-        guard.lock_lost_notified(),
-    )
-    .await;
-    let cycle_state_persisted = state.cycle_budget.cycle_state_persisted();
-    let recovery_required = cycle_timeout_requires_recovery(worker_stopped, cycle_state_persisted, fenced);
-    warn!(
-        target: "rustfs::scanner",
-        event = EVENT_SCANNER_CYCLE_STATE,
-        component = LOG_COMPONENT_SCANNER,
-        subsystem = LOG_SUBSYSTEM_RUNTIME,
-        state = "cycle_timeout",
-        worker_stopped,
-        cycle_state_persisted,
-        generation_fenced = fenced,
-        recovery_required,
-        "Scanner cycle deadline expired; durable cursor/generation fencing completed when possible"
-    );
-    global_metrics().record_scanner_cycle_timeout(recovery_required, state.cycle_budget.progress_age());
-    // Stop renewing before releasing the lease. A new leader can then claim the
-    // higher persisted generation instead of inheriting the expired worker.
-    guard.release();
-    global_metrics().set_cycle(None).await;
-}
-
 async fn mark_scan_cycle_idle(cycle_info: &mut CurrentCycle, cycle_metrics_guard: &mut ScannerCycleMetricsGuard) {
     cycle_info.current = 0;
     global_metrics().clear_current_scan_mode();
     cycle_metrics_guard.finish(cycle_info.clone()).await;
 }
 
-#[cfg(test)]
+#[instrument(skip_all)]
+#[hotpath::measure]
 async fn run_data_scanner_cycle(
     ctx: &CancellationToken,
     storeapi: &Arc<ECStore>,
     cycle_info: &mut CurrentCycle,
     cycle_revision: &mut DataUsageCacheRevision,
     leader_epoch: u64,
-) -> ScannerCycleOutcome {
-    let cycle_budget = ScannerCycleBudget::new(ctx, scanner_cycle_budget_config());
-    run_data_scanner_cycle_with_budget(ctx, storeapi, cycle_info, cycle_revision, leader_epoch, cycle_budget).await
-}
-
-#[instrument(skip_all)]
-#[hotpath::measure]
-async fn run_data_scanner_cycle_with_budget(
-    ctx: &CancellationToken,
-    storeapi: &Arc<ECStore>,
-    cycle_info: &mut CurrentCycle,
-    cycle_revision: &mut DataUsageCacheRevision,
-    leader_epoch: u64,
-    cycle_budget: Arc<ScannerCycleBudget>,
 ) -> ScannerCycleOutcome {
     let _activity_guard = ScannerActivityGuard::new();
     if let Err(err) = refresh_scanner_runtime_config_from_global() {
@@ -1163,11 +1012,7 @@ async fn run_data_scanner_cycle_with_budget(
     }
     let configured_cycle_interval = scanner_cycle_interval();
     let configured_bitrot_cycle = scanner_bitrot_cycle();
-    let cycle_budget_config = ScannerCycleBudgetConfig {
-        max_duration: cycle_budget.max_duration(),
-        max_objects: cycle_budget.max_objects(),
-        max_directories: cycle_budget.max_directories(),
-    };
+    let cycle_budget_config = scanner_cycle_budget_config();
     let usage_persist_timeout = data_usage_persist_timeout();
     global_metrics().record_scanner_cycle_config(
         configured_cycle_interval,
@@ -1238,6 +1083,7 @@ async fn run_data_scanner_cycle_with_budget(
     let (sender, receiver) = mpsc::channel::<DataUsageInfo>(1);
 
     let done_cycle = Metrics::time(Metric::ScanCycle);
+    let cycle_budget = ScannerCycleBudget::new(ctx, cycle_budget_config);
     let scan_result = storeapi
         .clone()
         .nsscanner_with_status(
@@ -1377,7 +1223,7 @@ async fn run_data_scanner_cycle_with_budget(
                 "Scanner cycle is recovering to a newer durable cache generation"
             );
             emit_scan_cycle_partial_with_source(cycle_start.elapsed(), ScanCyclePartialReason::Unknown, None);
-            let persisted = persist_required_scanner_cycle_floor(
+            return if persist_required_scanner_cycle_floor(
                 ctx,
                 storeapi.clone(),
                 cycle_info,
@@ -1386,9 +1232,8 @@ async fn run_data_scanner_cycle_with_budget(
                 required_cycle,
                 &mut cycle_metrics_guard,
             )
-            .await;
-            return if persisted {
-                cycle_budget.mark_cycle_state_persisted();
+            .await
+            {
                 ScannerCycleOutcome::Partial
             } else {
                 ScannerCycleOutcome::Failed
@@ -1446,7 +1291,7 @@ async fn run_data_scanner_cycle_with_budget(
             scan_cycle_partial_reason(budget_reason),
             scan_cycle_partial_source(budget_reason),
         );
-        let persisted = finalize_partial_scan_cycle(
+        return if finalize_partial_scan_cycle(
             ctx,
             storeapi.clone(),
             cycle_info,
@@ -1454,9 +1299,8 @@ async fn run_data_scanner_cycle_with_budget(
             leader_epoch,
             &mut cycle_metrics_guard,
         )
-        .await;
-        return if persisted {
-            cycle_budget.mark_cycle_state_persisted();
+        .await
+        {
             ScannerCycleOutcome::Partial
         } else {
             ScannerCycleOutcome::Failed
@@ -1531,7 +1375,7 @@ async fn run_data_scanner_cycle_with_budget(
                 );
             }
             emit_scan_cycle_partial_with_source(cycle_start.elapsed(), ScanCyclePartialReason::Unknown, None);
-            let persisted = finalize_partial_scan_cycle(
+            return if finalize_partial_scan_cycle(
                 ctx,
                 storeapi.clone(),
                 cycle_info,
@@ -1539,9 +1383,8 @@ async fn run_data_scanner_cycle_with_budget(
                 leader_epoch,
                 &mut cycle_metrics_guard,
             )
-            .await;
-            return if persisted {
-                cycle_budget.mark_cycle_state_persisted();
+            .await
+            {
                 ScannerCycleOutcome::Partial
             } else {
                 ScannerCycleOutcome::Failed
@@ -1582,7 +1425,6 @@ async fn run_data_scanner_cycle_with_budget(
             )
             .await
             {
-                cycle_budget.mark_cycle_state_persisted();
                 emit_scan_cycle_superseded(cycle_start.elapsed());
                 return ScannerCycleOutcome::Superseded;
             }
@@ -1615,7 +1457,6 @@ async fn run_data_scanner_cycle_with_budget(
         emit_scan_cycle_complete(false, cycle_start.elapsed());
         return ScannerCycleOutcome::Failed;
     }
-    cycle_budget.mark_cycle_state_persisted();
 
