fix(heal): cleanup consumed MRF replay journals

Do not retain Accepted or Merged replay intents as startup anchors after they have been handed to the heal manager. Only refused or still-pending replay records keep the journal on disk until a successor snapshot can persist them.

This keeps successor snapshots limited to the pending queue, which lets successful replay remove both authoritative and legacy journal paths and restores the crash-boundary tests around successor flush.

Co-Authored-By: heihutu <heihutu@gmail.com>

Co-Authored-By: zhi22915 <qiuzgang@gmail.com>
(cherry picked from commit d5b8f49c9d)

Cargo.lock

@@ -9906,6 +9906,7 @@ dependencies = [
  "regex",
  "rmp",
  "rmp-serde",
+ "rustfs-config",
  "rustfs-utils",
  "s3s",
  "serde",

crates/config/src/constants/api.rs

@@ -100,5 +100,10 @@ pub const DEFAULT_API_MAX_CONNECTIONS: usize = 0;
 /// Example: RUSTFS_API_OBJECT_MAX_VERSIONS=50000
 pub const ENV_API_OBJECT_MAX_VERSIONS: &str = "RUSTFS_API_OBJECT_MAX_VERSIONS";
 
-/// Default for `RUSTFS_API_OBJECT_MAX_VERSIONS`.
-pub const DEFAULT_API_OBJECT_MAX_VERSIONS: u64 = 9_223_372_036_854_775_807;
+/// Default and maximum accepted value for `RUSTFS_API_OBJECT_MAX_VERSIONS`.
+#[cfg(target_pointer_width = "64")]
+pub const DEFAULT_API_OBJECT_MAX_VERSIONS: usize = 9_223_372_036_854_775_807;
+
+/// Default and maximum accepted value for `RUSTFS_API_OBJECT_MAX_VERSIONS`.
+#[cfg(not(target_pointer_width = "64"))]
+pub const DEFAULT_API_OBJECT_MAX_VERSIONS: usize = usize::MAX;

crates/filemeta/Cargo.toml

@@ -44,6 +44,7 @@ tokio = { workspace = true, features = ["io-util", "macros", "sync", "fs", "rt-m
 xxhash-rust = { workspace = true, features = ["xxh64", "xxh3"] }
 bytes = { workspace = true, features = ["serde"] }
 rustfs-utils = { workspace = true, features = ["hash", "http"] }
+rustfs-config = { workspace = true, features = ["constants"] }
 byteorder = { workspace = true }
 tracing.workspace = true
 thiserror.workspace = true

crates/filemeta/src/filemeta.rs

@@ -70,12 +70,8 @@ const _XL_FLAG_INLINE_DATA: u8 = 1 << 2;
 const META_DATA_READ_DEFAULT: usize = 4 << 10;
 const MSGP_UINT32_SIZE: usize = 5;
 
-/// Default max object versions per object, aligned with MinIO's default.
-pub const DEFAULT_OBJECT_MAX_VERSIONS: usize = if usize::BITS >= 64 {
-    9_223_372_036_854_775_807
-} else {
-    usize::MAX
-};
+/// Default max object versions per object.
+pub const DEFAULT_OBJECT_MAX_VERSIONS: usize = rustfs_config::DEFAULT_API_OBJECT_MAX_VERSIONS;
 
 static OBJECT_MAX_VERSIONS: AtomicUsize = AtomicUsize::new(DEFAULT_OBJECT_MAX_VERSIONS);
 

crates/log-analyzer/src/rules/seed/disk.rs

@@ -68,17 +68,14 @@ pub(super) fn rules() -> Vec<Rule> {
             )
         },
         Rule {
-            anchors: strings(["Storage inventory probe failed; current drive health is unknown"]),
+            anchors: strings(["reporting peer disks offline after consecutive storage_info failures"]),
             ..base(
                 "peer-disks-offline",
                 P2Degraded,
                 "disk",
-                "peer 存储清单探测失败",
-                any([
-                    contains("Storage inventory probe failed; current drive health is unknown"),
-                    contains("reporting peer disks offline after consecutive storage_info failures"),
-                ]),
-                "某 peer 的 storage_info 探测失败,当前磁盘健康状态未知。",
+                "peer 磁盘被整体判定离线",
+                contains("reporting peer disks offline after consecutive storage_info failures"),
+                "对某 peer 连续 storage_info 失败,判定其磁盘整体离线。",
                 "检查该 peer 节点存活与 RPC 端口可达。",
             )
         },

