perf(ecstore): use AHashMap for FileInfo metadata fields (#6738)

crates/filemeta/Cargo.toml

@@ -51,6 +51,9 @@ s3s = { workspace = true, features = ["minio"] }
 regex.workspace = true
 arc-swap.workspace = true
 
+# High-performance hashing
+ahash = { workspace = true, features = ["serde"] }
+
 [dev-dependencies]
 criterion = { workspace = true, features = ["html_reports"] }
 tempfile = { workspace = true }

crates/filemeta/src/fileinfo.rs

@@ -13,6 +13,7 @@
 // limitations under the License.
 
 use crate::{Error, ReplicationState, ReplicationStatusType, Result, TRANSITION_COMPLETE, VersionPurgeStatusType};
+use ahash::AHashMap;
 use bytes::Bytes;
 use rmp_serde::Serializer;
 use rustfs_utils::HashAlgorithm;
@@ -67,7 +68,7 @@ pub struct ObjectPartInfo {
     // Index holds the index of the part in the erasure coding
     pub index: Option<Bytes>,
     // Checksums holds checksums of the part
-    pub checksums: Option<HashMap<String, String>>,
+    pub checksums: Option<AHashMap<String, String>>,
     pub error: Option<String>,
 }
 
@@ -268,7 +269,7 @@ pub struct FileInfo {
     pub mode: Option<u32>,
     // WrittenByVersion is the unix time stamp of the version that created this version of the object
     pub written_by_version: Option<u64>,
-    pub metadata: HashMap<String, String>,
+    pub metadata: AHashMap<String, String>,
     pub parts: Vec<ObjectPartInfo>,
     pub erasure: ErasureInfo,
     // MarkDeleted marks this version as deleted
@@ -301,7 +302,7 @@ fn is_sensitive_metadata_key(key: &str) -> bool {
             .any(|prefix| starts_with_ignore_ascii_case(key, prefix))
 }
 
-struct RedactedMetadata<'a>(&'a HashMap<String, String>);
+struct RedactedMetadata<'a>(&'a AHashMap<String, String>);
 
 impl std::fmt::Debug for RedactedMetadata<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
@@ -425,7 +426,7 @@ struct FileInfoMapDef {
     size: i64,
     mode: Option<u32>,
     written_by_version: Option<u64>,
-    metadata: HashMap<String, String>,
+    metadata: AHashMap<String, String>,
     parts: Vec<ObjectPartInfo>,
     erasure: ErasureInfo,
     mark_deleted: bool,
@@ -1079,7 +1080,7 @@ impl FileInfo {
         mod_time: Option<OffsetDateTime>,
         actual_size: i64,
         index: Option<Bytes>,
-        checksums: Option<HashMap<String, String>>,
+        checksums: Option<AHashMap<String, String>>,
     ) {
         let part = ObjectPartInfo {
             etag,
@@ -1457,7 +1458,7 @@ pub fn parse_restore_obj_status(restore_hdr: &str) -> Result<RestoreStatus> {
     Err(Error::other(ERR_RESTORE_HDR_MALFORMED))
 }
 
-pub fn is_restored_object_on_disk(meta: &HashMap<String, String>) -> bool {
+pub fn is_restored_object_on_disk<S: std::hash::BuildHasher>(meta: &HashMap<String, String, S>) -> bool {
     if let Some(restore_hdr) = meta.get(X_AMZ_RESTORE.as_str())
         && let Ok(restore_status) = parse_restore_obj_status(restore_hdr)
     {
@@ -2135,7 +2136,10 @@ mod tests {
             -1_000_000i64..=1_000_000i64,
             optional_timestamp_strategy(),
             proptest::option::of(bytes_strategy(16)),
-            proptest::option::of(hash_map(small_string_strategy(), small_string_strategy(), 0..=3)),
+            proptest::option::of(
+                hash_map(small_string_strategy(), small_string_strategy(), 0..=3)
+                    .prop_map(|m| m.into_iter().collect::<AHashMap<String, String>>()),
+            ),
             proptest::option::of(small_string_strategy()),
         )
             .prop_map(|(etag, number, size, actual_size, mod_time, index, checksums, error)| ObjectPartInfo {
@@ -2170,7 +2174,8 @@ mod tests {
                 -1_000_000i64..=1_000_000i64,
                 proptest::option::of(any::<u32>()),
                 proptest::option::of(any::<u64>()),
-                hash_map(small_string_strategy(), small_string_strategy(), 0..=4),
+                hash_map(small_string_strategy(), small_string_strategy(), 0..=4)
+                    .prop_map(|m| m.into_iter().collect::<AHashMap<String, String>>()),
                 vec(object_part_info_strategy(), 0..=3),
                 erasure_info_strategy(),
                 any::<bool>(),

