Merge branch 'main' into cxymds/fix-1934-orphan-discovery

SSE and compression wrap the payload so its length is unknown and
advertise HashReader::SIZE_PRESERVE_LAYER (-1). Every layer preserved
that sentinel except create_bitrot_writer, which clamped it to 0 before
calling DiskAPI::create_file. RemoteDisk forwards that size verbatim in
the put_file_stream query, so remote peers were told the body was empty.

Since the authenticated put-file trailer (#5868) the receiver used the
declared size to split body from trailer, turning the clamp into a fatal
"auth trailer has trailing data" failure for every SSE PUT on multi-node
deployments (rc.2). #6320 relaxed the receiver to only trust size > 0;
this change fixes the sender so the sentinel survives end to end and the
wire no longer conflates empty objects with unknown-length streams.

Refs #6331

CLAUDE.md

@@ -30,7 +30,8 @@ make build-docker BUILD_OS=ubuntu22.04
 - Crate membership: `Cargo.toml` `[workspace].members`
 - Architecture, layering, crate map: [ARCHITECTURE.md](ARCHITECTURE.md)
 - Migration guardrails & readiness contracts: [docs/architecture/](docs/architecture/README.md)
-- CI gates: `.github/workflows/ci.yml` (source of truth; never copy its steps into docs)
+- CI workflow steps: `.github/workflows/`; event, timeout, and required-status
+  matrix: [docs/testing/ci-gates.md](docs/testing/ci-gates.md)
 - Test-layer taxonomy, per-layer entry commands, serial/nextest rules, flake
   policy: [docs/testing/README.md](docs/testing/README.md)
 - Tier/ILM transition debugging (xl.meta inspection, versionId tracing):

CONTRIBUTING.md

@@ -70,6 +70,8 @@ make pre-pr
 
 > For the full test-layer taxonomy (unit / ecstore black-box / e2e / s3s-e2e / S3 compatibility / chaos / fuzz / bench), each layer's entry command, the naming conventions the migration gate depends on, and the serial/nextest rules, see [docs/testing/README.md](docs/testing/README.md).
 
+> For the event, timeout, required-status, and local reproduction matrix, see [docs/testing/ci-gates.md](docs/testing/ci-gates.md).
+
 ### 🔒 Automated Pre-commit Hooks
 #### What `make pre-commit` and `make pre-pr` actually run
 

Cargo.lock

@@ -1858,9 +1858,9 @@ dependencies = [
 
 [[package]]
 name = "cc"
-version = "1.4.3"
+version = "1.4.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "509591b7bcd67f4ef775afad7662703b4935daaa6ec0e5605cfb1090b32a2b6d"
+checksum = "0ad534f4357a5264cce5019c989cf66a4f0dc4e0d1b1d15f8aacec0ff7360273"
 dependencies = [
  "find-msvc-tools",
  "jobserver",
@@ -2522,12 +2522,6 @@ dependencies = [
  "subtle",
 ]
 
-[[package]]
-name = "cty"
-version = "0.2.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b365fabc795046672053e29c954733ec3b05e4be654ab130fe8f1f94d7051f35"
-
 [[package]]
 name = "curve25519-dalek"
 version = "4.1.3"
@@ -5988,15 +5982,6 @@ version = "0.2.16"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b6d2cec3eae94f9f509c767b45932f1ada8350c4bdb85af2fcab4a3c14807981"
 
-[[package]]
-name = "libmimalloc-sys"
-version = "0.1.49"
-source = "git+https://github.com/xonatius/mimalloc_rust.git?rev=6d4c41bb10c6d9da1d1b6f07b38c4cc051667f11#6d4c41bb10c6d9da1d1b6f07b38c4cc051667f11"
-dependencies = [
- "cc",
- "cty",
-]
-
 [[package]]
 name = "libredox"
 version = "0.1.20"
@@ -6397,14 +6382,6 @@ dependencies = [
  "synstructure 0.13.2",
 ]
 
-[[package]]
-name = "mimalloc"
-version = "0.1.52"
-source = "git+https://github.com/xonatius/mimalloc_rust.git?rev=6d4c41bb10c6d9da1d1b6f07b38c4cc051667f11#6d4c41bb10c6d9da1d1b6f07b38c4cc051667f11"
-dependencies = [
- "libmimalloc-sys",
-]
-
 [[package]]
 name = "mime"
 version = "0.3.17"
@@ -9162,13 +9139,11 @@ dependencies = [
  "insta",
  "jiff",
  "libc",
- "libmimalloc-sys",
  "libsystemd",
  "matchit 0.9.2",
  "md-5 0.11.0",
  "metrics",
  "metrics-util",
- "mimalloc",
  "mime_guess",
  "opentelemetry",
  "opentelemetry_sdk",
@@ -9204,6 +9179,8 @@ dependencies = [
  "rustfs-lock",
  "rustfs-log-analyzer",
  "rustfs-madmin",
+ "rustfs-mimalloc",
+ "rustfs-mimalloc-sys",
  "rustfs-notify",
  "rustfs-object-capacity",
  "rustfs-object-data-cache",
@@ -9875,6 +9852,24 @@ dependencies = [
  "tokio",
 ]
 
+[[package]]
+name = "rustfs-mimalloc"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a406f4aa07084301d485beec873af6dccc8e3f8762da244743df92038b1db1a6"
+dependencies = [
+ "rustfs-mimalloc-sys",
+]
+
+[[package]]
+name = "rustfs-mimalloc-sys"
+version = "0.5.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c3051b819175f58445d4c369a72f0ab88149f3885ba8bea2aff3be01f53fe7cd"
+dependencies = [
+ "cc",
+]
+
 [[package]]
 name = "rustfs-notify"
 version = "1.0.0-rc.3"

Cargo.toml

@@ -350,8 +350,8 @@ russh-sftp = "2.4.0"
 dav-server = "0.11.0"
 
 # Performance Analysis and Memory Profiling
-mimalloc = { version = "0.1.52", git = "https://github.com/xonatius/mimalloc_rust.git", rev = "6d4c41bb10c6d9da1d1b6f07b38c4cc051667f11" }
-libmimalloc-sys = { version = "0.1.49", git = "https://github.com/xonatius/mimalloc_rust.git", rev = "6d4c41bb10c6d9da1d1b6f07b38c4cc051667f11", features = ["extended"] }
+rustfs-mimalloc = { version = "0.5.0" }
+rustfs-mimalloc-sys = { version = "0.5.0" }
 hotpath = { version = "0.23.3", default-features = false }
 # Snapshot testing for output format regression detection
 insta = { version = "1.48" }

