Merge branch 'main' into cxymds/fix-1926-owner-selection

crates/config/src/constants/runtime.rs

@@ -57,6 +57,13 @@ pub const DEFAULT_MAX_IO_EVENTS_PER_TICK: usize = 1024;
 pub const DEFAULT_EVENT_INTERVAL: u32 = 61;
 pub const DEFAULT_RNG_SEED: Option<u64> = None; // None means random
 
+/// Dedicated blocking thread pool for fsync/fdatasync operations.
+/// When > 1, fsync operations are isolated from the main blocking pool to
+/// prevent device-bound fsync from starving read operations (pread/stat/open).
+/// Default 0 means auto (no isolation, use main runtime).
+pub const ENV_FSYNC_BLOCKING_THREADS: &str = "RUSTFS_RUNTIME_FSYNC_BLOCKING_THREADS";
+pub const DEFAULT_FSYNC_BLOCKING_THREADS: usize = 0;
+
 // Dial9 Tokio Telemetry Default values
 pub const DEFAULT_RUNTIME_DIAL9_ENABLED: bool = false; // Disabled by default
 pub const DEFAULT_RUNTIME_DIAL9_OUTPUT_DIR: &str = "/var/log/rustfs/telemetry";

crates/e2e_test/src/kms/bucket_default_encryption_test.rs

@@ -37,7 +37,7 @@ async fn test_bucket_default_sse_s3_put_object() -> Result<(), Box<dyn std::erro
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -159,7 +159,7 @@ async fn test_bucket_default_sse_kms_put_object() -> Result<(), Box<dyn std::err
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -278,7 +278,7 @@ async fn test_bucket_default_sse_kms_multipart_crc32() -> Result<(), Box<dyn std
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -475,7 +475,7 @@ async fn test_explicit_encryption_overrides_bucket_default() -> Result<(), Box<d
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -570,7 +570,7 @@ async fn test_sse_kms_without_key_id_populates_default() -> Result<(), Box<dyn s
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;

crates/e2e_test/src/kms/common.rs

@@ -189,69 +189,34 @@ pub async fn wait_for_kms_ready(
     access_key: &str,
     secret_key: &str,
 ) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
-    wait_for_kms_ready_with_timeout(base_url, access_key, secret_key, Duration::from_secs(5)).await
-}
-
-async fn wait_for_kms_ready_with_timeout(
-    base_url: &str,
-    access_key: &str,
-    secret_key: &str,
-    total_deadline: Duration,
-) -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
+    let total_deadline = Duration::from_secs(5);
     let start = tokio::time::Instant::now();
-    let deadline = start + total_deadline;
     let mut backoff = Duration::from_millis(200);
     let max_backoff = Duration::from_secs(1);
+    let mut first_attempt = true;
 
