refactor(object-capacity): remove the decorative SymlinkTracker and its no-op depth knob (#4571)

The tracker never influenced traversal: walkdir's follow behavior is
fixed up front by follow_links(), and should_follow() only gated the
tracker's own bookkeeping. Its 'depth limit' compared tree depth (not
symlink chain depth), so RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH was a
complete no-op while its telemetry claimed symlinks were skipped that
walkdir had in fact followed and counted; record_symlink was always
called with size 0, so tracked_bytes never left zero (S12).

Remove the tracker, its skipped/summary events, the symlink metric and
the depth env knob end to end (the env's 'as u8' truncation goes with
it), and document the real semantics at the walker: follow_links(true)
counts targets with walkdir's ancestor-loop detection breaking cycles,
follow_links(false) — the default — counts no symlink targets. The scan
root itself is pre-resolved since backlog#1015.

Ref: rustfs/backlog#1018 (S12 from audit rustfs/backlog#1010)

crates/config/src/constants/capacity.rs

@@ -45,9 +45,6 @@ pub const ENV_CAPACITY_METRICS_INTERVAL: &str = "RUSTFS_CAPACITY_METRICS_INTERVA
 /// Environment variable for following symbolic links during capacity calculation
 pub const ENV_CAPACITY_FOLLOW_SYMLINKS: &str = "RUSTFS_CAPACITY_FOLLOW_SYMLINKS";
 
-/// Environment variable for maximum symlink follow depth
-pub const ENV_CAPACITY_MAX_SYMLINK_DEPTH: &str = "RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH";
-
 /// Environment variable for enabling dynamic timeout calculation
 pub const ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT: &str = "RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT";
 
@@ -97,10 +94,6 @@ pub const DEFAULT_CAPACITY_METRICS_INTERVAL_SECS: u64 = 600;
 /// Default: false (disabled for safety)
 pub const DEFAULT_CAPACITY_FOLLOW_SYMLINKS: bool = false;
 
-/// Maximum symlink follow depth
-/// Default: 3 levels
-pub const DEFAULT_CAPACITY_MAX_SYMLINK_DEPTH: u8 = 3;
-
 /// Enable dynamic timeout calculation based on directory characteristics
 /// Default: true (enabled)
 pub const DEFAULT_CAPACITY_ENABLE_DYNAMIC_TIMEOUT: bool = true;
@@ -132,7 +125,6 @@ mod tests {
         assert_eq!(ENV_CAPACITY_SAMPLE_RATE, "RUSTFS_CAPACITY_SAMPLE_RATE");
         assert_eq!(ENV_CAPACITY_METRICS_INTERVAL, "RUSTFS_CAPACITY_METRICS_INTERVAL");
         assert_eq!(ENV_CAPACITY_FOLLOW_SYMLINKS, "RUSTFS_CAPACITY_FOLLOW_SYMLINKS");
-        assert_eq!(ENV_CAPACITY_MAX_SYMLINK_DEPTH, "RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH");
         assert_eq!(ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT, "RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT");
         assert_eq!(ENV_CAPACITY_MIN_TIMEOUT, "RUSTFS_CAPACITY_MIN_TIMEOUT");
         assert_eq!(ENV_CAPACITY_MAX_TIMEOUT, "RUSTFS_CAPACITY_MAX_TIMEOUT");

crates/io-metrics/src/capacity_metrics.rs

@@ -113,13 +113,6 @@ pub fn record_capacity_write_operation(write_frequency: usize) {
     gauge!("rustfs_capacity_write_frequency").set(write_frequency as f64);
 }
 
-/// Record symlink accounting.
-#[inline(always)]
-pub fn record_capacity_symlink(size_bytes: u64) {
-    counter!("rustfs_capacity_symlinks_encountered").increment(1);
-    histogram!("rustfs_capacity_symlinks_size_bytes").record(size_bytes as f64);
-}
-
 /// Record timeout fallback event.
 #[inline(always)]
 pub fn record_capacity_timeout_fallback() {

crates/io-metrics/src/lib.rs

@@ -121,7 +121,7 @@ pub use capacity_metrics::{
     record_capacity_dirty_disk_count, record_capacity_dynamic_timeout, record_capacity_refresh_inflight,
     record_capacity_refresh_joiner, record_capacity_refresh_request, record_capacity_refresh_result,
     record_capacity_refresh_scope, record_capacity_scan_disk, record_capacity_scan_mode, record_capacity_scan_sampling,
-    record_capacity_symlink, record_capacity_timeout_fallback, record_capacity_update_completed, record_capacity_update_failed,
+    record_capacity_timeout_fallback, record_capacity_update_completed, record_capacity_update_failed,
     record_capacity_write_operation, record_old_data_dir_cleanup,
 };
 

crates/object-capacity/README.md

@@ -99,7 +99,7 @@ This crate is intentionally not "timeout means hard failure":
 