crates/ecstore/src/io_support/bitrot.rs

@@ -784,6 +784,24 @@ pub(crate) fn create_deferred_bitrot_reader_with_stripe_handle(
 ///
 /// # Returns
 /// A Result containing the BitrotWriterWrapper or an error
+/// Size hint handed to `DiskAPI::create_file` for a bitrot-wrapped shard.
+///
+/// A known length is grown by one checksum per shard so the on-disk file size
+/// matches what the bitrot writer emits. A negative length is the
+/// unknown-size sentinel (`HashReader::SIZE_PRESERVE_LAYER`, used by SSE and
+/// compression) and must be preserved: `RemoteDisk::create_file` forwards it
+/// in the `put_file_stream` query, and the receiver only treats `size > 0` as
+/// a fixed body length when locating the authenticated trailer. Clamping it
+/// to `0` would claim an empty body and misframe the stream. `0` stays `0`
+/// because a genuinely empty object still means an empty body.
+fn bitrot_create_file_size(length: i64, shard_size: usize, checksum_algo: &HashAlgorithm) -> i64 {
+    if length <= 0 {
+        return length;
+    }
+    let length = length as usize;
+    (length.div_ceil(shard_size) * checksum_algo.size() + length) as i64
+}
+
 pub async fn create_bitrot_writer(
     is_inline_buffer: bool,
     disk: Option<&DiskStore>,
@@ -796,12 +814,7 @@ pub async fn create_bitrot_writer(
     let writer = if is_inline_buffer {
         CustomWriter::new_inline_buffer()
     } else if let Some(disk) = disk {
-        let length = if length > 0 {
-            let length = length as usize;
-            (length.div_ceil(shard_size) * checksum_algo.size() + length) as i64
-        } else {
-            0
-        };
+        let length = bitrot_create_file_size(length, shard_size, &checksum_algo);
 
         let file = disk.create_file("", volume, path, length).await?;
         #[cfg(feature = "hotpath")]
@@ -820,6 +833,25 @@ mod tests {
     use rustfs_rio::ChunkReader;
     use std::collections::VecDeque;
 
+    #[test]
+    fn bitrot_create_file_size_grows_known_length_by_checksums() {
+        // 10 bytes over 4-byte shards = 3 shards, each followed by a 32-byte hash.
+        assert_eq!(bitrot_create_file_size(10, 4, &HashAlgorithm::HighwayHash256), 10 + 3 * 32);
+        assert_eq!(bitrot_create_file_size(10, 4, &HashAlgorithm::None), 10);
+    }
+
+    #[test]
+    fn bitrot_create_file_size_keeps_empty_and_unknown_distinct() {
+        assert_eq!(bitrot_create_file_size(0, 4, &HashAlgorithm::HighwayHash256), 0);
+        // SSE/compression streams advertise SIZE_PRESERVE_LAYER (-1); the remote
+        // put_file_stream receiver relies on a non-positive size to parse the auth
+        // trailer from the stream tail, so the sentinel must survive untouched.
+        assert_eq!(
+            bitrot_create_file_size(rustfs_rio::HashReader::SIZE_PRESERVE_LAYER, 4, &HashAlgorithm::HighwayHash256),
+            rustfs_rio::HashReader::SIZE_PRESERVE_LAYER
+        );
+    }
+
     struct TestChunkReader {
         chunks: VecDeque<Bytes>,
     }

crates/ecstore/src/set_disk/ops/object.rs

@@ -2124,26 +2124,13 @@ impl SetDisks {
 
             let put_object_size = known_put_object_storage_size(data.size());
             let shard_file_size_raw = erasure.shard_file_size(put_object_size);
-            let is_inline_buffer =
-                storage_class_config.should_inline(shard_file_size_raw, erasure.data_shards, opts.versioned);
+            let is_inline_buffer = storage_class_config.should_inline(shard_file_size_raw, erasure.data_shards, opts.versioned);
 
             let collect_stage_timing = rustfs_io_metrics::put_stage_metrics_enabled() || issue3031_diag_enabled();
             let shard_file_size = shard_file_size_raw;
             let shard_size = erasure.shard_size();
             let write_path = classify_put_write_path(is_inline_buffer, put_object_size, fi.erasure.block_size);
             let direct_inline_commit = matches!(write_path, SmallWritePath::Inline);
-            {
-                use std::io::Write;
-                let msg = format!(
-                    "INLINE_DEBUG: bucket={} obj={} size={} shard_fs={} ds={} bs={} inline={} direct={} path={} iblock={} ver={}\n",
-                    bucket, object, put_object_size, shard_file_size_raw, erasure.data_shards, fi.erasure.block_size,
-                    is_inline_buffer, direct_inline_commit, write_path.metric_label(), storage_class_config.inline_block(), opts.versioned
-                );
-                if let Ok(mut f) = std::fs::OpenOptions::new().create(true).append(true).open("/tmp/rustfs_inline_debug.log") {
-                    let _ = f.write_all(msg.as_bytes());
-                }
-                let _ = std::io::stderr().write_all(msg.as_bytes());
-            }
             rustfs_io_metrics::record_put_object_path(write_path.metric_label());
             let writer_setup_stage_start = collect_stage_timing.then(Instant::now);
             let (mut writers, errors) = if direct_inline_commit {

crates/filemeta/src/metacache.rs

@@ -21,6 +21,7 @@ use arc_swap::ArcSwapOption;
 use rmp::Marker;
 use serde::{Deserialize, Serialize};
 use std::cmp::Ordering;
+use std::collections::{HashMap, HashSet};
 use std::str::from_utf8;
 use std::{
     fmt::Debug,
@@ -37,8 +38,13 @@ use tokio::io::{AsyncRead, AsyncReadExt, AsyncWrite, AsyncWriteExt};
 use tokio::spawn;
 use tokio::sync::Mutex;
 use tracing::{debug, warn};
+use uuid::Uuid;
 
 const SLASH_SEPARATOR: &str = "/";
+pub const MAX_META_CACHE_HEAL_CANDIDATES: usize = 1024;
+/// Keep truncation continuations bounded while still giving the scanner a
+/// safe object-level retry for versions that did not fit in the candidate set.
+pub const MAX_META_CACHE_HEAL_TRUNCATED_OBJECTS: usize = 64;
 
 #[derive(Clone, Debug, Default)]
 pub struct MetadataResolutionParams {
@@ -66,6 +72,50 @@ pub struct MetaCacheEntry {
     pub reusable: bool,
 }
 
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
+pub enum MetaCacheHealCandidateKind {
+    Object,
+    DeleteMarker,
+    UnversionedObject,
+}
+
+#[derive(Clone, Debug, PartialEq, Eq, Hash)]
+pub struct MetaCacheHealCandidate {
+    pub object: String,
+    pub version_id: Option<Uuid>,
+    pub kind: MetaCacheHealCandidateKind,
+    /// Number of raw disk entries that carried this validated version.
+    pub replica_count: usize,
+}
+
+impl MetaCacheHealCandidate {
+    pub fn validated_version(&self) -> Option<Uuid> {
+        match self.kind {
+            MetaCacheHealCandidateKind::Object | MetaCacheHealCandidateKind::DeleteMarker => self.version_id,
+            MetaCacheHealCandidateKind::UnversionedObject => None,
+        }
+    }
+
+    pub fn is_unversioned(&self) -> bool {
+        self.kind == MetaCacheHealCandidateKind::UnversionedObject
+    }
+}
+
+#[derive(Clone, Debug, Default, PartialEq, Eq)]
+pub struct MetaCacheHealDiscovery {
+    pub candidates: Vec<MetaCacheHealCandidate>,
+    pub unverified_count: usize,
+    pub truncated: bool,
+    /// Object names whose validated version set exceeded the candidate cap.
+    /// The scanner retries these names without a version and with destructive
+    /// healing disabled; this is an explicit bounded continuation, not a
+    /// version claim.
+    pub truncated_objects: Vec<String>,
+    /// Validated candidates beyond the main cap, retained with exact version
+    /// identities so callers never fall back to a latest-version request.
+    pub truncated_candidates: Vec<MetaCacheHealCandidate>,
+}
+
 impl MetaCacheEntry {
     pub fn marshal_msg(&self) -> Result<Vec<u8>> {
         let mut wr = Vec::new();
@@ -370,6 +420,185 @@ impl MetaCacheEntries {
         })
     }
 
