Signed-off-by: junxiang Mu <1948535941@qq.com>
This commit is contained in:
junxiang Mu
2025-07-24 11:08:32 +08:00
parent 2a8c46874d
commit 4fefd63a5b
12 changed files with 408 additions and 1398 deletions
+1 -4
View File
@@ -22,10 +22,7 @@ pub mod scanner;
pub use error::{Error, Result};
pub use heal::{channel::HealChannelProcessor, HealManager, HealOptions, HealPriority, HealRequest, HealType};
pub use scanner::{
BucketTargetUsageInfo, BucketUsageInfo, DataUsageInfo, Scanner, ScannerMetrics, load_data_usage_from_backend,
store_data_usage_in_backend,
};
pub use scanner::Scanner;
// Global cancellation token for AHM services (scanner and other background tasks)
static GLOBAL_AHM_SERVICES_CANCEL_TOKEN: OnceLock<CancellationToken> = OnceLock::new();
+2 -1
View File
@@ -527,6 +527,7 @@ impl Scanner {
mod_time: p.mod_time,
index: p.index.clone(),
checksums: p.checksums.clone(),
error: None,
})
.collect(),
erasure: rustfs_filemeta::ErasureInfo {
@@ -1392,8 +1393,8 @@ mod tests {
use rustfs_ecstore::endpoints::{EndpointServerPools, Endpoints, PoolEndpoints};
use rustfs_ecstore::store::ECStore;
use rustfs_ecstore::{
StorageAPI,
store_api::{MakeBucketOptions, ObjectIO, PutObjReader},
StorageAPI,
};
use serial_test::serial;
use std::fs;
+251 -222
View File
@@ -12,197 +12,258 @@
// See the License for the specific language governing permissions and
// limitations under the License.
use std::collections::HashMap;
use std::{
collections::HashMap,
sync::atomic::{AtomicU64, Ordering},
time::{Duration, SystemTime},
};
/// Size interval for object size histogram
#[derive(Debug, Clone)]
pub struct SizeInterval {
pub start: u64,
pub end: u64,
pub name: &'static str,
use serde::{Deserialize, Serialize};
use tracing::info;
/// Scanner metrics
#[derive(Debug, Clone, Default, Serialize, Deserialize)]
pub struct ScannerMetrics {
/// Total objects scanned since server start
pub objects_scanned: u64,
/// Total object versions scanned since server start
pub versions_scanned: u64,
/// Total directories scanned since server start
pub directories_scanned: u64,
/// Total bucket scans started since server start
pub bucket_scans_started: u64,
/// Total bucket scans finished since server start
pub bucket_scans_finished: u64,
/// Total objects with health issues found
pub objects_with_issues: u64,
/// Total heal tasks queued
pub heal_tasks_queued: u64,
/// Total heal tasks completed
pub heal_tasks_completed: u64,
/// Total heal tasks failed
pub heal_tasks_failed: u64,
/// Total healthy objects found
pub healthy_objects: u64,
/// Total corrupted objects found
pub corrupted_objects: u64,
/// Last scan activity time
pub last_activity: Option<SystemTime>,
/// Current scan cycle
pub current_cycle: u64,
/// Total scan cycles completed
pub total_cycles: u64,
/// Current scan duration
pub current_scan_duration: Option<Duration>,
/// Average scan duration
pub avg_scan_duration: Duration,
/// Objects scanned per second
pub objects_per_second: f64,
/// Buckets scanned per second
pub buckets_per_second: f64,
/// Storage metrics by bucket
pub bucket_metrics: HashMap<String, BucketMetrics>,
/// Disk metrics
pub disk_metrics: HashMap<String, DiskMetrics>,
}
/// Version interval for object versions histogram