 ### Symlink Handling
 
-- Symlinks are not followed by default: `RUSTFS_CAPACITY_FOLLOW_SYMLINKS=false`.
+- Symlinks are not followed by default: `RUSTFS_CAPACITY_FOLLOW_SYMLINKS=false`. When enabled, symlink targets are counted and cycles are broken by the walker's ancestor-loop detection.
 - If enabled, the scan applies circular-reference detection and a maximum follow depth.
 - The default maximum depth is `3`.
 
@@ -289,7 +289,6 @@ The configuration constants are defined in `crates/config/src/constants/capacity
 | `RUSTFS_CAPACITY_SAMPLE_RATE` | `200` | Overflow-file sampling interval |
 | `RUSTFS_CAPACITY_METRICS_INTERVAL` | `600s` | Runtime summary emission interval |
 | `RUSTFS_CAPACITY_FOLLOW_SYMLINKS` | `false` | Whether to follow symlinks |
-| `RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH` | `3` | Maximum symlink follow depth |
 | `RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT` | `true` | Whether to enable dynamic timeout scaling |
 | `RUSTFS_CAPACITY_MIN_TIMEOUT` | `2s` | Dynamic-timeout lower bound |
 | `RUSTFS_CAPACITY_MAX_TIMEOUT` | `15s` | Dynamic-timeout upper bound |

crates/object-capacity/README_ZH.md

@@ -99,7 +99,7 @@ crate 不是“超时就直接失败”的设计：
 
 ### 符号链接处理
 
-- 默认不跟随符号链接：`RUSTFS_CAPACITY_FOLLOW_SYMLINKS=false`。
+- 默认不跟随符号链接：`RUSTFS_CAPACITY_FOLLOW_SYMLINKS=false`。开启后会统计符号链接目标,循环由遍历器的祖先环检测阻断。
 - 开启后会做循环引用检测和最大深度限制。
 - 默认最大深度是 `3`。
 
@@ -289,7 +289,6 @@ get_capacity_manager()
 | `RUSTFS_CAPACITY_SAMPLE_RATE` | `200` | overflow 文件采样间隔 |
 | `RUSTFS_CAPACITY_METRICS_INTERVAL` | `600s` | runtime summary 打点间隔 |
 | `RUSTFS_CAPACITY_FOLLOW_SYMLINKS` | `false` | 是否跟随符号链接 |
-| `RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH` | `3` | 符号链接最大跟随深度 |
 | `RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT` | `true` | 是否启用动态超时 |
 | `RUSTFS_CAPACITY_MIN_TIMEOUT` | `2s` | 动态超时下界 |
 | `RUSTFS_CAPACITY_MAX_TIMEOUT` | `15s` | 动态超时上界 |