+    /// Discover validated object/delete-marker versions and safe unversioned
+    /// inspection candidates in the raw entries without applying read quorum.
+    /// This is intentionally separate from [`Self::resolve`]: a sub-quorum
+    /// version is a valid heal target even though it must not participate in
+    /// normal reads or writes.
+    ///
+    /// The validated list is bounded and deduplicated by object, version id,
+    /// and metadata kind; each candidate retains the number of raw disk
+    /// entries that carried it so callers can classify sub-quorum versions.
+    /// Entries whose xl.meta cannot be decoded are counted separately for
+    /// discovery accounting; they never become versionless destructive heal
+    /// requests and do not consume the validated quota. An
+    /// [`MetaCacheHealCandidateKind::UnversionedObject`] is always consumed by
+    /// a non-destructive scanner request.
+    pub fn discover_heal_candidates(&self, bucket: &str, max_candidates: usize) -> MetaCacheHealDiscovery {
+        let limit = max_candidates.min(MAX_META_CACHE_HEAL_CANDIDATES);
+        if limit == 0 || bucket.is_empty() {
+            return MetaCacheHealDiscovery::default();
+        }
+
+        let mut discovery = MetaCacheHealDiscovery {
+            candidates: Vec::<MetaCacheHealCandidate>::with_capacity(limit.min(self.0.len())),
+            unverified_count: 0,
+            truncated: false,
+            truncated_objects: Vec::with_capacity(MAX_META_CACHE_HEAL_TRUNCATED_OBJECTS.min(limit)),
+            truncated_candidates: Vec::new(),
+        };
+        let mut seen: HashMap<(String, Option<Uuid>, MetaCacheHealCandidateKind), usize> =
+            HashMap::with_capacity(limit.min(self.0.len()));
+
+        for entry in self.0.iter().flatten() {
+            if !valid_heal_candidate_name(bucket, entry) {
+                continue;
+            }
+
+            let meta = match FileMeta::load(&entry.metadata) {
+                Ok(meta) => meta,
+                Err(_) => {
+                    discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                    continue;
+                }
+            };
+            let mut entry_seen = HashSet::new();
+
+            for shallow in meta.versions {
+                let version = match shallow.parse_version_meta() {
+                    Ok(version) if version.valid() => version,
+                    Ok(_) | Err(_) => {
+                        discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                        continue;
+                    }
+                };
+                if version.free_version() {
+                    continue;
+                }
+
+                let payload_header = version.header();
+                if normalize_version_id(shallow.header.version_id) != normalize_version_id(payload_header.version_id)
+                    || shallow.header.version_type != payload_header.version_type
+                {
+                    discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                    continue;
+                }
+
+                let (kind, version_id) = match version.version_type {
+                    VersionType::Object
+                        if version.object.is_some() && version.delete_marker.is_none() && version.legacy_object.is_none() =>
+                    {
+                        match version.object.as_ref().and_then(|object| object.version_id) {
+                            Some(id) if !id.is_nil() => (MetaCacheHealCandidateKind::Object, Some(id)),
+                            Some(_) | None => (MetaCacheHealCandidateKind::UnversionedObject, None),
+                        }
+                    }
+                    VersionType::Delete
+                        if version.delete_marker.is_some() && version.object.is_none() && version.legacy_object.is_none() =>
+                    {
+                        let Some(id) = version.delete_marker.as_ref().and_then(|marker| marker.version_id) else {
+                            discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                            continue;
+                        };
+                        if id.is_nil() {
+                            discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                            continue;
+                        }
+                        (MetaCacheHealCandidateKind::DeleteMarker, Some(id))
+                    }
+                    VersionType::Legacy
+                        if version.legacy_object.is_some() && version.object.is_none() && version.delete_marker.is_none() =>
+                    {
+                        let Some(legacy) = version.legacy_object.as_ref() else {
+                            continue;
+                        };
+                        if legacy.version_id.is_empty() {
+                            (MetaCacheHealCandidateKind::UnversionedObject, None)
+                        } else {
+                            let Ok(id) = Uuid::parse_str(&legacy.version_id) else {
+                                discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                                continue;
+                            };
+                            if id.is_nil() {
+                                discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                                continue;
+                            }
+                            (MetaCacheHealCandidateKind::Object, Some(id))
+                        }
+                    }
+                    _ => {
+                        discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                        continue;
+                    }
+                };
+
+                if normalize_version_id(payload_header.version_id) != version_id {
+                    discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                    continue;
+                }
+
+                // `all_parts=true` is the trust-boundary check for versioned
+                // candidates. A null/legacy object may still need the old
+                // non-destructive inspection fallback when its part arrays
+                // are parseable but incomplete; never use that fallback for
+                // a candidate carrying a real version id.
+                let file_info = match version.clone().into_fileinfo(bucket, &entry.name, true) {
+                    Ok(file_info) => file_info,
+                    Err(_) if version_id.is_none() && matches!(kind, MetaCacheHealCandidateKind::UnversionedObject) => {
+                        discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                        match version.into_fileinfo(bucket, &entry.name, false) {
+                            Ok(file_info) => file_info,
+                            Err(_) => continue,
+                        }
+                    }
+                    Err(_) => {
+                        discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                        continue;
+                    }
+                };
+                if file_info.volume != bucket || file_info.name != entry.name {
+                    discovery.unverified_count = discovery.unverified_count.saturating_add(1);
+                    continue;
+                }
+
+                let candidate = MetaCacheHealCandidate {
+                    object: entry.name.clone(),
+                    version_id,
+                    kind,
+                    replica_count: 1,
+                };
+                let key = (candidate.object.clone(), candidate.version_id, candidate.kind.clone());
+                if entry_seen.contains(&key) {
+                    continue;
+                }
+                if let Some(index) = seen.get(&key).copied() {
+                    entry_seen.insert(key);
+                    discovery.candidates[index].replica_count = discovery.candidates[index].replica_count.saturating_add(1);
+                } else if discovery.candidates.len() >= limit {
+                    // Keep the validated candidate list bounded, but retain a
+                    // bounded object-level continuation so the scanner cannot
+                    // silently lose every version of a busy object.
+                    discovery.truncated = true;
+                    if discovery.truncated_objects.len() < MAX_META_CACHE_HEAL_TRUNCATED_OBJECTS
+                        && !discovery.truncated_objects.iter().any(|object| object == &candidate.object)
+                    {
+                        discovery.truncated_objects.push(candidate.object.clone());
+                    }
+                    if discovery.truncated_objects.iter().any(|object| object == &candidate.object) {
+                        discovery.truncated_candidates.push(candidate);
+                    }
+                    continue;
+                } else {
+                    entry_seen.insert(key.clone());
+                    seen.insert(key, discovery.candidates.len());
+                    discovery.candidates.push(candidate);
+                }
+            }
+        }
+
+        discovery
+    }
+
     fn resolve_inner(&self, mut params: MetadataResolutionParams, enforce_write_quorum: bool) -> Option<MetaCacheEntry> {
         if self.0.is_empty() {
             debug!(
@@ -546,6 +775,33 @@ impl MetaCacheEntries {
     }
 }
 