     done_cycle();
     emit_scan_cycle_complete(true, cycle_start.elapsed());
@@ -1680,7 +1521,7 @@ async fn run_data_scanner_with_maintenance_state(
 ) -> Result<(), ScannerError> {
     reset_scanner_cycle_schedule();
     // Acquire leader lock (write lock) to ensure only one scanner runs
-    let mut guard = match storeapi.new_ns_lock(RUSTFS_META_BUCKET, "leader.lock").await {
+    let guard = match storeapi.new_ns_lock(RUSTFS_META_BUCKET, "leader.lock").await {
         Ok(ns_lock) => match ns_lock.get_write_lock_quiet(get_lock_acquire_timeout()).await {
             Ok(guard) => {
                 record_scanner_leader_lock_state("acquired");
@@ -1765,22 +1606,40 @@ async fn run_data_scanner_with_maintenance_state(
         observe_scanner_activity(&storeapi, distributed, &mut scanner_activity_seen).await;
     }
 
-    let (mut cycle_info, mut leader_epoch, mut cycle_revision) =
-        match load_scanner_cycle_state_for_startup(storeapi.clone()).await {
-            ScannerCycleStateStartup::Ready {
-                cycle,
-                leader_epoch,
-                revision,
-            } => (cycle, leader_epoch, revision),
-            ScannerCycleStateStartup::Blocked => {
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleStateStartup::Transient(err) => {
-                global_metrics().set_cycle(None).await;
-                return Err(err);
-            }
-        };
+    let (buf, mut cycle_revision) = match read_config_with_revision(storeapi.clone(), DATA_USAGE_BLOOM_NAME_PATH.as_str()).await {
+        Ok((buf, revision)) => (buf.unwrap_or_default(), revision),
+        Err(err) => {
+            error!(
+                target: "rustfs::scanner",
+                event = EVENT_SCANNER_PERSIST_STATE,
+                component = LOG_COMPONENT_SCANNER,
+                subsystem = LOG_SUBSYSTEM_RUNTIME,
+                path = %&*DATA_USAGE_BLOOM_NAME_PATH,
+                state = "revision_load_failed",
+                error = %err,
+                "Scanner cycle state revision load failed"
+            );
+            global_metrics().set_cycle(None).await;
+            return Ok(());
+        }
+    };
+    let (mut cycle_info, mut leader_epoch) = match decode_scanner_cycle_state_for_startup(&buf) {
+        Ok(state) => state,
+        Err(err) => {
+            error!(
+                target: "rustfs::scanner",
+                event = EVENT_SCANNER_PERSIST_STATE,
+                component = LOG_COMPONENT_SCANNER,
+                subsystem = LOG_SUBSYSTEM_RUNTIME,
+                path = %&*DATA_USAGE_BLOOM_NAME_PATH,
+                state = "cycle_decode_failed",
+                error = %err,
+                "Scanner stopped because persisted cycle state is invalid"
+            );
+            global_metrics().set_cycle(None).await;
+            return Ok(());
+        }
+    };
     let usage_floor = match persisted_usage_floor(storeapi.clone()).await {
         Ok(floor) => floor,
         Err(err) => {
@@ -1845,49 +1704,13 @@ async fn run_data_scanner_with_maintenance_state(
             return Ok(());
         }
         let cycle_ctx = ctx.child_token();
-        let cycle_budget = ScannerCycleBudget::new_with_runtime_progress_tracking(&cycle_ctx, scanner_cycle_budget_config());
-        let initial_outcome = match await_scanner_cycle_with_budget_fence(
+        let initial_outcome = await_scanner_cycle_with_lock_fence(
             &cycle_ctx,
-            &cycle_budget,
-            run_data_scanner_cycle_with_budget(
-                &cycle_ctx,
-                &storeapi,
-                &mut cycle_info,
-                &mut cycle_revision,
-                leader_epoch,
-                cycle_budget.clone(),
-            ),
+            run_data_scanner_cycle(&cycle_ctx, &storeapi, &mut cycle_info, &mut cycle_revision, leader_epoch),
             guard.lock_lost_notified(),
         )
         .await
-        {
-            ScannerCycleWaitOutcome::Completed(outcome) => outcome,
-            ScannerCycleWaitOutcome::LockLost => {
-                record_scanner_leader_lock_lost("Scanner leader lock lost during the initial cycle").await;
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Cancelled => {
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Deadline { worker_stopped } => {
-                handle_scanner_cycle_deadline(
-                    &ctx,
-                    storeapi.clone(),
-                    ScannerCycleDeadlineState {
-                        cycle_info: &mut cycle_info,
-                        cycle_revision: &mut cycle_revision,
-                        leader_epoch: &mut leader_epoch,
-                        cycle_budget: &cycle_budget,
-                    },
-                    worker_stopped,
-                    &mut guard,
-                )
-                .await;
-                return Ok(());
-            }
-        };
+        .unwrap_or(ScannerCycleOutcome::Failed);
         superseded_backoff.record_retryable_cycle(initial_outcome == ScannerCycleOutcome::Superseded);
         deferred_backoff.record_retryable_cycle(matches!(initial_outcome, ScannerCycleOutcome::Deferred(_)));
         dirty_usage_generation_seen = dirty_generation_before_cycle;
@@ -2093,49 +1916,13 @@ async fn run_data_scanner_with_maintenance_state(
         }
         let dirty_generation_before_cycle = dirty_usage_generation();
         let cycle_ctx = ctx.child_token();
-        let cycle_budget = ScannerCycleBudget::new_with_runtime_progress_tracking(&cycle_ctx, scanner_cycle_budget_config());
-        let outcome = match await_scanner_cycle_with_budget_fence(
+        let outcome = await_scanner_cycle_with_lock_fence(
             &cycle_ctx,
-            &cycle_budget,
-            run_data_scanner_cycle_with_budget(
-                &cycle_ctx,
-                &storeapi,
-                &mut cycle_info,
-                &mut cycle_revision,
-                leader_epoch,
-                cycle_budget.clone(),
-            ),
+            run_data_scanner_cycle(&cycle_ctx, &storeapi, &mut cycle_info, &mut cycle_revision, leader_epoch),
             guard.lock_lost_notified(),
         )
         .await
-        {
-            ScannerCycleWaitOutcome::Completed(outcome) => outcome,
-            ScannerCycleWaitOutcome::LockLost => {
-                record_scanner_leader_lock_lost("Scanner leader lock lost during a scanner cycle").await;
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Cancelled => {
-                global_metrics().set_cycle(None).await;
-                return Ok(());
-            }
-            ScannerCycleWaitOutcome::Deadline { worker_stopped } => {
-                handle_scanner_cycle_deadline(
-                    &ctx,
-                    storeapi.clone(),
-                    ScannerCycleDeadlineState {
-                        cycle_info: &mut cycle_info,
-                        cycle_revision: &mut cycle_revision,
-                        leader_epoch: &mut leader_epoch,
-                        cycle_budget: &cycle_budget,
-                    },
-                    worker_stopped,
-                    &mut guard,
-                )
-                .await;
-                return Ok(());
-            }
-        };
+        .unwrap_or(ScannerCycleOutcome::Failed);
         superseded_backoff.record_retryable_cycle(outcome == ScannerCycleOutcome::Superseded);
         deferred_backoff.record_retryable_cycle(matches!(outcome, ScannerCycleOutcome::Deferred(_)));
         dirty_usage_generation_seen = dirty_generation_before_cycle;
@@ -2432,12 +2219,7 @@ pub(crate) use activity::{
 pub(crate) use activity::{ScannerCycleOutcome, scanner_cycle_outcome_with_pending_maintenance};
 #[cfg(test)]
 pub(crate) use cycle_state::encode_scanner_cycle_fence_for_test;
-pub use cycle_state::{
-    ScannerCycleRecoveryMarker, ScannerCycleRecoveryStatus, reset_scanner_cycle_recovery, scanner_cycle_recovery_status,
-};
-pub(crate) use cycle_state::{
-    current_scanner_leader_epoch, decode_persisted_scanner_cycle_fence, load_scanner_cycle_state_for_startup,
-};
+pub(crate) use cycle_state::{current_scanner_leader_epoch, decode_persisted_scanner_cycle_fence};
 pub use heal_info::{BackgroundHealInfo, read_background_heal_info, save_background_heal_info};
 pub use usage_store::store_data_usage_in_backend;
 