     loop {
-        match tokio::time::timeout_at(deadline, get_kms_status(base_url, access_key, secret_key)).await {
-            Ok(Ok(status)) => {
+        if !first_attempt {
+            if start.elapsed() >= total_deadline {
+                return Err("KMS failed to become ready within 5 seconds".into());
+            }
+            sleep(backoff).await;
+            backoff = (backoff * 2).min(max_backoff);
+        }
+        first_attempt = false;
+
+        match get_kms_status(base_url, access_key, secret_key).await {
+            Ok(status) => {
                 info!("KMS is ready (status: {})", status);
                 return Ok(());
             }
-            Ok(Err(e)) => {
-                let elapsed_ms = u64::try_from(start.elapsed().as_millis()).unwrap_or(u64::MAX);
-                warn!(error = %e, elapsed_ms, "KMS not ready yet, retrying…");
+            Err(e) => {
+                if start.elapsed() >= total_deadline {
+                    return Err(format!("KMS did not become ready within 5 s: last error: {e}").into());
+                }
+                warn!(error = %e, elapsed_ms = start.elapsed().as_millis() as u64, "KMS not ready yet, retrying…");
             }
-            Err(_) => return Err(format!("KMS failed to become ready within {} ms", total_deadline.as_millis()).into()),
         }
-
-        let now = tokio::time::Instant::now();
-        if now >= deadline {
-            return Err(format!("KMS failed to become ready within {} ms", total_deadline.as_millis()).into());
-        }
-        sleep((now + backoff).min(deadline) - now).await;
-        backoff = (backoff * 2).min(max_backoff);
-    }
-}
-
-#[cfg(test)]
-mod readiness_tests {
-    use super::wait_for_kms_ready_with_timeout;
-    use std::time::Duration;
-    use tokio::io::AsyncReadExt;
-    use tokio::net::TcpListener;
-
-    #[tokio::test]
-    async fn kms_readiness_deadline_covers_a_stalled_status_request() {
-        let listener = TcpListener::bind("127.0.0.1:0").await.expect("bind readiness test server");
-        let address = listener.local_addr().expect("read readiness test server address");
-        let server = tokio::spawn(async move {
-            let (mut socket, _) = listener.accept().await.expect("accept readiness request");
-            let mut request = [0_u8; 1024];
-            let _ = socket.read(&mut request).await.expect("read readiness request");
-            std::future::pending::<()>().await;
-        });
-
-        let result = tokio::time::timeout(
-            Duration::from_secs(1),
-            wait_for_kms_ready_with_timeout(&format!("http://{address}"), "access-key", "secret-key", Duration::from_millis(50)),
-        )
-        .await
-        .expect("readiness helper must enforce its own deadline");
-
-        server.abort();
-        assert!(result.is_err(), "a stalled status request must not outlive the readiness deadline");
     }
 }
 

crates/e2e_test/src/kms/copy_object_self_copy_sse_test.rs

@@ -61,7 +61,7 @@ async fn test_metadata_replace_self_copy_of_sse_object_stays_decryptable() {
         )
         .await
         .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let client = kms_env.base_env.create_s3_client();
     // Deliberately an UNVERSIONED bucket: that is the branch where the store layer can service
@@ -160,7 +160,7 @@ async fn test_metadata_replace_self_copy_dropping_sse_rewrites_plaintext() {
         )
         .await
         .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let client = kms_env.base_env.create_s3_client();
     // Unversioned, and deliberately WITHOUT a bucket default-encryption rule, so the copy below
@@ -256,7 +256,7 @@ async fn test_metadata_replace_self_copy_under_bucket_default_sse_stays_decrypta
         )
         .await
         .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let client = kms_env.base_env.create_s3_client();
     let bucket = "copy-object-self-copy-bucket-default-sse-test";

crates/e2e_test/src/kms/copy_object_version_restore_sse_test.rs

@@ -56,7 +56,7 @@ async fn test_self_copy_of_historical_sse_s3_version_is_readable() {
         )
         .await
         .expect("failed to start RustFS with local KMS");
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let client = kms_env.base_env.create_s3_client();
     let bucket = "copy-object-version-restore-sse-test";

crates/e2e_test/src/kms/encryption_metadata_test.rs

@@ -87,7 +87,7 @@ async fn test_head_reports_managed_metadata_for_sse_s3() -> Result<(), Box<dyn s
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -147,7 +147,7 @@ async fn test_head_reports_managed_metadata_for_sse_kms_and_copy() -> Result<(),
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -250,7 +250,7 @@ async fn test_multipart_upload_writes_encrypted_data() -> Result<(), Box<dyn std
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;

crates/e2e_test/src/kms/kms_comprehensive_test.rs

@@ -24,6 +24,7 @@ use super::common::{
     test_sse_kms_encryption, test_sse_s3_encryption,
 };
 use crate::common::{TEST_BUCKET, init_logging};
+use tokio::time::{Duration, sleep};
 use tracing::info;
 