crates/filemeta/src/filemeta.rs

@@ -174,10 +174,10 @@ fn valid_target_delete_marker_version(arn: &str, version_id: &str) -> bool {
 /// included in the quorum hash, so such a divergence does surface — but as a
 /// quorum failure on an otherwise healthy object, which is not a state worth
 /// reaching. Merge the RPC metadata carrier instead, and only ever insert.
-fn persist_target_delete_marker_versions(
+fn persist_target_delete_marker_versions<S: std::hash::BuildHasher>(
     meta_sys: &mut HashMap<String, Vec<u8>>,
     versions: &HashMap<String, String>,
-    transport_metadata: &HashMap<String, String>,
+    transport_metadata: &HashMap<String, String, S>,
 ) {
     let mut bounded = BTreeMap::new();
     // A corrupt carrier means the dual internal prefixes disagreed. Do not merge

crates/filemeta/src/filemeta/inline_data.rs

@@ -181,7 +181,7 @@ mod tests {
             data_dir: Some(data_dir),
             size: 64 * 1024,
             mod_time: Some(OffsetDateTime::now_utc()),
-            metadata,
+            metadata: metadata.into_iter().collect(),
             erasure: ErasureInfo {
                 algorithm: ErasureAlgo::ReedSolomon.to_string(),
                 data_blocks: 4,

crates/filemeta/src/filemeta/version.rs

@@ -26,7 +26,7 @@ use super::msgp_decode::{
     PrependByteReader, prealloc_hint, read_exact_vec, read_nil_or_array_len, read_nil_or_map_len, skip_msgp_value,
 };
 use super::*;
-use crate::{ChecksumInfo, TransitionVersionState};
+use crate::{AHashMap, ChecksumInfo, TransitionVersionState};
 use rustfs_utils::HashAlgorithm;
 use rustfs_utils::http::{
     RUSTFS_INTERNAL_PREFIX, SUFFIX_CRC, SUFFIX_FREE_VERSION, SUFFIX_INLINE_DATA, SUFFIX_PART_CHECKSUMS, SUFFIX_PURGESTATUS,
@@ -377,7 +377,7 @@ impl<'a> DerivedInternalMetadata<'a> {
     }
 }
 
-struct UniquePartChecksums(HashMap<String, String>);
+struct UniquePartChecksums(AHashMap<String, String>);
 
 impl<'de> serde::Deserialize<'de> for UniquePartChecksums {
     fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
@@ -397,7 +397,7 @@ impl<'de> serde::Deserialize<'de> for UniquePartChecksums {
             where
                 A: serde::de::SeqAccess<'de>,
             {
-                let mut checksums = HashMap::with_capacity(seq.size_hint().unwrap_or_default());
+                let mut checksums = AHashMap::with_capacity(seq.size_hint().unwrap_or_default());
                 while let Some((key, value)) = seq.next_element::<(String, String)>()? {
                     if checksums.insert(key, value).is_some() {
                         return Err(serde::de::Error::custom("duplicate part checksum name"));
@@ -1477,7 +1477,7 @@ pub struct MetaObjectV1 {
     #[serde(rename = "Erasure")]
     pub erasure: MetaObjectV1Erasure,
     #[serde(rename = "Meta")]
-    pub meta: HashMap<String, String>,
+    pub meta: AHashMap<String, String>,
     #[serde(rename = "Parts")]
     pub parts: Vec<MetaObjectV1Part>,
     #[serde(rename = "VersionID")]
@@ -1543,7 +1543,7 @@ pub struct MetaObjectV1Part {
     #[serde(rename = "i")]
     pub index: Option<Bytes>,
     #[serde(rename = "crc")]
-    pub checksums: Option<HashMap<String, String>>,
+    pub checksums: Option<AHashMap<String, String>>,
     #[serde(rename = "err")]
     pub error: Option<String>,
 }
@@ -1887,7 +1887,7 @@ impl MetaObjectV1Part {
                 "i" => self.index = Some(Bytes::from(read_msgp_bin(rd)?)),
                 "crc" => {
                     let len = rmp::decode::read_map_len(rd)? as usize;
-                    let mut checksums = HashMap::with_capacity(prealloc_hint(len));
+                    let mut checksums = AHashMap::with_capacity(prealloc_hint(len));
                     for _ in 0..len {
                         checksums.insert(read_msgp_string(rd)?, read_msgp_string(rd)?);
                     }
@@ -2570,7 +2570,7 @@ impl MetaObject {
             Vec::new()
         };
 