crates/log-analyzer/src/rules/seed/tests.rs

@@ -110,7 +110,7 @@ fn every_rule_has_a_positive_sample() {
         ("remote-peer-faulty", msg("Remote peer health check failed for node2: marking as faulty")),
         (
             "peer-disks-offline",
-            msg("Storage inventory probe failed; current drive health is unknown"),
+            msg("reporting peer disks offline after consecutive storage_info failures"),
         ),
         ("drive-faulty-error", msg("remote drive is faulty")),
         (
@@ -318,10 +318,6 @@ fn smoke_samples_hit_exact_rule_sets() {
         &["disk-marked-faulty"],
     );
     exact(&msg("erasure write quorum (required=8, achieved=5)"), &["ec-write-quorum"]);
-    exact(
-        &msg("reporting peer disks offline after consecutive storage_info failures"),
-        &["peer-disks-offline"],
-    );
     exact(
         &Sample {
             message: "Metacache listing quorum failed",

rustfs/src/startup_runtime.rs

@@ -17,7 +17,7 @@ use crate::{
     startup_runtime_hooks::{init_profiling_runtime, install_default_crypto_provider, log_startup_runtime_diagnostics},
     startup_tls_material::init_outbound_tls_material,
 };
-use rustfs_config::ENV_API_OBJECT_MAX_VERSIONS;
+use rustfs_config::{DEFAULT_API_OBJECT_MAX_VERSIONS, ENV_API_OBJECT_MAX_VERSIONS};
 use rustfs_utils::EnvParseOutcome;
 use std::io::{Error, Result};
 
@@ -32,13 +32,8 @@ pub(crate) async fn init_startup_runtime_foundation(config: &Config) -> Result<(
 
 fn init_object_max_versions_config() -> Result<()> {
     let limit = match rustfs_utils::get_env_parse_outcome::<u64>(ENV_API_OBJECT_MAX_VERSIONS) {
-        EnvParseOutcome::Absent => rustfs_filemeta::DEFAULT_OBJECT_MAX_VERSIONS,
-        EnvParseOutcome::Invalid => {
-            return Err(Error::other(format!(
-                "{ENV_API_OBJECT_MAX_VERSIONS} must be a positive integer no greater than {}",
-                usize::MAX
-            )));
-        }
+        EnvParseOutcome::Absent => DEFAULT_API_OBJECT_MAX_VERSIONS,
+        EnvParseOutcome::Invalid => return Err(object_max_versions_config_error()),
         EnvParseOutcome::Parsed(value) => object_max_versions_limit_from_u64(value)?,
     };
 
@@ -47,18 +42,20 @@ fn init_object_max_versions_config() -> Result<()> {
 
 fn object_max_versions_limit_from_u64(value: u64) -> Result<usize> {
     if value == 0 {
-        return Err(Error::other(format!(
-            "{ENV_API_OBJECT_MAX_VERSIONS} must be a positive integer no greater than {}",
-            usize::MAX
-        )));
+        return Err(object_max_versions_config_error());
     }
 
-    usize::try_from(value).map_err(|_| {
-        Error::other(format!(
-            "{ENV_API_OBJECT_MAX_VERSIONS} must be a positive integer no greater than {}",
-            usize::MAX
-        ))
-    })
+    let limit = usize::try_from(value).map_err(|_| object_max_versions_config_error())?;
+    if limit > DEFAULT_API_OBJECT_MAX_VERSIONS {
+        return Err(object_max_versions_config_error());
+    }
+    Ok(limit)
+}
+
+fn object_max_versions_config_error() -> Error {
+    Error::other(format!(
+        "{ENV_API_OBJECT_MAX_VERSIONS} must be a positive integer no greater than {DEFAULT_API_OBJECT_MAX_VERSIONS}"
+    ))
 }
 
 #[cfg(test)]