crates/object-capacity/src/capacity_manager.rs

@@ -19,14 +19,13 @@ use super::types::CapacityDiskRef;
 use crate::capacity_scope::{CapacityScope, CapacityScopeDisk, drain_global_dirty_scopes, take_capacity_scope};
 use futures::FutureExt;
 use rustfs_config::{
-    DEFAULT_CAPACITY_ENABLE_DYNAMIC_TIMEOUT, DEFAULT_CAPACITY_FOLLOW_SYMLINKS, DEFAULT_CAPACITY_MAX_SYMLINK_DEPTH,
-    DEFAULT_CAPACITY_MAX_TIMEOUT_SECS, DEFAULT_CAPACITY_METRICS_INTERVAL_SECS, DEFAULT_CAPACITY_MIN_TIMEOUT_SECS,
-    DEFAULT_FAST_UPDATE_THRESHOLD_SECS, DEFAULT_MAX_FILES_THRESHOLD, DEFAULT_SAMPLE_RATE, DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS,
-    DEFAULT_STAT_TIMEOUT_SECS, DEFAULT_WRITE_FREQUENCY_THRESHOLD, DEFAULT_WRITE_TRIGGER_DELAY_SECS,
-    ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT, ENV_CAPACITY_FAST_UPDATE_THRESHOLD, ENV_CAPACITY_FOLLOW_SYMLINKS,
-    ENV_CAPACITY_MAX_FILES_THRESHOLD, ENV_CAPACITY_MAX_SYMLINK_DEPTH, ENV_CAPACITY_MAX_TIMEOUT, ENV_CAPACITY_METRICS_INTERVAL,
-    ENV_CAPACITY_MIN_TIMEOUT, ENV_CAPACITY_SAMPLE_RATE, ENV_CAPACITY_SCHEDULED_INTERVAL, ENV_CAPACITY_STAT_TIMEOUT,
-    ENV_CAPACITY_WRITE_FREQUENCY_THRESHOLD, ENV_CAPACITY_WRITE_TRIGGER_DELAY,
+    DEFAULT_CAPACITY_ENABLE_DYNAMIC_TIMEOUT, DEFAULT_CAPACITY_FOLLOW_SYMLINKS, DEFAULT_CAPACITY_MAX_TIMEOUT_SECS,
+    DEFAULT_CAPACITY_METRICS_INTERVAL_SECS, DEFAULT_CAPACITY_MIN_TIMEOUT_SECS, DEFAULT_FAST_UPDATE_THRESHOLD_SECS,
+    DEFAULT_MAX_FILES_THRESHOLD, DEFAULT_SAMPLE_RATE, DEFAULT_SCHEDULED_UPDATE_INTERVAL_SECS, DEFAULT_STAT_TIMEOUT_SECS,
+    DEFAULT_WRITE_FREQUENCY_THRESHOLD, DEFAULT_WRITE_TRIGGER_DELAY_SECS, ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT,
+    ENV_CAPACITY_FAST_UPDATE_THRESHOLD, ENV_CAPACITY_FOLLOW_SYMLINKS, ENV_CAPACITY_MAX_FILES_THRESHOLD, ENV_CAPACITY_MAX_TIMEOUT,
+    ENV_CAPACITY_METRICS_INTERVAL, ENV_CAPACITY_MIN_TIMEOUT, ENV_CAPACITY_SAMPLE_RATE, ENV_CAPACITY_SCHEDULED_INTERVAL,
+    ENV_CAPACITY_STAT_TIMEOUT, ENV_CAPACITY_WRITE_FREQUENCY_THRESHOLD, ENV_CAPACITY_WRITE_TRIGGER_DELAY,
 };
 use rustfs_io_metrics::capacity_metrics::{
     record_capacity_current_bytes, record_capacity_degraded_reading, record_capacity_dirty_disk_count,
@@ -82,8 +81,6 @@ struct CachedCapacityConfig {
     metrics_interval: Duration,
     /// Follow symlinks flag
     follow_symlinks: bool,
-    /// Max symlink depth
-    max_symlink_depth: u8,
     /// Enable dynamic timeout flag
     enable_dynamic_timeout: bool,
     /// Min timeout
@@ -138,7 +135,6 @@ impl CachedCapacityConfig {
                 DEFAULT_CAPACITY_METRICS_INTERVAL_SECS,
             )),
             follow_symlinks: get_env_bool(ENV_CAPACITY_FOLLOW_SYMLINKS, DEFAULT_CAPACITY_FOLLOW_SYMLINKS),
-            max_symlink_depth: get_env_u64(ENV_CAPACITY_MAX_SYMLINK_DEPTH, DEFAULT_CAPACITY_MAX_SYMLINK_DEPTH as u64) as u8,
             enable_dynamic_timeout: get_env_bool(ENV_CAPACITY_ENABLE_DYNAMIC_TIMEOUT, DEFAULT_CAPACITY_ENABLE_DYNAMIC_TIMEOUT),
             min_timeout: Duration::from_secs(env_u64_at_least(ENV_CAPACITY_MIN_TIMEOUT, DEFAULT_CAPACITY_MIN_TIMEOUT_SECS, 1)),
             max_timeout: Duration::from_secs(env_u64_at_least(ENV_CAPACITY_MAX_TIMEOUT, DEFAULT_CAPACITY_MAX_TIMEOUT_SECS, 1)),
@@ -267,18 +263,6 @@ pub fn get_follow_symlinks() -> bool {
     get_cached_config().follow_symlinks
 }
 
-/// Get max symlink depth from environment or default
-#[cfg(not(test))]
-pub fn get_max_symlink_depth() -> u8 {
-    get_cached_config().max_symlink_depth
-}
-
-/// Get max symlink depth from environment or default (test mode)
-#[cfg(test)]
-pub fn get_max_symlink_depth() -> u8 {
-    get_cached_config().max_symlink_depth
-}
-
 /// Get enable dynamic timeout flag from environment or default
 #[cfg(not(test))]
 pub fn get_enable_dynamic_timeout() -> bool {