#[derive(Debug, Clone)]
pub struct VersionInterval {
pub start: u64,
pub end: u64,
pub name: &'static str,
/// Bucket-specific metrics
#[derive(Debug, Clone, Default, Serialize, Deserialize)]
pub struct BucketMetrics {
/// Bucket name
pub bucket: String,
/// Total objects in bucket
pub total_objects: u64,
/// Total size of objects in bucket (bytes)
pub total_size: u64,
/// Objects with health issues
pub objects_with_issues: u64,
/// Last scan time
pub last_scan_time: Option<SystemTime>,
/// Scan duration
pub scan_duration: Option<Duration>,
/// Heal tasks queued for this bucket
pub heal_tasks_queued: u64,
/// Heal tasks completed for this bucket
pub heal_tasks_completed: u64,
/// Heal tasks failed for this bucket
pub heal_tasks_failed: u64,
}
/// Object size histogram intervals
pub const OBJECTS_HISTOGRAM_INTERVALS: &[SizeInterval] = &[
SizeInterval {
start: 0,
end: 1024 - 1,
name: "LESS_THAN_1_KiB",
},
SizeInterval {
start: 1024,
end: 1024 * 1024 - 1,
name: "1_KiB_TO_1_MiB",
},
SizeInterval {
start: 1024 * 1024,
end: 10 * 1024 * 1024 - 1,
name: "1_MiB_TO_10_MiB",
},
SizeInterval {
start: 10 * 1024 * 1024,
end: 64 * 1024 * 1024 - 1,
name: "10_MiB_TO_64_MiB",
},
SizeInterval {
start: 64 * 1024 * 1024,
end: 128 * 1024 * 1024 - 1,
name: "64_MiB_TO_128_MiB",
},
SizeInterval {
start: 128 * 1024 * 1024,
end: 512 * 1024 * 1024 - 1,
name: "128_MiB_TO_512_MiB",
},
SizeInterval {
start: 512 * 1024 * 1024,
end: u64::MAX,
name: "MORE_THAN_512_MiB",
},
];
/// Object version count histogram intervals
pub const OBJECTS_VERSION_COUNT_INTERVALS: &[VersionInterval] = &[
VersionInterval {
start: 1,
end: 1,
name: "1_VERSION",
},
VersionInterval {
start: 2,
end: 10,
name: "2_TO_10_VERSIONS",
},
VersionInterval {
start: 11,
end: 100,
name: "11_TO_100_VERSIONS",
},
VersionInterval {
start: 101,
end: 1000,
name: "101_TO_1000_VERSIONS",
},
VersionInterval {
start: 1001,
end: u64::MAX,
name: "MORE_THAN_1000_VERSIONS",
},
];
/// Size histogram for object size distribution
#[derive(Debug, Clone, Default)]
pub struct SizeHistogram {
counts: Vec<u64>,
/// Disk-specific metrics
#[derive(Debug, Clone, Default, Serialize, Deserialize)]
pub struct DiskMetrics {
/// Disk path
pub disk_path: String,
/// Total disk space (bytes)
pub total_space: u64,
/// Used disk space (bytes)
pub used_space: u64,
/// Free disk space (bytes)
pub free_space: u64,
/// Objects scanned on this disk
pub objects_scanned: u64,
/// Objects with issues on this disk
pub objects_with_issues: u64,
/// Last scan time
pub last_scan_time: Option<SystemTime>,
/// Whether disk is online
pub is_online: bool,
/// Whether disk is being scanned
pub is_scanning: bool,
}
/// Versions histogram for object version count distribution
#[derive(Debug, Clone, Default)]
pub struct VersionsHistogram {
counts: Vec<u64>,
/// Thread-safe metrics collector
pub struct MetricsCollector {
/// Atomic counters for real-time metrics
objects_scanned: AtomicU64,
versions_scanned: AtomicU64,
directories_scanned: AtomicU64,
bucket_scans_started: AtomicU64,
bucket_scans_finished: AtomicU64,
objects_with_issues: AtomicU64,
heal_tasks_queued: AtomicU64,
heal_tasks_completed: AtomicU64,
heal_tasks_failed: AtomicU64,
current_cycle: AtomicU64,
total_cycles: AtomicU64,