+fn valid_heal_candidate_name(bucket: &str, entry: &MetaCacheEntry) -> bool {
+    if bucket.is_empty()
+        || entry.name.is_empty()
+        || entry.is_dir()
+        || (cfg!(windows) && entry.name.contains('\\'))
+        || entry.name.chars().any(char::is_control)
+    {
+        return false;
+    }
+
+    // Validate raw key components without normalizing them. The scanner maps
+    // accepted keys to filesystem paths later, so dot components and empty
+    // internal components must be rejected before that boundary. A final
+    // empty component is retained for valid keys ending in '/'.
+    let mut components = entry.name.split('/').peekable();
+    while let Some(component) = components.next() {
+        if component == "." || component == ".." || (component.is_empty() && components.peek().is_some()) {
+            return false;
+        }
+    }
+    true
+}
+
+fn normalize_version_id(version_id: Option<Uuid>) -> Option<Uuid> {
+    version_id.filter(|id| !id.is_nil())
+}
+
 #[derive(Debug, Default)]
 pub struct MetaCacheEntriesSortedResult {
     pub entries: Option<MetaCacheEntriesSorted>,
@@ -991,7 +1247,7 @@ impl<T: Clone + Debug + Send + Sync + 'static> Cache<T> {
 mod tests {
     use super::*;
     use crate::test_data::create_real_xlmeta;
-    use crate::{FileMetaVersion, MetaDeleteMarker, TRANSITION_COMPLETE};
+    use crate::{FileMetaVersion, MetaDeleteMarker, MetaObjectV1, MetaObjectV1Erasure, MetaObjectV1Stat, TRANSITION_COMPLETE};
     use std::collections::HashMap;
     use std::io::Cursor;
     use std::sync::{
@@ -1592,6 +1848,381 @@ mod tests {
         );
     }
 