crates/scanner/src/scanner/leadership.rs

@@ -196,7 +196,7 @@ pub(super) async fn claim_scanner_leadership(
         if ctx.is_cancelled() {
             return false;
         }
-        let Some(claimed_epoch) = persisted_epoch.checked_add(1).filter(|epoch| *epoch < u64::MAX) else {
+        let Some(claimed_epoch) = persisted_epoch.checked_add(1) else {
             error!(
                 target: "rustfs::scanner",
                 event = EVENT_SCANNER_PERSIST_STATE,

crates/scanner/src/scanner_budget.rs

@@ -14,16 +14,17 @@
 
 use std::sync::{
     Arc,
-    atomic::{AtomicBool, AtomicU8, AtomicU64, Ordering},
+    atomic::{AtomicU8, AtomicU64, Ordering},
 };
-use tokio::time::{Duration, Instant};
+use std::time::Instant;
+
+use tokio::time::Duration;
 use tokio_util::sync::CancellationToken;
 
 const BUDGET_REASON_NONE: u8 = 0;
 const BUDGET_REASON_RUNTIME: u8 = 1;
 const BUDGET_REASON_OBJECTS: u8 = 2;
 const BUDGET_REASON_DIRECTORIES: u8 = 3;
-const PROGRESS_CLOCK_SAMPLE_INTERVAL: u64 = 128;
 
 #[derive(Clone, Copy, Debug, Default, PartialEq, Eq)]
 pub(crate) struct ScannerCycleBudgetConfig {
@@ -62,51 +63,29 @@ pub struct ScannerCycleBudget {
     token: CancellationToken,
     reason: Arc<AtomicU8>,
     started_at: Instant,
-    deadline: Option<Instant>,
     max_duration: Option<Duration>,
     max_objects: Option<u64>,
     max_directories: Option<u64>,
     track_progress: bool,
-    track_unbounded_counts: bool,
     objects_scanned: AtomicU64,
     directories_started: AtomicU64,
     entries_visited: AtomicU64,
-    last_progress_millis: AtomicU64,
-    cycle_state_persisted: AtomicBool,
 }
 
 impl ScannerCycleBudget {
-    #[cfg(test)]
     pub(crate) fn new(parent: &CancellationToken, config: ScannerCycleBudgetConfig) -> Arc<Self> {
-        Self::new_inner(parent, config, false, false)
+        Self::new_inner(parent, config, false)
     }
 
     pub(crate) fn new_with_progress_tracking(parent: &CancellationToken, config: ScannerCycleBudgetConfig) -> Arc<Self> {
-        Self::new_inner(parent, config, true, true)
+        Self::new_inner(parent, config, true)
     }
 