crates/object-capacity/src/scan.rs

@@ -14,18 +14,18 @@
 
 use super::capacity_manager::{
     CapacityUpdate, DiskCapacityUpdate, HybridCapacityManager, get_enable_dynamic_timeout, get_follow_symlinks,
-    get_max_files_threshold, get_max_symlink_depth, get_max_timeout, get_min_timeout, get_sample_rate, get_stat_timeout,
+    get_max_files_threshold, get_max_timeout, get_min_timeout, get_sample_rate, get_stat_timeout,
 };
 use super::types::{CapacityDiskRef, CapacityScanResult, CapacityScanSummary};
 use crate::capacity_scope::CapacityScopeDisk;
 use futures::{StreamExt, stream};
 use rustfs_io_metrics::capacity_metrics::{
     record_capacity_dynamic_timeout, record_capacity_scan_disk, record_capacity_scan_mode, record_capacity_scan_sampling,
-    record_capacity_symlink, record_capacity_timeout_fallback,
+    record_capacity_timeout_fallback,
 };
 use std::collections::HashSet;
 use std::ffi::OsStr;
-use std::path::{Component, Path, PathBuf};
+use std::path::{Component, Path};
 use std::sync::Arc;
 use std::sync::atomic::{AtomicBool, Ordering};
 use std::time::{Duration, Instant};
@@ -41,8 +41,6 @@ const LOG_SUBSYSTEM_SAMPLING: &str = "sampling";
 const EVENT_CAPACITY_SCAN_DISK_COMPLETED: &str = "capacity_scan_disk_completed";
 const EVENT_CAPACITY_SCAN_DISK_FAILED: &str = "capacity_scan_disk_failed";
 const EVENT_CAPACITY_SCAN_SUMMARY: &str = "capacity_scan_summary";
-const EVENT_CAPACITY_SCAN_SYMLINK_SKIPPED: &str = "capacity_scan_symlink_skipped";
-const EVENT_CAPACITY_SCAN_SYMLINK_SUMMARY: &str = "capacity_scan_symlink_summary";
 const EVENT_CAPACITY_SCAN_DYNAMIC_TIMEOUT: &str = "capacity_scan_dynamic_timeout";
 const EVENT_CAPACITY_SCAN_TIMEOUT: &str = "capacity_scan_timeout";
 const EVENT_CAPACITY_SCAN_SAMPLING_CLAMPED: &str = "capacity_scan_sampling_clamped";
@@ -365,69 +363,6 @@ pub async fn scan_used_capacity_disks(
     Ok(calculate_data_dir_used_capacity(disks).await?.into())
 }
 
-/// Tracker for symlink resolution with circular reference detection.
-struct SymlinkTracker {
-    visited: HashSet<PathBuf>,
-    symlink_count: usize,
-    symlink_size: u64,
-    max_depth: u8,
-}
-
-impl SymlinkTracker {
-    fn new(max_depth: u8) -> Self {
-        Self {
-            visited: HashSet::new(),
-            symlink_count: 0,
-            symlink_size: 0,
-            max_depth,
-        }
-    }
-
-    fn should_follow(&self, path: &Path, depth: u8) -> bool {
-        if depth >= self.max_depth {
-            debug!(
-                event = EVENT_CAPACITY_SCAN_SYMLINK_SKIPPED,
-                component = LOG_COMPONENT_CAPACITY,
-                subsystem = LOG_SUBSYSTEM_SCAN,
-                result = "skipped",
-                reason = "depth_limit",
-                depth,
-                max_depth = self.max_depth,
-                path = ?path,
-                "capacity scan symlink skipped"
-            );
-            return false;
-        }
-
-        if self.visited.contains(path) {
-            warn!(
-                event = EVENT_CAPACITY_SCAN_SYMLINK_SKIPPED,
-                component = LOG_COMPONENT_CAPACITY,
-                subsystem = LOG_SUBSYSTEM_SCAN,
-                result = "skipped",
-                reason = "cycle_detected",
-                path = ?path,
-                "capacity scan symlink skipped"
-            );
-            return false;
-        }
-
-        true
-    }
-
-    fn record_symlink(&mut self, path: PathBuf, size: u64) {
-        if self.visited.insert(path) {
-            self.symlink_count += 1;
-            self.symlink_size += size;
-            record_capacity_symlink(size);
-        }
-    }
-
-    fn get_stats(&self) -> (usize, u64) {
-        (self.symlink_count, self.symlink_size)
-    }
-}
-
 /// Monitor for directory traversal progress with a cooperative timeout.
 ///
 /// The old in-loop stall detector was structurally unreachable: cooperative
@@ -549,7 +484,6 @@ struct ScanLimits {
     sample_rate: usize,
     enable_dynamic_timeout: bool,
     follow_symlinks: bool,
-    max_symlink_depth: u8,
 }
 