 /// Comprehensive test: Full KMS workflow with all encryption types
@@ -34,7 +35,7 @@ async fn test_comprehensive_kms_full_workflow() -> Result<(), Box<dyn std::error
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -102,7 +103,7 @@ async fn test_comprehensive_stress_test() -> Result<(), Box<dyn std::error::Erro
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -136,7 +137,7 @@ async fn test_comprehensive_key_isolation() -> Result<(), Box<dyn std::error::Er
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -207,7 +208,7 @@ async fn test_comprehensive_concurrent_operations() -> Result<(), Box<dyn std::e
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -252,7 +253,7 @@ async fn test_comprehensive_performance_benchmark() -> Result<(), Box<dyn std::e
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;

crates/e2e_test/src/kms/kms_edge_cases_test.rs

@@ -44,7 +44,7 @@ async fn test_kms_zero_byte_file_encryption() -> Result<(), Box<dyn std::error::
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -117,7 +117,7 @@ async fn test_kms_single_byte_file_encryption() -> Result<(), Box<dyn std::error
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -209,7 +209,7 @@ async fn test_kms_multipart_boundary_conditions() -> Result<(), Box<dyn std::err
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -284,7 +284,7 @@ async fn test_kms_invalid_key_scenarios() -> Result<(), Box<dyn std::error::Erro
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -371,7 +371,7 @@ async fn test_kms_concurrent_encryption() -> Result<(), Box<dyn std::error::Erro
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = Arc::new(kms_env.base_env.create_s3_client());
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -478,7 +478,7 @@ async fn test_kms_key_validation_security() -> Result<(), Box<dyn std::error::Er
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;

crates/e2e_test/src/kms/kms_fault_recovery_test.rs

@@ -37,7 +37,7 @@ async fn test_kms_key_directory_unavailable() -> Result<(), Box<dyn std::error::
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -127,7 +127,7 @@ async fn test_kms_corrupted_key_files() -> Result<(), Box<dyn std::error::Error
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -218,7 +218,7 @@ async fn test_kms_multipart_upload_interruption() -> Result<(), Box<dyn std::err
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -401,7 +401,7 @@ async fn test_kms_resource_constraints() -> Result<(), Box<dyn std::error::Error
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;

crates/e2e_test/src/kms/kms_local_test.rs

@@ -46,7 +46,7 @@ async fn test_local_kms_end_to_end() -> Result<(), Box<dyn std::error::Error + S
         .expect("Failed to start RustFS with Local KMS");
 
     // Wait a moment for RustFS to fully start up and initialize KMS
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);
 
@@ -127,7 +127,7 @@ async fn test_local_kms_key_isolation() {
         .expect("Failed to start RustFS with Local KMS");
 
     // Wait a moment for RustFS to fully start up and initialize KMS
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);
 
@@ -227,7 +227,7 @@ async fn test_local_kms_large_file() {
         .expect("Failed to start RustFS with Local KMS");
 
     // Wait a moment for RustFS to fully start up and initialize KMS
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);
 
@@ -309,7 +309,7 @@ async fn test_local_kms_multipart_upload() {
         .expect("Failed to start RustFS with Local KMS");
 
     // Wait for KMS initialization
-    kms_env.wait_for_kms_ready().await.expect("KMS ready");
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     info!("RustFS started with KMS auto-configuration, default_key_id: {}", default_key_id);
 

crates/e2e_test/src/kms/kms_vault_test.rs

@@ -19,6 +19,7 @@
 //! multipart upload behaviour.
 
 use crate::common::{TEST_BUCKET, init_logging};
+use tokio::time::{Duration, sleep};
 use tracing::{error, info};
 
 use super::common::{
@@ -44,8 +45,8 @@ impl VaultKmsTestContext {
 
         start_kms(&env.base_env.url, &env.base_env.access_key, &env.base_env.secret_key).await?;
 
-        // Wait for KMS to finish initialising.
-        super::common::wait_for_kms_ready(&env.base_env.url, &env.base_env.access_key, &env.base_env.secret_key).await?;
+        // Allow Vault to finish initialising token auth and transit engine.
+        sleep(Duration::from_secs(2)).await;
 