+    #[test]
+    fn discover_heal_candidates_keeps_sub_quorum_versions_and_deduplicates() {
+        let now = OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp");
+        let entries = MetaCacheEntries(vec![
+            Some(metacache_entry_single_version(1, now, "one")),
+            Some(metacache_entry_single_version(2, now, "two")),
+            Some(metacache_entry_single_version(2, now, "two")),
+            Some(metacache_entry_single_version(3, now, "three")),
+        ]);
+
+        let discovery = entries.discover_heal_candidates("bucket", 16);
+        let ids: std::collections::HashSet<Uuid> = discovery
+            .candidates
+            .iter()
+            .filter_map(|candidate| candidate.version_id)
+            .collect();
+        assert_eq!(
+            ids,
+            [Uuid::from_u128(1), Uuid::from_u128(2), Uuid::from_u128(3)]
+                .into_iter()
+                .collect()
+        );
+        assert_eq!(discovery.candidates.len(), 3, "duplicate tied versions must be emitted once");
+        assert_eq!(
+            discovery
+                .candidates
+                .iter()
+                .find(|candidate| candidate.version_id == Some(Uuid::from_u128(2)))
+                .expect("duplicate version should be discovered")
+                .replica_count,
+            2
+        );
+    }
+
+    #[test]
+    fn discover_heal_candidates_does_not_count_duplicate_versions_within_one_entry() {
+        let now = OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp");
+        let mut meta = FileMeta::load(&metacache_entry_single_version(1, now, "duplicate").metadata)
+            .expect("duplicate fixture should decode");
+        meta.versions.push(meta.versions[0].clone());
+        let entry = MetaCacheEntry {
+            name: "object".to_string(),
+            metadata: meta.marshal_msg().expect("duplicate metadata should marshal"),
+            cached: Some(meta),
+            reusable: false,
+        };
+
+        let discovery = MetaCacheEntries(vec![Some(entry)]).discover_heal_candidates("bucket", 16);
+        let candidate = discovery
+            .candidates
+            .iter()
+            .find(|candidate| candidate.version_id == Some(Uuid::from_u128(1)))
+            .expect("duplicate fixture should be discovered");
+        assert_eq!(candidate.replica_count, 1);
+    }
+
+    #[test]
+    fn discover_heal_candidates_covers_divergent_quorum_boundaries_n2_n4_n6() {
+        let now = OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp");
+
+        for (disk_count, quorum) in [(2usize, 1usize), (4, 2), (6, 3)] {
+            let target_id = Uuid::from_u128(0x1000 + disk_count as u128);
+            for target_replicas in [quorum.saturating_sub(1), quorum, quorum + 1] {
+                let entries = (0..disk_count)
+                    .map(|disk| {
+                        let version_id = if disk < target_replicas {
+                            target_id
+                        } else {
+                            Uuid::from_u128(0x2000 + disk as u128)
+                        };
+                        Some(metacache_entry_single_version(version_id.as_u128(), now, "divergent"))
+                    })
+                    .collect();
+                let discovery = MetaCacheEntries(entries).discover_heal_candidates("bucket", 32);
+                let target = discovery
+                    .candidates
+                    .iter()
+                    .find(|candidate| candidate.version_id == Some(target_id));
+                assert_eq!(target.is_some(), target_replicas > 0, "N={disk_count}, replicas={target_replicas}");
+                if let Some(target) = target {
+                    assert_eq!(target.replica_count, target_replicas);
+                }
+            }
+        }
+    }
+
+    #[test]
+    fn discover_heal_candidates_separates_delete_markers_and_preserves_unversioned_objects() {
+        let mut marker_meta = FileMeta::new();
+        marker_meta
+            .add_version(FileInfo {
+                volume: "bucket".to_string(),
+                name: "object".to_string(),
+                version_id: Some(Uuid::from_u128(99)),
+                deleted: true,
+                mod_time: Some(OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp")),
+                ..Default::default()
+            })
+            .expect("delete marker should be added");
+        let marker = MetaCacheEntry {
+            name: "object".to_string(),
+            metadata: marker_meta.marshal_msg().expect("delete marker metadata should marshal"),
+            cached: Some(marker_meta),
+            reusable: false,
+        };
+
+        let unversioned_entry = metacache_entry_with_mod_time(
+            OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp"),
+            "unversioned",
+        );
+        let discovery = MetaCacheEntries(vec![Some(marker), Some(unversioned_entry)]).discover_heal_candidates("bucket", 16);
+        assert!(discovery.candidates.iter().any(|candidate| {
+            candidate.kind == MetaCacheHealCandidateKind::DeleteMarker && candidate.version_id == Some(Uuid::from_u128(99))
+        }));
+        assert!(discovery.candidates.iter().any(|candidate| {
+            candidate.kind == MetaCacheHealCandidateKind::UnversionedObject && candidate.version_id.is_none()
+        }));
+    }
+
+    #[test]
+    fn discover_heal_candidates_rejects_delete_markers_without_ids() {
+        let mut marker_meta = FileMeta::new();
+        marker_meta
+            .add_version(FileInfo {
+                volume: "bucket".to_string(),
+                name: "object".to_string(),
+                deleted: true,
+                mod_time: Some(OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp")),
+                ..Default::default()
+            })
+            .expect("nil delete marker should be added");
+        let discovery = MetaCacheEntries(vec![Some(MetaCacheEntry {
+            name: "object".to_string(),
+            metadata: marker_meta.marshal_msg().expect("nil marker metadata should marshal"),
+            cached: Some(marker_meta),
+            reusable: false,
+        })])
+        .discover_heal_candidates("bucket", 16);
+
+        assert!(
+            !discovery
+                .candidates
+                .iter()
+                .any(|candidate| candidate.kind == MetaCacheHealCandidateKind::DeleteMarker)
+        );
+        assert!(discovery.unverified_count >= 1);
+    }
+
+    #[test]
+    fn discover_heal_candidates_skips_free_versions() {
+        let object_id = Uuid::from_u128(100);
+        let free_id = Uuid::from_u128(101);
+        let mut meta = FileMeta::new();
+        meta.add_version(FileInfo {
+            volume: "bucket".to_string(),
+            name: "object".to_string(),
+            version_id: Some(object_id),
+            transition_status: TRANSITION_COMPLETE.to_string(),
+            transitioned_objname: "remote/object".to_string(),
+            transition_version_id: Some(Uuid::from_u128(102)),
+            transition_tier: "WARM".to_string(),
+            mod_time: Some(OffsetDateTime::now_utc()),
+            ..Default::default()
+        })
+        .expect("transitioned object should be added");
+        let mut delete = FileInfo {
+            volume: "bucket".to_string(),
+            name: "object".to_string(),
+            version_id: Some(object_id),
+            mod_time: Some(OffsetDateTime::now_utc()),
+            ..Default::default()
+        };
+        delete.set_tier_free_version_id(&free_id.to_string());
+        meta.delete_version(&delete).expect("free version should be persisted");
+
+        let discovery = MetaCacheEntries(vec![Some(MetaCacheEntry {
+            name: "object".to_string(),
+            metadata: meta.marshal_msg().expect("free version metadata should marshal"),
+            cached: Some(meta),
+            reusable: false,
+        })])
+        .discover_heal_candidates("bucket", 16);
+        assert!(discovery.candidates.is_empty());
+    }
+
+    #[test]
+    fn discover_heal_candidates_preserves_unversioned_legacy_object() {
+        let legacy = MetaObjectV1 {
+            version: "1.0.1".to_string(),
+            format: "xl".to_string(),
+            stat: MetaObjectV1Stat {
+                size: 1,
+                mod_time: Some(OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp")),
+                name: "object".to_string(),
+                ..Default::default()
+            },
+            erasure: MetaObjectV1Erasure {
+                data_blocks: 4,
+                parity_blocks: 2,
+                index: 1,
+                distribution: vec![1, 2, 3, 4, 5, 6],
+                ..Default::default()
+            },
+            ..Default::default()
+        };
+        let version = FileMetaVersion {
+            version_type: VersionType::Legacy,
+            legacy_object: Some(legacy),
+            ..Default::default()
+        };
+        let mut meta = FileMeta::new();
+        meta.versions
+            .push(FileMetaShallowVersion::try_from(version).expect("legacy metadata should marshal"));
+        let discovery = MetaCacheEntries(vec![Some(MetaCacheEntry {
+            name: "object".to_string(),
+            metadata: meta.marshal_msg().expect("legacy metadata should marshal"),
+            cached: Some(meta),
+            reusable: false,
+        })])
+        .discover_heal_candidates("bucket", 16);
+        assert!(discovery.candidates.iter().any(|candidate| {
+            candidate.kind == MetaCacheHealCandidateKind::UnversionedObject && candidate.version_id.is_none()
+        }));
+    }
+
+    #[test]
+    fn discover_heal_candidates_rejects_nil_and_malformed_metadata_and_is_bounded() {
+        let now = OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp");
+        let mut nil = metacache_entry_single_version(1, now, "nil");
+        let mut nil_meta = FileMeta::load(&nil.metadata).expect("nil fixture should decode");
+        let mut nil_version = nil_meta.versions[0]
+            .parse_version_meta()
+            .expect("nil fixture version should decode");
+        nil_version.object.as_mut().expect("object fixture").version_id = Some(Uuid::nil());
+        nil_meta.versions[0] = FileMetaShallowVersion::try_from(nil_version).expect("nil fixture should marshal");
+        nil.metadata = nil_meta.marshal_msg().expect("nil fixture metadata should marshal");
+
+        let mut mismatched = metacache_entry_single_version(2, now, "mismatched");
+        let mut mismatched_meta = FileMeta::load(&mismatched.metadata).expect("mismatched fixture should decode");
+        mismatched_meta.versions[0].header.version_id = Some(Uuid::from_u128(200));
+        mismatched.metadata = mismatched_meta.marshal_msg().expect("mismatched metadata should marshal");
+
+        let mut short_parts = metacache_entry_single_version(3, now, "short-parts");
+        let mut short_parts_meta = FileMeta::load(&short_parts.metadata).expect("short-parts fixture should decode");
+        let mut short_parts_version = short_parts_meta.versions[0]
+            .parse_version_meta()
+            .expect("short-parts fixture version should decode");
+        let object = short_parts_version.object.as_mut().expect("object fixture");
+        object.part_numbers = vec![1];
+        object.part_actual_sizes = vec![1];
+        object.part_sizes.clear();
+        short_parts_meta.versions[0] = FileMetaShallowVersion::try_from(short_parts_version).expect("short-parts should marshal");
+        short_parts.metadata = short_parts_meta.marshal_msg().expect("short-parts metadata should marshal");
+
+        let mut short_unversioned = metacache_entry_with_mod_time(now, "short-unversioned");
+        let mut short_unversioned_meta =
+            FileMeta::load(&short_unversioned.metadata).expect("short-unversioned fixture should decode");
+        let mut short_unversioned_version = short_unversioned_meta.versions[0]
+            .parse_version_meta()
+            .expect("short-unversioned version should decode");
+        let unversioned_object = short_unversioned_version.object.as_mut().expect("unversioned object fixture");
+        unversioned_object.part_numbers = vec![1];
+        unversioned_object.part_actual_sizes = vec![1];
+        unversioned_object.part_sizes.clear();
+        short_unversioned_meta.versions[0] =
+            FileMetaShallowVersion::try_from(short_unversioned_version).expect("short-unversioned should marshal");
+        short_unversioned.metadata = short_unversioned_meta
+            .marshal_msg()
+            .expect("short-unversioned metadata should marshal");
+
+        let mut malformed = nil.clone();
+        malformed.name = "malformed".to_string();
+        malformed.metadata = vec![1, 2, 3];
+
+        let entries = MetaCacheEntries(
+            std::iter::once(Some(nil))
+                .chain(std::iter::once(Some(mismatched)))
+                .chain(std::iter::once(Some(short_parts)))
+                .chain(std::iter::once(Some(short_unversioned)))
+                .chain(std::iter::once(Some(malformed)))
+                .chain((0..32).map(|id| Some(metacache_entry_single_version(id + 10, now, "bounded"))))
+                .collect(),
+        );
+        let discovery = entries.discover_heal_candidates("bucket", 5);
+        assert!(discovery.candidates.len() <= 5);
+        assert!(discovery.truncated, "bounded discovery must expose dropped candidates");
+        assert!(
+            discovery
+                .truncated_candidates
+                .iter()
+                .all(|candidate| candidate.version_id.is_some()),
+            "overflow candidates must retain exact version identities"
+        );
+        assert!(
+            discovery.truncated_objects.iter().any(|object| object == "object"),
+            "bounded discovery must expose an object-level safe continuation"
+        );
+        assert!(
+            !discovery
+                .candidates
+                .iter()
+                .any(|candidate| candidate.version_id == Some(Uuid::nil()))
+        );
+        assert!(
+            !discovery
+                .candidates
+                .iter()
+                .any(|candidate| candidate.version_id == Some(Uuid::from_u128(2)))
+        );
+        assert!(
+            !discovery
+                .candidates
+                .iter()
+                .any(|candidate| candidate.version_id == Some(Uuid::from_u128(3)))
+        );
+        assert!(discovery.candidates.iter().any(|candidate| {
+            candidate.kind == MetaCacheHealCandidateKind::UnversionedObject && candidate.version_id.is_none()
+        }));
+        assert!(
+            discovery.unverified_count >= 1,
+            "malformed and rejected metadata must remain observable during discovery"
+        );
+
+        for invalid_name in [
+            "../object",
+            "./object",
+            "object/../other",
+            "object//name",
+            "object\u{0001}name",
+            "object\0name",
+        ] {
+            let mut entry = metacache_entry_single_version(400, now, invalid_name);
+            entry.name = invalid_name.to_string();
+            let discovery = MetaCacheEntries(vec![Some(entry)]).discover_heal_candidates("bucket", 5);
+            assert!(
+                discovery.candidates.is_empty(),
+                "invalid key should not become a heal candidate: {invalid_name:?}"
+            );
+        }
+
+        #[cfg(windows)]
+        {
+            let mut entry = metacache_entry_single_version(400, now, "object\\name");
+            entry.name = "object\\name".to_string();
+            assert!(
+                MetaCacheEntries(vec![Some(entry)])
+                    .discover_heal_candidates("bucket", 5)
+                    .candidates
+                    .is_empty(),
+                "backslash is a path separator on Windows"
+            );
+        }
+
+        #[cfg(not(windows))]
+        {
+            let mut entry = metacache_entry_single_version(400, now, "object\\name");
+            entry.name = "object\\name".to_string();
+            assert_eq!(
+                MetaCacheEntries(vec![Some(entry)])
+                    .discover_heal_candidates("bucket", 5)
+                    .candidates
+                    .len(),
+                1,
+                "backslash is object-key data on Unix"
+            );
+        }
+
+        for valid_name in ["trailing/", "prefix/object"] {
+            let mut entry = metacache_entry_single_version(401, now, valid_name);
+            entry.name = valid_name.to_string();
+            let discovery = MetaCacheEntries(vec![Some(entry)]).discover_heal_candidates("bucket", 5);
+            assert_eq!(discovery.candidates.len(), 1, "raw S3 key should remain opaque: {valid_name:?}");
+        }
+    }
+
     #[test]
     fn resolve_rejects_partial_latest_and_returns_committed_previous_metadata() {
         let old_mod_time = OffsetDateTime::from_unix_timestamp(1_705_312_300).expect("valid timestamp");