 impl ScanLimits {
@@ -579,7 +513,6 @@ impl ScanLimits {
             sample_rate: effective_sample_rate,
             enable_dynamic_timeout: get_enable_dynamic_timeout(),
             follow_symlinks: get_follow_symlinks(),
-            max_symlink_depth: get_max_symlink_depth(),
         }
     }
 }
@@ -723,7 +656,6 @@ fn scan_dir_blocking(path: &Path, limits: &ScanLimits, cancelled: &AtomicBool) -
         sample_rate: effective_sample_rate,
         enable_dynamic_timeout,
         follow_symlinks,
-        max_symlink_depth,
     } = *limits;
     {
         if !path.exists() {
@@ -756,13 +688,17 @@ fn scan_dir_blocking(path: &Path, limits: &ScanLimits, cancelled: &AtomicBool) -
         // elapses before the exact prefix fills (early sampling entry).
         let mut effective_threshold = max_files_threshold;
 
-        let mut symlink_tracker = SymlinkTracker::new(max_symlink_depth);
         let mut progress_monitor = ProgressMonitor::new(base_timeout, min_timeout, max_timeout, enable_dynamic_timeout);
 
-        let walker = WalkDir::new(path)
-            .follow_links(follow_symlinks)
-            .follow_root_links(follow_symlinks)
-            .into_iter();
+        // With follow_links(true), symlink targets are counted and walkdir's
+        // own ancestor-loop detection breaks cycles (diamond-shaped links may
+        // still be counted once per path). With the default (false), symlink
+        // targets are not counted. The old SymlinkTracker layered on top of
+        // this never influenced traversal, reported tree depth as chain depth
+        // and always tracked 0 bytes, so it was removed along with the no-op
+        // RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH knob (backlog#1018 S12). The root
+        // itself is pre-resolved above (backlog#1015).
+        let walker = WalkDir::new(path).follow_links(follow_symlinks).into_iter();
 
         for entry_result in walker {
             if cancelled.load(Ordering::Relaxed) {
@@ -864,14 +800,6 @@ fn scan_dir_blocking(path: &Path, limits: &ScanLimits, cancelled: &AtomicBool) -
                 }
             };
 
-            if follow_symlinks
-                && entry.path_is_symlink()
-                && let Ok(target) = std::fs::read_link(entry.path())
-                && symlink_tracker.should_follow(&target, entry.depth().min(u8::MAX as usize) as u8)
-            {
-                symlink_tracker.record_symlink(target, 0);
-            }
-
             let file_type = entry.file_type();
             if file_type.is_dir() {
                 continue;
@@ -977,19 +905,6 @@ fn scan_dir_blocking(path: &Path, limits: &ScanLimits, cancelled: &AtomicBool) -
             }
         }
 
-        let (symlink_count, symlink_size) = symlink_tracker.get_stats();
-        if symlink_count > 0 {
-            info!(
-                event = EVENT_CAPACITY_SCAN_SYMLINK_SUMMARY,
-                component = LOG_COMPONENT_CAPACITY,
-                subsystem = LOG_SUBSYSTEM_SCAN,
-                result = "observed",
-                symlink_count,
-                tracked_bytes = symlink_size,
-                "capacity scan symlink summary"
-            );
-        }
-
         if file_count > effective_threshold && sampled_count > 0 {
             let overflow_count = file_count - effective_threshold;
             let estimated_overflow = estimate_overflow_bytes(overflow_sampled_bytes, overflow_count as u64, sampled_count as u64);
@@ -1369,7 +1284,6 @@ mod tests {
             sample_rate: 1,
             enable_dynamic_timeout: false,
             follow_symlinks: false,
-            max_symlink_depth: 40,
         }
     }
 

rustfs/src/capacity/mod.rs

@@ -32,7 +32,6 @@
 //! - `RUSTFS_CAPACITY_SAMPLE_RATE` - Sampling rate for metrics (default: 200)
 //! - `RUSTFS_CAPACITY_METRICS_INTERVAL` - Metrics summary logging interval (default: 600s)
 //! - `RUSTFS_CAPACITY_FOLLOW_SYMLINKS` - Follow symlinks during traversal (default: false)
-//! - `RUSTFS_CAPACITY_MAX_SYMLINK_DEPTH` - Max symlink depth (default: 3)
 //! - `RUSTFS_CAPACITY_ENABLE_DYNAMIC_TIMEOUT` - Enable dynamic timeout (default: true)
 //! - `RUSTFS_CAPACITY_MIN_TIMEOUT` - Minimum timeout (default: 2s)
 //! - `RUSTFS_CAPACITY_MAX_TIMEOUT` - Maximum timeout (default: 15s)