         Ok(Self { env })
     }

crates/e2e_test/src/kms/multipart_encryption_test.rs

@@ -33,7 +33,7 @@ async fn test_step1_basic_single_file_encryption() -> Result<(), Box<dyn std::er
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -89,7 +89,7 @@ async fn test_step2_basic_multipart_upload_without_encryption() -> Result<(), Bo
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -187,7 +187,7 @@ async fn test_step3_multipart_upload_with_sse_s3() -> Result<(), Box<dyn std::er
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -310,7 +310,7 @@ async fn test_step4_large_multipart_upload_with_encryption() -> Result<(), Box<d
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;
@@ -435,7 +435,7 @@ async fn test_step5_all_encryption_types_multipart() -> Result<(), Box<dyn std::
 
     let mut kms_env = LocalKMSTestEnvironment::new().await?;
     let _default_key_id = kms_env.start_rustfs_for_local_kms().await?;
-    kms_env.wait_for_kms_ready().await?;
+    tokio::time::sleep(tokio::time::Duration::from_secs(3)).await;
 
     let s3_client = kms_env.base_env.create_s3_client();
     kms_env.base_env.create_test_bucket(TEST_BUCKET).await?;

crates/ecstore/src/disk/os.rs

@@ -315,7 +315,7 @@ pub async fn fsync_dir(dir: impl AsRef<Path>) -> io::Result<()> {
     #[cfg(unix)]
     {
         let dir = dir.as_ref().to_path_buf();
-        tokio::task::spawn_blocking(move || fsync_dir_std(dir)).await?
+        fsync_spawn_blocking(move || fsync_dir_std(dir)).await?
     }
 
     #[cfg(not(unix))]
@@ -683,7 +683,7 @@ async fn fsync_open_dst_dir_group(group: &DstDirFsyncGroup) -> io::Result<()> {
     #[cfg(test)]
     let dir = group.dir.clone();
     let dir_file = group.dir_file.clone();
-    tokio::task::spawn_blocking(move || {
+    fsync_spawn_blocking(move || {
         #[cfg(test)]
         {
             if let Some(kind) = fsync_dir_recorder::take_grouped_failure(&dir) {
@@ -1080,6 +1080,44 @@ const TEST_GLOBAL_FILE_SYNCS: usize = 64;
 
 static FILE_SYNC_PERMITS: LazyLock<Semaphore> = LazyLock::new(|| Semaphore::new(global_file_sync_limit()));
 static DISK_FILE_SYNC_LIMITERS: LazyLock<Mutex<HashMap<PathBuf, Weak<Semaphore>>>> = LazyLock::new(|| Mutex::new(HashMap::new()));
+
+/// Dedicated tokio runtime for fsync/fdatasync blocking operations. When
+/// configured with >1 threads, isolates device-bound fsync from the main
+/// blocking pool so reads (pread/stat/open) are not starved. `None` means
+/// fall back to the main runtime (zero behavior change).
+static FSYNC_RUNTIME: LazyLock<Option<tokio::runtime::Runtime>> = LazyLock::new(|| {
+    let threads =
+        rustfs_utils::get_env_usize(rustfs_config::ENV_FSYNC_BLOCKING_THREADS, rustfs_config::DEFAULT_FSYNC_BLOCKING_THREADS);
+    if threads <= 1 {
+        return None;
+    }
+    let mut builder = tokio::runtime::Builder::new_multi_thread();
+    builder
+        .worker_threads(num_cpus::get().min(8))
+        .max_blocking_threads(threads)
+        .thread_name("rustfs-fsync")
+        .thread_stack_size(512 * 1024)
+        .enable_all();
+    match builder.build() {
+        Ok(rt) => {
+            tracing::info!(threads, "fsync dedicated blocking pool enabled");
+            Some(rt)
+        }
+        Err(err) => {
+            tracing::warn!(%err, "failed to build fsync runtime, falling back to main pool");
+            None
+        }
+    }
+});
+
+/// Spawn a blocking task on the fsync-dedicated runtime if configured,
+/// otherwise fall back to the main tokio blocking pool.
+fn fsync_spawn_blocking<T: Send + 'static>(f: impl FnOnce() -> T + Send + 'static) -> tokio::task::JoinHandle<T> {
+    match FSYNC_RUNTIME.as_ref() {
+        Some(rt) => rt.spawn_blocking(f),
+        None => tokio::task::spawn_blocking(f),
+    }
+}
 static DISK_VOLUME_MUTATION_LOCKS: LazyLock<Mutex<HashMap<PathBuf, Weak<RwLock<()>>>>> =
     LazyLock::new(|| Mutex::new(HashMap::new()));
 type NamespaceMutationLock = AsyncMutex<()>;
@@ -1217,7 +1255,7 @@ where
     F: FnOnce() -> io::Result<T> + Send + 'static,
 {
     let (disk_permit, global_permit) = acquire_file_sync_permits(disk_permits).await?;
-    let result = tokio::task::spawn_blocking(move || {
+    let result = fsync_spawn_blocking(move || {
         let _disk_permit = disk_permit;
         work()
     })
@@ -2146,7 +2184,7 @@ async fn run_blocking_namespace_file_sync_operation_with_global<T: Send + 'stati
         wait_started,
     );
     let disk_permit = admission.disk_permit.clone();
-    let result = tokio::task::spawn_blocking(move || {
+    let result = fsync_spawn_blocking(move || {
         let _lease = lease;
         let _disk_permit = disk_permit;
         operation()