crates/scanner/src/scanner_folder.rs

@@ -43,7 +43,10 @@ use rustfs_common::metrics::{
     UpdateCurrentPathFn, current_path_updater, global_metrics,
 };
 use rustfs_common::trace_bus::{TraceEvent, TraceFunc, TraceKind, trace_emit, trace_subscriber_count};
-use rustfs_filemeta::{MetaCacheEntries, MetaCacheEntry, MetadataResolutionParams};
+use rustfs_filemeta::{
+    MAX_META_CACHE_HEAL_CANDIDATES, MAX_META_CACHE_HEAL_TRUNCATED_OBJECTS, MetaCacheEntries, MetaCacheEntry,
+    MetaCacheHealCandidateKind,
+};
 use rustfs_utils::path::{SLASH_SEPARATOR, path_join_buf};
 use s3s::dto::{BucketLifecycleConfiguration, ObjectLockConfiguration, VersioningConfiguration};
 use time::OffsetDateTime;
@@ -96,6 +99,11 @@ const METRIC_SCANNER_EXCESS_OBJECT_VERSION_SIZE_TOTAL: &str = "rustfs_scanner_ex
 const METRIC_SCANNER_EXCESS_FOLDERS_TOTAL: &str = "rustfs_scanner_excess_folders_total";
 const METRIC_SCANNER_PENDING_HEAL_PRUNE_TOTAL: &str = "rustfs_scanner_pending_heal_prune_total";
 const METRIC_SCANNER_PENDING_HEAL_MALFORMED_TOTAL: &str = "rustfs_scanner_pending_heal_malformed_total";
+const METRIC_SCANNER_HEAL_DISCOVERY_CANDIDATES_TOTAL: &str = "rustfs_scanner_heal_discovery_candidates_total";
+const METRIC_SCANNER_HEAL_DISCOVERY_SUB_QUORUM_TOTAL: &str = "rustfs_scanner_heal_discovery_sub_quorum_total";
+const METRIC_SCANNER_HEAL_DISCOVERY_UNVERIFIED_TOTAL: &str = "rustfs_scanner_heal_discovery_unverified_total";
+const METRIC_SCANNER_HEAL_DISCOVERY_QUEUED_TOTAL: &str = "rustfs_scanner_heal_discovery_queued_total";
+const METRIC_SCANNER_HEAL_DISCOVERY_TRUNCATED_TOTAL: &str = "rustfs_scanner_heal_discovery_truncated_total";
 const MAX_PENDING_SCANNER_HEAL_RETRIES_PER_BUCKET: usize = 128;
 
 // --- scanner excess alerts as S3 notification events (rustfs/backlog#1868) --
@@ -883,7 +891,7 @@ impl FolderScanner {
         object: Option<String>,
         version_id: Option<String>,
         request: HealChannelRequest,
-    ) -> Result<(), ScannerError> {
+    ) -> Result<HealAdmissionResult, ScannerError> {
         let candidate_type = pending_scanner_heal_candidate_type(kind);
         let priority = request.priority;
         let scan_mode = request.scan_mode.unwrap_or(self.scan_mode);
@@ -911,7 +919,7 @@ impl FolderScanner {
                     error = %err,
                     "Scanner deferred heal request after channel error"
                 );
-                return Ok(());
+                return Ok(HealAdmissionResult::Full);
             }
         };
         self.update_pending_scanner_heal_after_admission(
@@ -923,7 +931,7 @@ impl FolderScanner {
             result,
         );
         if result.is_admitted() {
-            return Ok(());
+            return Ok(result);
         }
 
         record_high_priority_heal_escalation(candidate_type, priority, result);
@@ -944,7 +952,7 @@ impl FolderScanner {
             state = "high_priority_not_admitted",
             "Scanner high-priority heal admission failed"
         );
-        Ok(())
+        Ok(result)
     }
 
     pub fn set_heal_object_select(&mut self, prob: u32) {
@@ -1736,14 +1744,7 @@ impl FolderScanner {
                 break;
             }
 