-    pub(crate) fn new_with_runtime_progress_tracking(parent: &CancellationToken, config: ScannerCycleBudgetConfig) -> Arc<Self> {
-        let track_progress = config.max_duration.is_some();
-        Self::new_inner(parent, config, track_progress, false)
-    }
-
-    fn new_inner(
-        parent: &CancellationToken,
-        config: ScannerCycleBudgetConfig,
-        track_progress: bool,
-        track_unbounded_counts: bool,
-    ) -> Arc<Self> {
+    fn new_inner(parent: &CancellationToken, config: ScannerCycleBudgetConfig, track_progress: bool) -> Arc<Self> {
         let token = parent.child_token();
         let reason = Arc::new(AtomicU8::new(BUDGET_REASON_NONE));
-        let started_at = Instant::now();
-        let deadline = config.max_duration.map(|duration| match started_at.checked_add(duration) {
-            Some(deadline) => deadline,
-            // Runtime config rejects this range, but keep programmatic callers
-            // fail-closed instead of panicking or silently disabling the wall clock.
-            None => started_at,
-        });
 
-        if let Some(deadline) = deadline {
+        if let Some(duration) = config.max_duration {
             let parent = parent.clone();
             let token_wait = token.clone();
             let token_cancel = token.clone();
@@ -115,7 +94,7 @@ impl ScannerCycleBudget {
                 tokio::select! {
                     _ = parent.cancelled() => {}
                     _ = token_wait.cancelled() => {}
-                    _ = tokio::time::sleep_until(deadline) => {
+                    _ = tokio::time::sleep(duration) => {
                         Self::cancel_for_reason(&reason, &token_cancel, ScannerCycleBudgetReason::Runtime);
                     }
                 }
@@ -125,18 +104,14 @@ impl ScannerCycleBudget {
         Arc::new(Self {
             token,
             reason,
-            started_at,
-            deadline,
+            started_at: Instant::now(),
             max_duration: config.max_duration,
             max_objects: config.max_objects,
             max_directories: config.max_directories,
             track_progress,
-            track_unbounded_counts,
             objects_scanned: AtomicU64::new(0),
             directories_started: AtomicU64::new(0),
             entries_visited: AtomicU64::new(0),
-            last_progress_millis: AtomicU64::new(0),
-            cycle_state_persisted: AtomicBool::new(false),
         })
     }
 
@@ -156,14 +131,6 @@ impl ScannerCycleBudget {
         self.max_duration
     }
 
-    pub(crate) fn deadline(&self) -> Option<Instant> {
-        self.deadline
-    }
-
-    pub(crate) fn cancel_for_runtime(&self) {
-        self.cancel_for(ScannerCycleBudgetReason::Runtime);
-    }
-
     pub(crate) fn max_objects(&self) -> Option<u64> {
         self.max_objects
     }
@@ -206,43 +173,15 @@ impl ScannerCycleBudget {
         self.entries_visited.load(Ordering::Relaxed)
     }
 
-    pub(crate) fn mark_cycle_state_persisted(&self) {
-        self.cycle_state_persisted.store(true, Ordering::Release);
-    }
-
-    pub(crate) fn cycle_state_persisted(&self) -> bool {
-        self.cycle_state_persisted.load(Ordering::Acquire)
-    }
-
-    pub(crate) fn progress_age(&self) -> Duration {
-        let elapsed_millis = u64::try_from(self.started_at.elapsed().as_millis()).unwrap_or(u64::MAX);
-        let last_progress = self.last_progress_millis.load(Ordering::Relaxed);
-        Duration::from_millis(elapsed_millis.saturating_sub(last_progress))
-    }
-
-    fn record_progress_sample(&self, event: u64) {
-        // Clock reads are sampled at batch/count boundaries; the scanner's
-        // per-object path does not add a second progress atomic.
-        if event == 0 || (event != 1 && !event.is_multiple_of(PROGRESS_CLOCK_SAMPLE_INTERVAL)) {
-            return;
-        }
-        let elapsed_millis = u64::try_from(self.started_at.elapsed().as_millis()).unwrap_or(u64::MAX);
-        self.last_progress_millis.store(elapsed_millis, Ordering::Relaxed);
-    }
-
     pub(crate) fn record_entries_visited(&self, entries_visited: u64) {
         if self.track_progress {
-            let entries = saturating_fetch_add(&self.entries_visited, entries_visited);
-            self.record_progress_sample(entries);
+            saturating_fetch_add(&self.entries_visited, entries_visited);
         }
     }
 
     pub(crate) fn record_remote_progress(&self, objects_scanned: u64, directories_started: u64) {
         if self.track_progress || self.max_objects.is_some() {
             let objects = saturating_fetch_add(&self.objects_scanned, objects_scanned);
-            if self.track_progress {
-                self.record_progress_sample(objects);
-            }
             if self.max_objects.is_some_and(|max_objects| objects >= max_objects) {
                 self.cancel_for(ScannerCycleBudgetReason::Objects);
             }
@@ -250,12 +189,9 @@ impl ScannerCycleBudget {
 
         if self.track_progress || self.max_directories.is_some() {
             let directories = saturating_fetch_add(&self.directories_started, directories_started);
-            if self.track_progress {
-                self.record_progress_sample(directories);
-            }
             if self
                 .max_directories
-                .is_some_and(|max_directories| directory_budget_exhausted(directories, max_directories))
+                .is_some_and(|max_directories| directories > max_directories)
             {
                 self.cancel_for(ScannerCycleBudgetReason::Directories);
             }
@@ -271,17 +207,14 @@ impl ScannerCycleBudget {
     }
 
     pub(crate) fn try_start_directory(&self) -> bool {
-        if self.max_directories.is_none() && !self.track_unbounded_counts {
+        if !self.track_progress && self.max_directories.is_none() {
             return true;
         }
 