healthy_objects: AtomicU64,
corrupted_objects: AtomicU64,
}
impl SizeHistogram {
/// Create a new size histogram
impl MetricsCollector {
/// Create a new metrics collector
pub fn new() -> Self {
Self {
counts: vec![0; OBJECTS_HISTOGRAM_INTERVALS.len()],
objects_scanned: AtomicU64::new(0),
versions_scanned: AtomicU64::new(0),
directories_scanned: AtomicU64::new(0),
bucket_scans_started: AtomicU64::new(0),
bucket_scans_finished: AtomicU64::new(0),
objects_with_issues: AtomicU64::new(0),
heal_tasks_queued: AtomicU64::new(0),
heal_tasks_completed: AtomicU64::new(0),
heal_tasks_failed: AtomicU64::new(0),
current_cycle: AtomicU64::new(0),
total_cycles: AtomicU64::new(0),
healthy_objects: AtomicU64::new(0),
corrupted_objects: AtomicU64::new(0),
}
}
/// Add a size to the histogram
pub fn add(&mut self, size: u64) {
for (idx, interval) in OBJECTS_HISTOGRAM_INTERVALS.iter().enumerate() {
if size >= interval.start && size <= interval.end {
self.counts[idx] += 1;
break;
}
/// Increment objects scanned count
pub fn increment_objects_scanned(&self, count: u64) {
self.objects_scanned.fetch_add(count, Ordering::Relaxed);
}
/// Increment versions scanned count
pub fn increment_versions_scanned(&self, count: u64) {
self.versions_scanned.fetch_add(count, Ordering::Relaxed);
}
/// Increment directories scanned count
pub fn increment_directories_scanned(&self, count: u64) {
self.directories_scanned.fetch_add(count, Ordering::Relaxed);
}
/// Increment bucket scans started count
pub fn increment_bucket_scans_started(&self, count: u64) {
self.bucket_scans_started.fetch_add(count, Ordering::Relaxed);
}
/// Increment bucket scans finished count
pub fn increment_bucket_scans_finished(&self, count: u64) {
self.bucket_scans_finished.fetch_add(count, Ordering::Relaxed);
}
/// Increment objects with issues count
pub fn increment_objects_with_issues(&self, count: u64) {
self.objects_with_issues.fetch_add(count, Ordering::Relaxed);
}
/// Increment heal tasks queued count
pub fn increment_heal_tasks_queued(&self, count: u64) {
self.heal_tasks_queued.fetch_add(count, Ordering::Relaxed);
}
/// Increment heal tasks completed count
pub fn increment_heal_tasks_completed(&self, count: u64) {
self.heal_tasks_completed.fetch_add(count, Ordering::Relaxed);
}
/// Increment heal tasks failed count
pub fn increment_heal_tasks_failed(&self, count: u64) {
self.heal_tasks_failed.fetch_add(count, Ordering::Relaxed);
}
/// Set current cycle
pub fn set_current_cycle(&self, cycle: u64) {
self.current_cycle.store(cycle, Ordering::Relaxed);
}
/// Increment total cycles
pub fn increment_total_cycles(&self) {
self.total_cycles.fetch_add(1, Ordering::Relaxed);
}
/// Increment healthy objects count
pub fn increment_healthy_objects(&self) {
self.healthy_objects.fetch_add(1, Ordering::Relaxed);
}
/// Increment corrupted objects count
pub fn increment_corrupted_objects(&self) {
self.corrupted_objects.fetch_add(1, Ordering::Relaxed);
}
/// Get current metrics snapshot
pub fn get_metrics(&self) -> ScannerMetrics {
ScannerMetrics {
objects_scanned: self.objects_scanned.load(Ordering::Relaxed),
versions_scanned: self.versions_scanned.load(Ordering::Relaxed),
directories_scanned: self.directories_scanned.load(Ordering::Relaxed),