-            let mut resolver = MetadataResolutionParams {
-                dir_quorum: self.disks_quorum,
-                obj_quorum: self.disks_quorum,
-                bucket: "".to_string(),
-                strict: false,
-                ..Default::default()
-            };
-
+            let mut previous_bucket = String::new();
             for name in abandoned_children {
                 if !self.should_heal().await {
                     break;
@@ -1751,7 +1752,7 @@ impl FolderScanner {
 
                 let (bucket, prefix) = path2_bucket_object(name.as_str());
 
-                if bucket != resolver.bucket {
+                if bucket != previous_bucket {
                     self.send_required_scanner_heal_request(
                         PendingScannerHealKind::Bucket,
                         bucket.clone(),
@@ -1760,10 +1761,9 @@ impl FolderScanner {
                         build_bucket_heal_request(bucket.clone(), HealChannelPriority::High),
                     )
                     .await?;
+                    previous_bucket = bucket.clone();
                 }
 
-                resolver.bucket = bucket.clone();
-
                 let child_ctx = ctx.child_token();
 
                 let (agreed_tx, mut agreed_rx) = mpsc::channel::<String>(1);
@@ -1880,6 +1880,8 @@ impl FolderScanner {
                 let mut agreed_closed = false;
                 let mut partial_closed = false;
                 let mut finished_closed = false;
+                let mut seen_heal_candidates: HashSet<(String, Option<String>, MetaCacheHealCandidateKind)> = HashSet::new();
+                let mut seen_truncated_objects: HashSet<String> = HashSet::new();
 
                 loop {
                     if agreed_closed && partial_closed && finished_closed {
@@ -1904,65 +1906,112 @@ impl FolderScanner {
                                 break;
                             }
 
-                            let Some(entry) = resolve_object_heal_entry(&entries, resolver.clone()) else {
-                                continue;
-                            };
-
-                            (self.update_current_path)(&entry.name).await;
-
-                            if entry.is_dir() {
-                                continue;
+                            let discovery = entries.discover_heal_candidates(&bucket, MAX_META_CACHE_HEAL_CANDIDATES);
+                            counter!(METRIC_SCANNER_HEAL_DISCOVERY_CANDIDATES_TOTAL)
+                                .increment(u64::try_from(discovery.candidates.len()).unwrap_or(u64::MAX));
+                            counter!(METRIC_SCANNER_HEAL_DISCOVERY_SUB_QUORUM_TOTAL).increment(
+                                u64::try_from(
+                                    discovery
+                                        .candidates
+                                        .iter()
+                                        .filter(|candidate| candidate.replica_count < disks_quorum)
+                                        .count(),
+                                )
+                                .unwrap_or(u64::MAX),
+                            );
+                            counter!(METRIC_SCANNER_HEAL_DISCOVERY_UNVERIFIED_TOTAL).increment(
+                                u64::try_from(discovery.unverified_count).unwrap_or(u64::MAX),
+                            );
+                            if discovery.truncated {
+                                counter!(METRIC_SCANNER_HEAL_DISCOVERY_TRUNCATED_TOTAL).increment(1);
                             }
 
-                            let fivs = match entry.file_info_versions(&bucket) {
-                                Ok(fivs) => fivs,
-                                Err(e) => {
-                                    error!(
-                                        target: "rustfs::scanner::folder",
-                                        event = EVENT_SCANNER_FOLDER_STATE,
-                                        component = LOG_COMPONENT_SCANNER,
-                                        subsystem = LOG_SUBSYSTEM_FOLDER,
-                                        bucket = %bucket,
-                                        entry = %entry.name,
-                                        state = "file_info_versions_failed",
-                                        error = %e,
-                                        "Scanner list_path_raw failed to resolve file versions"
-                                    );
-                                    self.send_required_scanner_heal_request(
-                                        PendingScannerHealKind::Object,
-                                        bucket.clone(),
-                                        Some(entry.name.clone()),
-                                        None,
-                                        build_object_heal_request(
-                                            bucket.clone(),
-                                            entry.name.clone(),
-                                            None,
-                                            self.scan_mode,
-                                            HealChannelPriority::High,
-                                        ),
-                                    )
-                                    .await?;
-                                    found_objects = true;
+                            for candidate in discovery.candidates {
+                                let sub_quorum_candidate = candidate.replica_count < disks_quorum;
+                                let version_id = candidate.validated_version().map(|id| id.to_string());
+                                let identity = (candidate.object.clone(), version_id.clone(), candidate.kind.clone());
+                                if seen_heal_candidates.len() >= MAX_META_CACHE_HEAL_CANDIDATES
+                                    && !seen_heal_candidates.contains(&identity)
+                                {
                                     continue;
                                 }
-                            };
-
-                            for fiv in fivs.versions {
-                                let version_id = fiv.version_id.and_then(|v| if v.is_nil() { None } else { Some(v.to_string()) });
-                                self.send_required_scanner_heal_request(
-                                    PendingScannerHealKind::Object,
-                                    bucket.clone(),
-                                    Some(entry.name.clone()),
-                                    version_id.clone(),
-                                    build_object_heal_request(
+                                if !seen_heal_candidates.insert(identity) {
+                                    continue;
+                                }
+                                let request = if candidate.is_unversioned() {
+                                    build_non_destructive_object_heal_request(
                                         bucket.clone(),
-                                        entry.name.clone(),
-                                        version_id,
+                                        candidate.object.clone(),
                                         self.scan_mode,
                                         HealChannelPriority::High,
-                                    ),
+                                    )
+                                } else {
+                                    build_object_heal_request(
+                                        bucket.clone(),
+                                        candidate.object.clone(),
+                                        version_id.clone(),
+                                        self.scan_mode,
+                                        HealChannelPriority::High,
+                                    )
+                                };
+                                (self.update_current_path)(&candidate.object).await;
+                                let admission = self.send_required_scanner_heal_request(
+                                    PendingScannerHealKind::Object,
+                                    bucket.clone(),
+                                    Some(candidate.object.clone()),
+                                    version_id.clone(),
+                                    request,
                                 )
                                 .await?;
+                                if admission.is_admitted() {
+                                    counter!(METRIC_SCANNER_HEAL_DISCOVERY_QUEUED_TOTAL).increment(1);
+                                } else if sub_quorum_candidate {
+                                    self.mark_pending_scanner_heal_reason(
+                                        PendingScannerHealKind::Object,
+                                        &bucket,
+                                        Some(&candidate.object),
+                                        version_id.as_deref(),
+                                        "sub_quorum_metadata",
+                                    );
+                                }
+                                found_objects = true;
+                            }
+
+                            // Candidates beyond the main cap remain exact
+                            // version requests; never downgrade them to a
+                            // latest-version (version_id=None) heal.
+                            for candidate in discovery.truncated_candidates {
+                                let version_id = candidate.validated_version().map(|id| id.to_string());
+                                let identity = (candidate.object.clone(), version_id.clone(), candidate.kind.clone());
+                                if seen_truncated_objects.len() >= MAX_META_CACHE_HEAL_TRUNCATED_OBJECTS
+                                    && !seen_truncated_objects.contains(&candidate.object)
+                                {
+                                    continue;
+                                }
+                                seen_truncated_objects.insert(candidate.object.clone());
+                                if !seen_heal_candidates.insert(identity) {
+                                    continue;
+                                }
+                                let request = build_object_heal_request(
+                                    bucket.clone(),
+                                    candidate.object.clone(),
+                                    version_id.clone(),
+                                    self.scan_mode,
+                                    HealChannelPriority::High,
+                                );
+                                (self.update_current_path)(&candidate.object).await;
+                                let admission = self
+                                    .send_required_scanner_heal_request(
+                                        PendingScannerHealKind::Object,
+                                        bucket.clone(),
+                                        Some(candidate.object.clone()),
+                                        version_id,
+                                        request,
+                                    )
+                                    .await?;
+                                if admission.is_admitted() {
+                                    counter!(METRIC_SCANNER_HEAL_DISCOVERY_QUEUED_TOTAL).increment(1);
+                                }
                                 found_objects = true;
                             }
 