crates/ecstore/src/store/heal.rs

@@ -297,10 +297,16 @@ impl ECStore {
 #[cfg(test)]
 mod tests {
     use super::*;
+    use crate::bucket::metadata_sys;
     use crate::core::pools::{PoolDecommissionInfo, PoolStatus};
-    use crate::disk::{DiskOption, format::FormatV3, new_disk};
-    use crate::layout::endpoints::{Endpoints, PoolEndpoints};
+    use crate::disk::{DeleteOptions, DiskOption, format::FormatV3, new_disk};
+    use crate::layout::endpoints::{EndpointServerPools, Endpoints, PoolEndpoints};
+    use crate::runtime::instance::InstanceContext;
+    use crate::storage_api_contracts::bucket::{BucketOperations, MakeBucketOptions};
+    use crate::storage_api_contracts::object::{ObjectIO as _, ObjectOperations};
     use crate::store::init_format::{load_format_erasure, save_format_file};
+    use crate::store::init_local_disks_with_instance_ctx;
+    use tokio_util::sync::CancellationToken;
 
     async fn minimal_heal_pool(pool_idx: usize) -> Arc<Sets> {
         let format = FormatV3::new(1, 1);
@@ -347,6 +353,51 @@ mod tests {
         }
     }
 
+    async fn multi_pool_heal_store() -> (tempfile::TempDir, Arc<ECStore>, CancellationToken) {
+        let temp_dir = tempfile::tempdir().expect("multi-pool heal test directory should be created");
+        let mut pool_endpoints = Vec::new();
+        for pool_index in 0..2 {
+            let mut endpoints = Vec::new();
+            for disk_index in 0..4 {
+                let disk_path = temp_dir.path().join(format!("pool{pool_index}-disk{disk_index}"));
+                tokio::fs::create_dir_all(&disk_path)
+                    .await
+                    .expect("multi-pool heal test disk should be created");
+                let mut endpoint = Endpoint::try_from(disk_path.to_str().expect("disk path should be utf8"))
+                    .expect("test endpoint should parse");
+                endpoint.set_pool_index(pool_index);
+                endpoint.set_set_index(0);
+                endpoint.set_disk_index(disk_index);
+                endpoints.push(endpoint);
+            }
+            pool_endpoints.push(PoolEndpoints {
+                legacy: false,
+                set_count: 1,
+                drives_per_set: 4,
+                endpoints: Endpoints::from(endpoints),
+                cmd_line: format!("heal-owner-pool-{pool_index}"),
+                platform: "test".to_string(),
+            });
+        }
+
+        let endpoint_pools = EndpointServerPools::from(pool_endpoints);
+        let instance_ctx = Arc::new(InstanceContext::new());
+        init_local_disks_with_instance_ctx(&instance_ctx, endpoint_pools.clone())
+            .await
+            .expect("multi-pool local disks should initialize");
+        let shutdown = CancellationToken::new();
+        let store = ECStore::new_with_instance_ctx(
+            "127.0.0.1:0".parse().expect("test address should parse"),
+            endpoint_pools,
+            shutdown.clone(),
+            instance_ctx,
+        )
+        .await
+        .expect("multi-pool test store should initialize");
+        metadata_sys::init_bucket_metadata_sys(store.clone(), Vec::new()).await;
+        (temp_dir, store, shutdown)
+    }
+
     #[tokio::test]
     async fn heal_object_pool_scope_selects_only_requested_pool() {
         let store = minimal_heal_store().await;
@@ -506,6 +557,204 @@ mod tests {
         }
     }
 