bucket_scans_started: self.bucket_scans_started.load(Ordering::Relaxed),
bucket_scans_finished: self.bucket_scans_finished.load(Ordering::Relaxed),
objects_with_issues: self.objects_with_issues.load(Ordering::Relaxed),
heal_tasks_queued: self.heal_tasks_queued.load(Ordering::Relaxed),
heal_tasks_completed: self.heal_tasks_completed.load(Ordering::Relaxed),
heal_tasks_failed: self.heal_tasks_failed.load(Ordering::Relaxed),
healthy_objects: self.healthy_objects.load(Ordering::Relaxed),
corrupted_objects: self.corrupted_objects.load(Ordering::Relaxed),
last_activity: Some(SystemTime::now()),
current_cycle: self.current_cycle.load(Ordering::Relaxed),
total_cycles: self.total_cycles.load(Ordering::Relaxed),
current_scan_duration: None, // Will be set by scanner
avg_scan_duration: Duration::ZERO, // Will be calculated
objects_per_second: 0.0, // Will be calculated
buckets_per_second: 0.0, // Will be calculated
bucket_metrics: HashMap::new(), // Will be populated by scanner
disk_metrics: HashMap::new(), // Will be populated by scanner
}
}
/// Get the histogram as a map
pub fn to_map(&self) -> HashMap<String, u64> {
let mut result = HashMap::new();
for (idx, count) in self.counts.iter().enumerate() {
let interval = &OBJECTS_HISTOGRAM_INTERVALS[idx];
result.insert(interval.name.to_string(), *count);
}
result
}
/// Reset all metrics
pub fn reset(&self) {
self.objects_scanned.store(0, Ordering::Relaxed);
self.versions_scanned.store(0, Ordering::Relaxed);
self.directories_scanned.store(0, Ordering::Relaxed);
self.bucket_scans_started.store(0, Ordering::Relaxed);
self.bucket_scans_finished.store(0, Ordering::Relaxed);
self.objects_with_issues.store(0, Ordering::Relaxed);
self.heal_tasks_queued.store(0, Ordering::Relaxed);
self.heal_tasks_completed.store(0, Ordering::Relaxed);
self.heal_tasks_failed.store(0, Ordering::Relaxed);
self.current_cycle.store(0, Ordering::Relaxed);
self.total_cycles.store(0, Ordering::Relaxed);
self.healthy_objects.store(0, Ordering::Relaxed);
self.corrupted_objects.store(0, Ordering::Relaxed);
/// Merge another histogram into this one
pub fn merge(&mut self, other: &SizeHistogram) {
for (idx, count) in other.counts.iter().enumerate() {
self.counts[idx] += count;
}
}
/// Get total count
pub fn total_count(&self) -> u64 {
self.counts.iter().sum()
}
/// Reset the histogram
pub fn reset(&mut self) {
for count in &mut self.counts {
*count = 0;
}
info!("Scanner metrics reset");
}
}
impl VersionsHistogram {
/// Create a new versions histogram
pub fn new() -> Self {
Self {
counts: vec![0; OBJECTS_VERSION_COUNT_INTERVALS.len()],
}
}
/// Add a version count to the histogram
pub fn add(&mut self, versions: u64) {
for (idx, interval) in OBJECTS_VERSION_COUNT_INTERVALS.iter().enumerate() {
if versions >= interval.start && versions <= interval.end {
self.counts[idx] += 1;
break;
}
}
}
/// Get the histogram as a map
pub fn to_map(&self) -> HashMap<String, u64> {
let mut result = HashMap::new();
for (idx, count) in self.counts.iter().enumerate() {
let interval = &OBJECTS_VERSION_COUNT_INTERVALS[idx];
result.insert(interval.name.to_string(), *count);
}
result
}
/// Merge another histogram into this one
pub fn merge(&mut self, other: &VersionsHistogram) {