         let directories = saturating_fetch_add(&self.directories_started, 1);
-        if self.track_progress {
-            self.record_progress_sample(directories);
-        }
         if self
             .max_directories
-            .is_some_and(|max_directories| directory_budget_exhausted(directories, max_directories))
+            .is_some_and(|max_directories| directories > max_directories)
         {
             self.cancel_for(ScannerCycleBudgetReason::Directories);
             return false;
@@ -291,14 +224,11 @@ impl ScannerCycleBudget {
     }
 
     pub(crate) fn record_object_scanned(&self) {
-        if self.max_objects.is_none() && !self.track_unbounded_counts {
+        if !self.track_progress && self.max_objects.is_none() {
             return;
         }
 
         let objects = saturating_fetch_add(&self.objects_scanned, 1);
-        if self.track_progress {
-            self.record_progress_sample(objects);
-        }
         if self.max_objects.is_some_and(|max_objects| objects >= max_objects) {
             self.cancel_for(ScannerCycleBudgetReason::Objects);
         }
@@ -329,13 +259,6 @@ fn saturating_fetch_add(value: &AtomicU64, delta: u64) -> u64 {
     }
 }
 
-fn directory_budget_exhausted(directories: u64, max_directories: u64) -> bool {
-    // Saturation hides a remote max+1 update when the configured limit is the
-    // largest representable counter. Treat that boundary as exhausted rather
-    // than allowing work to continue indefinitely.
-    directories > max_directories || (directories == u64::MAX && max_directories == u64::MAX)
-}
-
 impl Drop for ScannerCycleBudget {
     fn drop(&mut self) {
         self.token.cancel();
@@ -478,35 +401,6 @@ mod tests {
         assert_eq!(directory_budget.reason(), Some(ScannerCycleBudgetReason::Directories));
     }
 
-    #[test]
-    fn directory_budget_fails_closed_when_progress_saturates() {
-        let parent = CancellationToken::new();
-        let budget = ScannerCycleBudget::new(
-            &parent,
-            ScannerCycleBudgetConfig {
-                max_directories: Some(u64::MAX),
-                ..Default::default()
-            },
-        );
-
-        budget.record_remote_progress(0, u64::MAX);
-
-        assert_eq!(budget.reason(), Some(ScannerCycleBudgetReason::Directories));
-        assert!(budget.token().is_cancelled());
-
-        let local_budget = ScannerCycleBudget::new(
-            &parent,
-            ScannerCycleBudgetConfig {
-                max_directories: Some(u64::MAX),
-                ..Default::default()
-            },
-        );
-        local_budget.record_remote_progress(0, u64::MAX - 1);
-        assert!(!local_budget.budget_elapsed());
-        assert!(!local_budget.try_start_directory());
-        assert_eq!(local_budget.reason(), Some(ScannerCycleBudgetReason::Directories));
-    }
-
     #[test]
     fn explicit_progress_tracking_counts_unbounded_remote_work_without_cancelling() {
         let parent = CancellationToken::new();
@@ -567,29 +461,4 @@ mod tests {
         assert!(object_limited.requires_serial_progress_accounting());
         assert!(directory_limited.requires_serial_progress_accounting());
     }
-
-    #[tokio::test(start_paused = true)]
-    async fn progress_age_uses_virtual_time_and_sampled_progress() {
-        let parent = CancellationToken::new();
-        let budget = ScannerCycleBudget::new_with_runtime_progress_tracking(
-            &parent,
-            ScannerCycleBudgetConfig {
-                max_duration: Some(Duration::from_secs(60)),
-                ..Default::default()
-            },
-        );
-
-        tokio::time::advance(Duration::from_secs(5)).await;
-        assert_eq!(budget.progress_age(), Duration::from_secs(5));
-        budget.record_entries_visited(1);
-        assert_eq!(budget.progress_age(), Duration::ZERO);
-
-        tokio::time::advance(Duration::from_secs(2)).await;
-        for _ in 0..126 {
-            budget.record_entries_visited(1);
-        }
-        assert_eq!(budget.progress_age(), Duration::from_secs(2));
-        budget.record_entries_visited(1);
-        assert_eq!(budget.progress_age(), Duration::ZERO);
-    }
 }

crates/scanner/src/scanner_io.rs

@@ -48,7 +48,6 @@ use time::OffsetDateTime;
 use tokio::sync::{Mutex, Notify, Semaphore, mpsc};
 use tokio::time::Duration;
 use tokio_util::sync::CancellationToken;
-use tokio_util::task::AbortOnDropHandle;
 use tracing::{debug, error, warn};
 
 use crate::ScannerObjectInfo as ObjectInfo;

crates/scanner/src/scanner_io/io_cache.rs

@@ -314,7 +314,7 @@ impl ScannerIOCache for SetDisks {
         let ctx_clone = ctx.clone();
         let completed_bucket_count = Arc::new(AtomicUsize::new(0));
         let completed_bucket_count_clone = completed_bucket_count.clone();
-        let collect_bucket_results_fut = AbortOnDropHandle::new(tokio::spawn(async move {
+        let collect_bucket_results_fut = tokio::spawn(async move {
             let mut cancelled = false;
 
             loop {
@@ -333,7 +333,7 @@ impl ScannerIOCache for SetDisks {
                     }
                 }
             }
-        }));
+        });
 
         let mut futs = Vec::new();
 