+    #[tokio::test]
+    #[serial_test::serial]
+    async fn unscoped_heal_object_suspended_owner_semantics() {
+        let (_temp_dir, store, shutdown) = multi_pool_heal_store().await;
+        let bucket = format!("heal-owner-{}", Uuid::new_v4().simple());
+        let active_object = "active-owner";
+        let suspended_only_object = "suspended-only";
+        let duplicate_object = "duplicate-owner";
+        let marker_object = "marker-owner";
+        let quorum_object = "quorum-owner";
+        store
+            .make_bucket(&bucket, &MakeBucketOptions::default())
+            .await
+            .expect("bucket should be created in all pools");
+
+        let mut active_reader = PutObjReader::from_vec(b"active owner".to_vec());
+        store.pools[0]
+            .put_object(&bucket, active_object, &mut active_reader, &ObjectOptions::default())
+            .await
+            .expect("active owner object should be written");
+        let active_disks = store.pools[0].disk_set[0].disks.read().await.clone();
+        let missing_active_disk = active_disks[0].clone().expect("active disk should be online");
+        missing_active_disk
+            .delete(
+                &bucket,
+                active_object,
+                DeleteOptions {
+                    recursive: true,
+                    immediate: true,
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("active owner shard should be removed for repair");
+        assert!(
+            missing_active_disk.read_xl(&bucket, active_object, false).await.is_err(),
+            "the active owner fixture must start with one missing metadata copy"
+        );
+
+        let mut suspended_reader = PutObjReader::from_vec(b"suspended owner".to_vec());
+        store.pools[1]
+            .put_object(&bucket, suspended_only_object, &mut suspended_reader, &ObjectOptions::default())
+            .await
+            .expect("suspended owner object should be written");
+        for (pool_index, mod_time) in [1_i64, 2_i64].into_iter().enumerate() {
+            let mut duplicate_reader = PutObjReader::from_vec(format!("duplicate-pool-{pool_index}").into_bytes());
+            store.pools[pool_index]
+                .put_object(
+                    &bucket,
+                    duplicate_object,
+                    &mut duplicate_reader,
+                    &ObjectOptions {
+                        mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(mod_time)),
+                        ..Default::default()
+                    },
+                )
+                .await
+                .expect("duplicate owner object should be written");
+        }
+        let history_version = Uuid::new_v4();
+        let mut history_reader = PutObjReader::from_vec(b"marker history".to_vec());
+        store.pools[0]
+            .put_object(
+                &bucket,
+                marker_object,
+                &mut history_reader,
+                &ObjectOptions {
+                    versioned: true,
+                    version_id: Some(history_version.to_string()),
+                    mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(1)),
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("versioned marker history should be written");
+        store.pools[0]
+            .delete_object(
+                &bucket,
+                marker_object,
+                ObjectOptions {
+                    versioned: true,
+                    mod_time: Some(OffsetDateTime::UNIX_EPOCH + time::Duration::seconds(2)),
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("delete marker should be written");
+        let mut quorum_reader = PutObjReader::from_vec(b"quorum boundary".to_vec());
+        store.pools[0]
+            .put_object(&bucket, quorum_object, &mut quorum_reader, &ObjectOptions::default())
+            .await
+            .expect("quorum boundary object should be written");
+        {
+            let mut pool_meta = store.pool_meta.write().await;
+            let mut next = PoolMeta::new(&store.pools, &pool_meta);
+            next.pools[1].decommission = Some(PoolDecommissionInfo {
+                start_time: Some(OffsetDateTime::UNIX_EPOCH),
+                ..Default::default()
+            });
+            *pool_meta = next;