for (idx, count) in other.counts.iter().enumerate() {
self.counts[idx] += count;
}
}
/// Get total count
pub fn total_count(&self) -> u64 {
self.counts.iter().sum()
}
/// Reset the histogram
pub fn reset(&mut self) {
for count in &mut self.counts {
*count = 0;
}
impl Default for MetricsCollector {
fn default() -> Self {
Self::new()
}
}
@@ -211,67 +272,35 @@ mod tests {
use super::*;
#[test]
fn test_size_histogram() {
let mut histogram = SizeHistogram::new();
// Add some sizes
histogram.add(512); // LESS_THAN_1_KiB
histogram.add(1024); // 1_KiB_TO_1_MiB
histogram.add(1024 * 1024); // 1_MiB_TO_10_MiB
histogram.add(5 * 1024 * 1024); // 1_MiB_TO_10_MiB
let map = histogram.to_map();
assert_eq!(map.get("LESS_THAN_1_KiB"), Some(&1));
assert_eq!(map.get("1_KiB_TO_1_MiB"), Some(&1));
assert_eq!(map.get("1_MiB_TO_10_MiB"), Some(&2));
assert_eq!(map.get("10_MiB_TO_64_MiB"), Some(&0));
fn test_metrics_collector_creation() {
let collector = MetricsCollector::new();
let metrics = collector.get_metrics();
assert_eq!(metrics.objects_scanned, 0);
assert_eq!(metrics.versions_scanned, 0);
}
#[test]
fn test_versions_histogram() {
let mut histogram = VersionsHistogram::new();
fn test_metrics_increment() {
let collector = MetricsCollector::new();
// Add some version counts
histogram.add(1); // 1_VERSION
histogram.add(5); // 2_TO_10_VERSIONS
histogram.add(50); // 11_TO_100_VERSIONS
histogram.add(500); // 101_TO_1000_VERSIONS
collector.increment_objects_scanned(10);
collector.increment_versions_scanned(5);
collector.increment_objects_with_issues(2);
let map = histogram.to_map();
assert_eq!(map.get("1_VERSION"), Some(&1));
assert_eq!(map.get("2_TO_10_VERSIONS"), Some(&1));
assert_eq!(map.get("11_TO_100_VERSIONS"), Some(&1));
assert_eq!(map.get("101_TO_1000_VERSIONS"), Some(&1));
let metrics = collector.get_metrics();
assert_eq!(metrics.objects_scanned, 10);
assert_eq!(metrics.versions_scanned, 5);
assert_eq!(metrics.objects_with_issues, 2);
}
#[test]
fn test_histogram_merge() {
let mut histogram1 = SizeHistogram::new();
histogram1.add(1024);
histogram1.add(1024 * 1024);
fn test_metrics_reset() {
let collector = MetricsCollector::new();
let mut histogram2 = SizeHistogram::new();
histogram2.add(1024);
histogram2.add(5 * 1024 * 1024);
collector.increment_objects_scanned(10);
collector.reset();
histogram1.merge(&histogram2);
let map = histogram1.to_map();
assert_eq!(map.get("1_KiB_TO_1_MiB"), Some(&2)); // 1 from histogram1 + 1 from histogram2
assert_eq!(map.get("1_MiB_TO_10_MiB"), Some(&2)); // 1 from histogram1 + 1 from histogram2
}
#[test]
fn test_histogram_reset() {
let mut histogram = SizeHistogram::new();
histogram.add(1024);
histogram.add(1024 * 1024);
assert_eq!(histogram.total_count(), 2);
histogram.reset();
assert_eq!(histogram.total_count(), 0);
let metrics = collector.get_metrics();
assert_eq!(metrics.objects_scanned, 0);
}
}
-5
View File
@@ -13,13 +13,8 @@
// limitations under the License.
pub mod data_scanner;
pub mod data_usage;
pub mod histogram;
pub mod metrics;
// Re-export main types for convenience
pub use data_scanner::Scanner;
pub use data_usage::{
BucketTargetUsageInfo, BucketUsageInfo, DataUsageInfo, load_data_usage_from_backend, store_data_usage_in_backend,
};
pub use metrics::ScannerMetrics;