-        let mut metadata = HashMap::with_capacity(self.meta_user.len() + self.meta_sys.len());
+        let mut metadata = AHashMap::with_capacity(self.meta_user.len() + self.meta_sys.len());
         for (k, v) in &self.meta_user {
             if k == AMZ_META_UNENCRYPTED_CONTENT_LENGTH || k == AMZ_META_UNENCRYPTED_CONTENT_MD5 {
                 continue;
@@ -2861,7 +2861,7 @@ impl From<FileInfo> for MetaObject {
     }
 }
 
-fn get_internal_replication_state(metadata: &HashMap<String, String>) -> Option<ReplicationState> {
+fn get_internal_replication_state<S: std::hash::BuildHasher>(metadata: &HashMap<String, String, S>) -> Option<ReplicationState> {
     let mut rs = ReplicationState::default();
     let mut has = false;
 
@@ -2942,7 +2942,7 @@ impl MetaDeleteMarker {
     }
 
     pub fn into_fileinfo(&self, volume: &str, path: &str, _all_parts: bool) -> Result<FileInfo> {
-        let metadata = self
+        let metadata: AHashMap<String, String> = self
             .meta_sys
             .clone()
             .into_iter()
@@ -5500,10 +5500,10 @@ mod tests {
     /// entirely, silently dropping the whole legacy body on re-marshal.
     #[test]
     fn legacy_version_body_round_trips_through_encode() {
-        let mut meta = HashMap::new();
+        let mut meta = AHashMap::new();
         meta.insert("content-type".to_string(), "application/octet-stream".to_string());
 
-        let mut crc = HashMap::new();
+        let mut crc = AHashMap::new();
         crc.insert("crc32c".to_string(), "deadbeef".to_string());
 
         let legacy = MetaObjectV1 {

crates/filemeta/src/lib.rs

@@ -22,6 +22,12 @@ mod replication;
 
 pub mod test_data;
 
+/// High-performance HashMap type alias using ahash instead of SipHash.
+pub type AHashMap<K, V> = ahash::AHashMap<K, V>;
+
+/// High-performance HashSet type alias using ahash.
+pub type AHashSet<K> = ahash::AHashSet<K>;
+
 pub use error::*;
 pub use fileinfo::*;
 pub use filemeta::*;

crates/filemeta/src/metacache.rs

@@ -1676,7 +1676,7 @@ mod tests {
     }
 
     fn metacache_entry_with_mod_time(mod_time: OffsetDateTime, etag: &str) -> MetaCacheEntry {
-        let mut metadata = HashMap::new();
+        let mut metadata = ahash::AHashMap::new();
         metadata.insert("etag".to_string(), etag.to_string());
 
         let mut meta = FileMeta::new();
@@ -1705,7 +1705,7 @@ mod tests {
         data_blocks: usize,
         parity_blocks: usize,
     ) -> MetaCacheEntry {
-        let mut metadata = HashMap::new();
+        let mut metadata = ahash::AHashMap::new();
         metadata.insert("etag".to_string(), etag.to_string());
 
         let mut fi = FileInfo::new("object", data_blocks, parity_blocks);
@@ -1730,7 +1730,7 @@ mod tests {
     fn metacache_entry_with_erasure_versions(versions: &[(OffsetDateTime, &str, usize, usize)]) -> MetaCacheEntry {
         let mut meta = FileMeta::new();
         for (idx, (mod_time, etag, data_blocks, parity_blocks)) in versions.iter().enumerate() {
-            let mut metadata = HashMap::new();
+            let mut metadata = ahash::AHashMap::new();
             metadata.insert("etag".to_string(), (*etag).to_string());
 
             let mut fi = FileInfo::new("object", *data_blocks, *parity_blocks);
@@ -1776,7 +1776,7 @@ mod tests {
     /// Build an entry holding a single object version with an explicit version id
     /// and mod_time, so a set of these can model DISJOINT per-disk version sets.
     fn metacache_entry_single_version(version_u128: u128, mod_time: OffsetDateTime, etag: &str) -> MetaCacheEntry {
-        let mut metadata = HashMap::new();
+        let mut metadata = ahash::AHashMap::new();
         metadata.insert("etag".to_string(), etag.to_string());
 
         let mut fi = FileInfo::new("object", 4, 2);