+        }
+
+        let (_, duplicate_owner) = store
+            .get_latest_object_info_with_idx(&bucket, duplicate_object, &ObjectOptions::default())
+            .await
+            .expect("duplicate owner should resolve");
+        assert_eq!(duplicate_owner, 1, "latest duplicate must win when all pools are eligible");
+        let (_, active_duplicate_owner) = store
+            .get_latest_object_info_with_idx(
+                &bucket,
+                duplicate_object,
+                &ObjectOptions {
+                    skip_decommissioned: true,
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("active duplicate owner should resolve");
+        assert_eq!(
+            active_duplicate_owner, 0,
+            "suspended duplicate must be excluded from active owner selection"
+        );
+        let (marker_info, marker_owner) = store
+            .get_latest_object_info_with_idx(
+                &bucket,
+                marker_object,
+                &ObjectOptions {
+                    skip_decommissioned: true,
+                    versioned: true,
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("latest delete marker should resolve");
+        assert_eq!(marker_owner, 0);
+        assert!(marker_info.delete_marker, "latest version must preserve delete-marker semantics");
+
+        let (active_result, active_err) = store
+            .handle_heal_object(&bucket, active_object, "", &HealOpts::default())
+            .await
+            .expect("unscoped active-owner heal should complete");
+        assert_eq!(active_result.object, active_object);
+        assert!(active_err.is_none(), "active owner must be selected even with a suspended pool");
+        assert!(
+            missing_active_disk.read_xl(&bucket, active_object, false).await.is_ok(),
+            "active owner heal must write the missing disk metadata: result={active_result:?}, err={active_err:?}"
+        );
+        assert!(
+            store.pools[1]
+                .get_object_info(&bucket, active_object, &ObjectOptions::default())
+                .await
+                .is_err(),
+            "the suspended pool must not be written for an active-owner object"
+        );
+
+        let (suspended_result, suspended_err) = store
+            .handle_heal_object(&bucket, suspended_only_object, "", &HealOpts::default())
+            .await
+            .expect("unscoped suspended-only heal should return a terminal result");
+        assert!(suspended_result.object.is_empty());
+        assert!(matches!(suspended_err, Some(Error::FileNotFound)));
+        assert!(
+            store.pools[1]
+                .get_object_info(&bucket, suspended_only_object, &ObjectOptions::default())
+                .await
+                .is_ok(),
+            "suspended-only data must remain untouched when unscoped heal reports absent"
+        );
+
+        let (_, explicit_err) = store
+            .handle_heal_object(
+                &bucket,
+                suspended_only_object,
+                "",
+                &HealOpts {
+                    pool: Some(1),
+                    ..Default::default()
+                },
+            )
+            .await
+            .expect("explicit suspended-owner heal should return a mapped error");
+        assert!(matches!(explicit_err, Some(Error::SlowDown)));
+
+        let original_quorum_disks = store.pools[0].disk_set[0].disks.read().await.clone();
+        let surviving_quorum_disk = original_quorum_disks[3].clone();
+        *store.pools[0].disk_set[0].disks.write().await = vec![None, None, None, surviving_quorum_disk];
+        let (_, quorum_err) = store
+            .handle_heal_object(&bucket, quorum_object, "", &HealOpts::default())
+            .await
+            .expect("quorum boundary heal should return a mapped result");
+        *store.pools[0].disk_set[0].disks.write().await = original_quorum_disks;
+        assert!(
+            matches!(quorum_err, Some(Error::ErasureReadQuorum)),
+            "quorum-boundary heal must preserve quorum error, got {quorum_err:?}"
+        );
+        shutdown.cancel();
+    }
+
     #[tokio::test]
     async fn handle_heal_format_continues_after_a_pool_error() {
         let canonical_format = FormatV3::new(1, 3);