@@ -365,7 +365,7 @@ impl ScannerIOCache for SetDisks {
                 NamespaceScannerWorkerMode::RemoteV4(server_epoch) => Some(server_epoch),
                 NamespaceScannerWorkerMode::Coordinator => None,
             };
-            futs.push(AbortOnDropHandle::new(tokio::spawn(async move {
+            futs.push(tokio::spawn(async move {
                 let remote_session_id = uuid::Uuid::new_v4();
                 let mut remote_session_sequence = 0_u64;
                 loop {
@@ -1038,7 +1038,7 @@ impl ScannerIOCache for SetDisks {
                         );
                     }
                 }
-            })));
+            }));
         }
         drop(bucket_tx);
         drop(bucket_result_tx);

crates/scanner/src/scanner_io/io_cycle.rs

@@ -242,7 +242,7 @@ impl ScannerIOCycle for ECStore {
                         results[results_index_clone] = result;
                     }
                 });
-                wait_futs.push(AbortOnDropHandle::new(receiver_fut));
+                wait_futs.push(receiver_fut);
 
                 let scan_plan = ScannerBucketScanPlan {
                     buckets: set_buckets,
@@ -318,7 +318,7 @@ impl ScannerIOCycle for ECStore {
                         record_set_scan_failure(&mut first_err, e);
                     }
                 });
-                wait_futs.push(AbortOnDropHandle::new(scanner_fut));
+                wait_futs.push(scanner_fut);
             }
         }
 

crates/utils/src/envs.rs

@@ -268,7 +268,7 @@ where
         .parse::<T>()
         .map_err(|_| {
             log_once(&format!("env_invalid_value:{used_key}"), || {
-                format!("Invalid {} value for {used_key}. Treating as unset.", type_name::<T>())
+                format!("Invalid {} value for {used_key}: {value}. Treating as unset.", type_name::<T>())
             });
         })
         .ok()
@@ -570,7 +570,7 @@ where
         Ok(parsed) => EnvParseOutcome::Parsed(parsed),
         Err(_) => {
             log_once(&format!("env_invalid_value:{used_key}"), || {
-                format!("Invalid {} value for {used_key}. Treating as unset.", type_name::<T>())
+                format!("Invalid {} value for {used_key}: {value}. Treating as unset.", type_name::<T>())
             });
             EnvParseOutcome::Invalid
         }

docs/operations/scanner-runtime-controls.md

@@ -52,7 +52,7 @@ The `/v3/scanner/status` response reports each effective runtime value with a
 | `scanner.max_wait` | `RUSTFS_SCANNER_MAX_WAIT_SECS` | seconds | preset-derived | Caps one scanner sleep. |
 | `scanner.cycle` | `RUSTFS_SCANNER_CYCLE` | seconds | preset-derived | Sets the interval between scanner cycles. |
 | `scanner.start_delay` | `RUSTFS_SCANNER_START_DELAY_SECS` | seconds | unset | Sets startup delay and, for compatibility, the cycle interval when `scanner.cycle` is unset. |
-| `scanner.cycle_max_duration` | `RUSTFS_SCANNER_CYCLE_MAX_DURATION_SECS` | seconds | `1800` | Caps one cycle's runtime. An explicit `0` disables this budget. |
+| `scanner.cycle_max_duration` | `RUSTFS_SCANNER_CYCLE_MAX_DURATION_SECS` | seconds | `0` | Caps one cycle's runtime. `0` disables this budget. |
 | `scanner.cycle_max_objects` | `RUSTFS_SCANNER_CYCLE_MAX_OBJECTS` | objects | `0` | Caps objects processed by one cycle. `0` disables this budget. |
 | `scanner.cycle_max_directories` | `RUSTFS_SCANNER_CYCLE_MAX_DIRECTORIES` | directories | `0` | Caps directories entered by one cycle. `0` disables this budget. |
 | `heal.bitrot_cycle` | `RUSTFS_SCANNER_BITROT_CYCLE_SECS` | seconds | `2592000` | Controls periodic deep bitrot scans. `false`, `off`, `no`, or `disabled` disables periodic deep scans; `0`, `true`, `on`, or `yes` runs deep mode every scanner cycle. |
@@ -70,21 +70,6 @@ sleep multiplier, maximum wait, and cycle interval. Use `scanner.delay`,
 `scanner.max_wait`, and `scanner.cycle` when the preset is close but one axis
 needs a precise override.
 