scripts/check_test_wiring.py

@@ -206,6 +206,13 @@ def check_runner_selection(root: Path) -> list[str]:
     return errors
 
 
+def check_s3_tests_runner(root: Path) -> list[str]:
+    runner = (root / "scripts/s3-tests/run.sh").read_text()
+    if "--showlocals" in runner:
+        return ["scripts/s3-tests/run.sh: pytest failure diagnostics must not dump local values"]
+    return []
+
+
 def profile_selection(root: Path, profile: str) -> str:
     if not re.fullmatch(r"e2e-[a-z0-9-]+", profile):
         raise ValueError(f"invalid e2e profile name: {profile}")
@@ -272,6 +279,7 @@ def validate(root: Path) -> list[str]:
     errors.extend(check_e2e_modules(root))
     errors.extend(check_fuzz_targets(root))
     errors.extend(check_runner_selection(root))
+    errors.extend(check_s3_tests_runner(root))
     errors.extend(check_profile_definitions(root))
     return errors
 
@@ -341,6 +349,23 @@ class SelfTests(unittest.TestCase):
             )
             self.assertEqual(len(check_fuzz_targets(root)), 1)
 
+    def test_s3_runner_rejects_unbounded_failure_locals(self) -> None:
+        with tempfile.TemporaryDirectory() as tmp:
+            root = Path(tmp)
+            runner = root / "scripts/s3-tests/run.sh"
+            runner.parent.mkdir(parents=True)
+            runner.write_text("tox -- -vv -ra --tb=long\n")
+            self.assertEqual(check_s3_tests_runner(root), [])
+            runner.write_text("tox -- -vv -ra --showlocals --tb=long\n")
+            self.assertEqual(len(check_s3_tests_runner(root)), 1)
+            with (
+                mock.patch(__name__ + ".check_e2e_modules", return_value=[]),
+                mock.patch(__name__ + ".check_fuzz_targets", return_value=[]),
+                mock.patch(__name__ + ".check_runner_selection", return_value=[]),
+                mock.patch(__name__ + ".check_profile_definitions", return_value=[]),
+            ):
+                self.assertEqual(len(validate(root)), 1)
+
     def test_profile_listing_enforces_selection(self) -> None:
         with tempfile.TemporaryDirectory() as tmp:
             root = Path(tmp)
@@ -411,7 +436,7 @@ def main() -> int:
         for error in errors:
             print(f"ERROR: {error}", file=sys.stderr)
         return 1
-    print("OK: e2e modules, runner selection, fuzz matrices, and profile guards are wired")
+    print("OK: e2e modules, runner selection, fuzz matrices, profiles, and bounded diagnostics are wired")
     return 0
 
 

scripts/s3-tests/run.sh

@@ -1028,10 +1028,11 @@ else
 fi
 
 # Run tests from s3tests/functional
+# Failure locals can contain multi-MiB request bodies; keep tracebacks without expanding local values.
 set +e
 S3TEST_CONF="${CONF_OUTPUT_PATH}" \
     tox -- \
-    -vv -ra --showlocals --tb=long \
+    -vv -ra --tb=long \
     --maxfail="${MAXFAIL}" \
     --timeout="${TEST_TIMEOUT}" \
     --junitxml="${ARTIFACTS_DIR}/junit.xml" \