-When the cycle duration control is unset, RustFS uses a finite 1800-second
-(30-minute) default, matching the scanner benchmark guidance. An explicit `0`
-preserves the compatibility behavior of an unbounded cycle; object and
-directory budgets likewise remain unbounded when explicitly set to `0`. Invalid
-or overflowing duration environment values are configuration errors rather than
-silent fallback values.
-
-When a finite deadline expires, RustFS cancels cooperative scanner work and
-waits only for the existing bounded shutdown window. A non-yielding I/O future
-is dropped after that window. RustFS then attempts a higher leadership epoch so
-late cycle, usage, cache, and remote writes from the old generation fail closed.
-If the worker cannot stop cooperatively, the cycle state was not confirmed
-durable, or that epoch fence cannot be durably persisted, the scanner reports
-`recovery-required`; it does not claim an uncooperative cursor was saved.
-
 An explicit `scanner.cycle` or `RUSTFS_SCANNER_CYCLE` is a minimum inter-cycle
 cadence: dirty-usage notifications do not bypass that configured interval.
 The default adaptive policy continues to use dirty-usage notifications to wake
@@ -159,10 +144,6 @@ metrics.maintenance_control.primary_control
 metrics.source_work
 metrics.replication_repair
 metrics.scan_checkpoint
-metrics.cycle_timeout_total
-metrics.cycle_last_progress_age
-metrics.leader_lease_without_progress
-metrics.cycle_recovery_required_total
 ```
 
 ## Reading Pacing Pressure

rustfs/src/admin/handlers/mod.rs

@@ -126,7 +126,6 @@ mod tests {
         let _list_remote_target_handler = replication::ListRemoteTargetHandler {};
         let _remove_remote_target_handler = replication::RemoveRemoteTargetHandler {};
         let _scanner_status_handler = scanner::ScannerStatusHandler {};
-        let _scanner_cycle_state_reset_handler = scanner::ScannerCycleStateResetHandler {};
         let _ilm_expiry_status_handler = scanner::IlmExpiryStatusHandler {};
         let _manual_transition_handler = ilm_transition::ManualTransitionRunHandler {};
         let _manual_transition_status_handler = ilm_transition::ManualTransitionJobStatusHandler {};

rustfs/src/admin/handlers/scanner.rs

@@ -13,11 +13,8 @@
 // limitations under the License.
 
 use crate::admin::auth::authorize_admin_request;
-use crate::admin::handlers::supervise_admin_mutation;
 use crate::admin::router::{AdminOperation, Operation, S3Router};
-use crate::admin::runtime_sources::{
-    app_context_from_req, current_object_store_handle_for_context, current_scanner_metrics_report,
-};
+use crate::admin::runtime_sources::current_scanner_metrics_report;
 use crate::module_switches::{ENV_SCANNER_ENABLED, scanner_enabled_from_env};
 use crate::server::ADMIN_PREFIX;
 use chrono::Utc;
@@ -25,13 +22,11 @@ use http::{HeaderMap, HeaderValue};
 use hyper::{Method, StatusCode};
 use matchit::Params;
 use rustfs_common::metrics::{ScannerLifecycleExpirySnapshot, ScannerMaintenanceControlSnapshot, ScannerMetricsReport};
-use rustfs_config::MAX_ADMIN_REQUEST_BODY_SIZE;
 use rustfs_credentials::Credentials;
 use rustfs_policy::policy::action::{Action, AdminAction};
 use s3s::header::CONTENT_TYPE;
 use s3s::{Body, S3Error, S3ErrorCode, S3Request, S3Response, S3Result, s3_error};
-use serde::{Deserialize, Serialize};
-use tokio_util::sync::CancellationToken;
+use serde::Serialize;
 
 const JSON_CONTENT_TYPE: &str = "application/json";
 
@@ -43,13 +38,6 @@ struct ScannerStatusResponse {
     metrics: ScannerMetricsReport,
     cycle_schedule: rustfs_scanner::ScannerCycleScheduleStatus,
     runtime_config: rustfs_scanner::runtime_config::ScannerRuntimeConfigStatus,
-    cycle_recovery: rustfs_scanner::ScannerCycleRecoveryStatus,
-}
-
-#[derive(Debug, Deserialize)]
-#[serde(deny_unknown_fields)]
-struct ScannerCycleResetRequest {
-    mode: String,
 }
 
 #[derive(Debug, Serialize)]
@@ -129,7 +117,6 @@ fn scanner_status_response(
         metrics,
         cycle_schedule,
         runtime_config,
-        cycle_recovery: rustfs_scanner::scanner::scanner_cycle_recovery_status(),
     }
 }
 
@@ -157,11 +144,6 @@ pub fn register_scanner_route(r: &mut S3Router<AdminOperation>) -> std::io::Resu
         format!("{ADMIN_PREFIX}/v3/scanner/status").as_str(),
         AdminOperation(&ScannerStatusHandler {}),
     )?;
-    r.insert(
-        Method::POST,
-        format!("{ADMIN_PREFIX}/v3/scanner/cycle-state/reset").as_str(),
-        AdminOperation(&ScannerCycleStateResetHandler {}),
-    )?;
     r.insert(
         Method::GET,
         format!("{ADMIN_PREFIX}/v3/ilm/expiry/status").as_str(),
@@ -181,13 +163,6 @@ async fn validate_scanner_status_request(req: &S3Request<Body>) -> S3Result<Cred
     authorize_admin_request(req, vec![Action::AdminAction(AdminAction::ServerInfoAdminAction)]).await
 }
 
-async fn validate_scanner_reset_request(req: &S3Request<Body>) -> S3Result<Credentials> {
-    if req.credentials.is_none() {
-        return Err(s3_error!(InvalidRequest, "missing credentials"));
-    }
-    authorize_admin_request(req, vec![Action::AdminAction(AdminAction::ConfigUpdateAdminAction)]).await
-}
-
 fn json_response(body: Vec<u8>) -> S3Result<S3Response<(StatusCode, Body)>> {
     let mut headers = HeaderMap::new();
     let content_type = HeaderValue::from_str(JSON_CONTENT_TYPE)
@@ -217,37 +192,6 @@ impl Operation for ScannerStatusHandler {
 
 pub struct IlmExpiryStatusHandler {}
 
-pub struct ScannerCycleStateResetHandler {}
-
-#[async_trait::async_trait]
-impl Operation for ScannerCycleStateResetHandler {
-    async fn call(&self, mut req: S3Request<Body>, _params: Params<'_, '_>) -> S3Result<S3Response<(StatusCode, Body)>> {
-        let _cred = validate_scanner_reset_request(&req).await?;
-        let body = req
-            .input
-            .store_all_limited(MAX_ADMIN_REQUEST_BODY_SIZE)
-            .await
-            .map_err(|err| S3Error::with_message(S3ErrorCode::InvalidRequest, format!("invalid reset request body: {err}")))?;
-        let reset = serde_json::from_slice::<ScannerCycleResetRequest>(&body)
-            .map_err(|err| S3Error::with_message(S3ErrorCode::InvalidRequest, format!("invalid reset request body: {err}")))?;
-        if reset.mode != "full-rescan" {
-            return Err(S3Error::with_message(S3ErrorCode::InvalidRequest, "reset mode must be full-rescan"));
-        }
-        let context = app_context_from_req(&req)
-            .ok_or_else(|| S3Error::with_message(S3ErrorCode::InternalError, "storage layer not initialized"))?;
-        let store = current_object_store_handle_for_context(Some(context.as_ref()))
-            .ok_or_else(|| S3Error::with_message(S3ErrorCode::InternalError, "storage layer not initialized"))?;
-        supervise_admin_mutation("scanner cycle state reset", async move {
-            rustfs_scanner::scanner::reset_scanner_cycle_recovery(CancellationToken::new(), store)
-                .await
-                .map_err(|err| S3Error::with_message(S3ErrorCode::InternalError, err.to_string()))?;
-            Ok::<_, S3Error>(())
-        })
-        .await?;
-        json_response(br#"{"status":"reset","mode":"full-rescan"}"#.to_vec())
-    }
-}
-
 #[async_trait::async_trait]
 impl Operation for IlmExpiryStatusHandler {
     async fn call(&self, req: S3Request<Body>, _params: Params<'_, '_>) -> S3Result<S3Response<(StatusCode, Body)>> {
@@ -293,38 +237,6 @@ mod tests {
         assert_eq!(err.message(), Some("missing credentials"));
     }
 
-    #[tokio::test]
-    async fn scanner_reset_gate_rejects_missing_credentials() {
-        let req = S3Request {
-            input: Body::from(String::new()),
-            method: Method::POST,
-            uri: http::Uri::from_static("/rustfs/admin/v3/scanner/cycle-state/reset"),
-            headers: HeaderMap::new(),
-            extensions: http::Extensions::new(),
-            credentials: None,
-            region: None,
-            service: None,
-            trailing_headers: None,
-        };
-
-        let err = validate_scanner_reset_request(&req)
-            .await
-            .expect_err("a reset request without credentials must be rejected");
-        assert_eq!(err.code(), &S3ErrorCode::InvalidRequest);
-        assert_eq!(err.message(), Some("missing credentials"));
-    }
-
-    #[test]
-    fn admin_reset_requires_full_rescan_or_verified_cursor() {
-        let full_rescan: ScannerCycleResetRequest =
-            serde_json::from_str(r#"{"mode":"full-rescan"}"#).expect("full rescan must be accepted");
-        assert_eq!(full_rescan.mode, "full-rescan");
-        let cursor: ScannerCycleResetRequest =
-            serde_json::from_str(r#"{"mode":"cursor"}"#).expect("mode validation belongs to the handler");
-        assert_ne!(cursor.mode, "full-rescan");
-        assert!(serde_json::from_str::<ScannerCycleResetRequest>(r#"{"mode":"full-rescan","cursor":"untrusted"}"#).is_err());
-    }
-
     #[test]
     fn scanner_disabled_reason_reports_startup_env_key() {
         assert_eq!(scanner_disabled_reason(true), None);
@@ -392,11 +304,6 @@ mod tests {
         assert_eq!(encoded["cycle_schedule"]["effective_interval_seconds"], 0);
         assert_eq!(encoded["cycle_schedule"]["clean_idle_backoff_enabled"], false);
         assert_eq!(encoded["cycle_schedule"]["clean_idle_backoff_multiplier"], 1);
-        assert_eq!(encoded["cycle_recovery"]["state"], "healthy");
-        assert_eq!(
-            encoded["cycle_recovery"]["quarantine_path"],
-            rustfs_scanner::DATA_USAGE_BLOOM_RECOVERY_PATH.as_str()
-        );
     }
 
     #[test]

rustfs/src/admin/route_policy.rs

@@ -428,12 +428,6 @@ pub const ADMIN_ROUTE_POLICY_SPECS: &[AdminRouteSpec] = &[
     admin(HttpMethod::Get, "/rustfs/admin/v3/config", CONFIG_UPDATE, RouteRiskLevel::High),
     admin(HttpMethod::Put, "/rustfs/admin/v3/config", CONFIG_UPDATE, RouteRiskLevel::High),
     admin(HttpMethod::Get, "/rustfs/admin/v3/scanner/status", SERVER_INFO, RouteRiskLevel::Sensitive),
-    admin(
-        HttpMethod::Post,
-        "/rustfs/admin/v3/scanner/cycle-state/reset",
-        CONFIG_UPDATE,
-        RouteRiskLevel::High,
-    ),
     admin(
         HttpMethod::Get,
         "/rustfs/admin/v3/ilm/expiry/status",
@@ -2026,12 +2020,6 @@ mod tests {
         assert_not_action(HttpMethod::Get, "/rustfs/admin/v3/ilm/expiry/status", SET_TIER);
     }
 
-    #[test]
-    fn route_policy_requires_config_update_for_scanner_cycle_reset() {
-        assert_action(HttpMethod::Post, "/rustfs/admin/v3/scanner/cycle-state/reset", CONFIG_UPDATE);
-        assert_not_action(HttpMethod::Post, "/rustfs/admin/v3/scanner/cycle-state/reset", SERVER_INFO);
-    }
-
     #[test]
     fn route_policy_uses_tier_actions_for_transition_routes() {
         assert_action(HttpMethod::Post, "/rustfs/admin/v3/ilm/transition/run", SET_TIER);

rustfs/src/admin/route_registration_test.rs

@@ -243,7 +243,6 @@ fn expected_admin_route_matrix() -> Vec<RouteMatrixEntry> {
         admin_route(Method::GET, "/v3/config"),
         admin_route(Method::PUT, "/v3/config"),
         admin_route(Method::GET, "/v3/scanner/status"),
-        admin_route(Method::POST, "/v3/scanner/cycle-state/reset"),
         admin_route(Method::GET, "/v3/audit/target/list"),
         admin_route_sample(
             Method::PUT,
@@ -880,7 +879,6 @@ fn test_register_routes_cover_representative_admin_paths() {
     assert_route(&router, Method::GET, &admin_path("/v3/config"));
     assert_route(&router, Method::PUT, &admin_path("/v3/config"));
     assert_route(&router, Method::GET, &admin_path("/v3/scanner/status"));
-    assert_route(&router, Method::POST, &admin_path("/v3/scanner/cycle-state/reset"));
     assert_route(&router, Method::GET, &admin_path("/v3/ilm/expiry/status"));
     assert_route(&router, Method::POST, &admin_path("/v3/ilm/transition/run"));
     assert_route(
@@ -1369,7 +1367,6 @@ fn test_admin_alias_paths_match_existing_admin_routes() {
         (Method::GET, compat_admin_alias_path("/v3/config")),
         (Method::PUT, compat_admin_alias_path("/v3/config")),
         (Method::GET, compat_admin_alias_path("/v3/scanner/status")),
-        (Method::POST, compat_admin_alias_path("/v3/scanner/cycle-state/reset")),
         (Method::GET, compat_admin_alias_path("/v3/ilm/expiry/status")),
     ] {
         assert!(