crates/scanner/src/scanner_folder/item_actions.rs

@@ -13,6 +13,8 @@
 // limitations under the License.
 /// Per-object scan actions: ScannerItem, the get-size failure policy, and the heal/ILM admission helpers.
 use super::*;
+#[cfg(test)]
+use rustfs_filemeta::MetadataResolutionParams;
 
 /// Cached folder information for scanning
 #[derive(Clone, Debug)]
@@ -88,6 +90,22 @@ pub(super) fn build_object_heal_request(
     }
 }
 
+/// Build the versionless inspection request used when discovery cannot prove
+/// a destructive version identity (for example an unversioned object or a
+/// bounded candidate overflow). The explicit flag is the fail-closed safety
+/// boundary; callers must not reconstruct it with the destructive default.
+pub(super) fn build_non_destructive_object_heal_request(
+    bucket: String,
+    object: String,
+    scan_mode: HealScanMode,
+    priority: HealChannelPriority,
+) -> HealChannelRequest {
+    let mut request = build_object_heal_request(bucket, object, None, scan_mode, priority);
+    request.remove_corrupted = Some(false);
+    request
+}
+
+#[cfg(test)]
 pub(super) fn resolve_object_heal_entry(
     entries: &MetaCacheEntries,
     resolver: MetadataResolutionParams,

crates/scanner/src/scanner_folder/ledger.rs

@@ -105,6 +105,29 @@ impl FolderScanner {
         }
     }
 
+    /// Preserve the discovery reason when a candidate could not be admitted
+    /// immediately. The existing string field is intentionally reused so the
+    /// scanner's map-encoded cache schema stays backward compatible.
+    pub(super) fn mark_pending_scanner_heal_reason(
+        &mut self,
+        kind: PendingScannerHealKind,
+        bucket: &str,
+        object: Option<&str>,
+        version_id: Option<&str>,
+        reason: &str,
+    ) {
+        if let Some(entry) = self
+            .new_cache
+            .info
+            .pending_heals
+            .iter_mut()
+            .find(|entry| pending_scanner_heal_matches(entry, kind, bucket, object, version_id))
+        {
+            entry.last_admission_reason = reason.to_string();
+            self.sync_pending_heals();
+        }
+    }
+
     pub(super) fn prune_pending_scanner_heals(&mut self) {
         let now = Self::now_secs();
         let before_expiry = self.new_cache.info.pending_heals.len();
@@ -305,13 +328,22 @@ pub(super) fn build_pending_scanner_heal_request(entry: &PendingScannerHeal) ->
     match entry.kind {
         PendingScannerHealKind::Bucket => Some(build_bucket_heal_request(entry.bucket.clone(), HealChannelPriority::High)),
         PendingScannerHealKind::Object => entry.object.as_ref().map(|object| {
-            build_object_heal_request(
-                entry.bucket.clone(),
-                object.clone(),
-                entry.version_id.clone(),
-                entry.scan_mode,
-                HealChannelPriority::High,
-            )
+            if entry.version_id.is_none() {
+                build_non_destructive_object_heal_request(
+                    entry.bucket.clone(),
+                    object.clone(),
+                    entry.scan_mode,
+                    HealChannelPriority::High,
+                )
+            } else {
+                build_object_heal_request(
+                    entry.bucket.clone(),
+                    object.clone(),
+                    entry.version_id.clone(),
+                    entry.scan_mode,
+                    HealChannelPriority::High,
+                )
+            }
         }),
     }
 }

crates/scanner/src/scanner_folder/tests.rs

@@ -17,7 +17,7 @@ use crate::SCANNER_SLEEPER;
 use super::*;
 use crate::storage_api::VersionPurgeStatusType;
 use crate::{DiskOption, Endpoint, STORAGE_FORMAT_FILE, TierStats, new_disk, storageclass};
-use rustfs_filemeta::{FileInfo, FileMeta};
+use rustfs_filemeta::{FileInfo, FileMeta, MetadataResolutionParams};
 use std::io::Write;
 #[cfg(unix)]
 use std::os::unix::fs::{PermissionsExt, symlink};
@@ -982,6 +982,21 @@ fn test_build_object_heal_request_omits_nil_version_id() {
     assert_eq!(request.recreate_missing, Some(false));
 }
 
+#[test]
+fn test_build_non_destructive_object_heal_request_disables_removal() {
+    let request = build_non_destructive_object_heal_request(
+        "bucket".to_string(),
+        "path/to/object".to_string(),
+        HealScanMode::Deep,
+        HealChannelPriority::High,
+    );
+
+    assert_eq!(request.object_version_id, None);
+    assert_eq!(request.remove_corrupted, Some(false));
+    assert_eq!(request.recreate_missing, Some(false));
+    assert_eq!(request.source, HealRequestSource::Scanner);
+}
+
 #[test]
 fn test_build_bucket_heal_request_disables_recreate_for_scanner() {
     let request = build_bucket_heal_request("bucket".to_string(), HealChannelPriority::Low);
@@ -1121,6 +1136,42 @@ fn test_pending_heal_reconstructs_object_request_with_version() {
     assert_eq!(request.source, HealRequestSource::Scanner);
 }
 
+#[test]
+fn test_pending_heal_reconstructs_unversioned_request_without_removal() {
+    let pending = pending_heal(PendingScannerHealKind::Object, "bucket", Some("object"), None, 1, 1);
+
+    let request = build_pending_scanner_heal_request(&pending).expect("unversioned object request should rebuild");
+
+    assert!(request.object_version_id.is_none());
+    assert_eq!(request.remove_corrupted, Some(false));
+    assert_eq!(request.recreate_missing, Some(false));
+}
+
+#[tokio::test]
+async fn test_pending_heal_reason_preserves_sub_quorum_discovery() {
+    let (mut scanner, temp_dir) = build_test_scanner().await;
+    let _guard = TestGuard::new(u64::MAX, usize::MAX, &mut scanner, temp_dir);
+
+    scanner.update_pending_scanner_heal_after_admission(
+        PendingScannerHealKind::Object,
+        "bucket",
+        Some("object"),
+        Some("version-a"),
+        HealScanMode::Deep,
+        HealAdmissionResult::Full,
+    );
+    scanner.mark_pending_scanner_heal_reason(
+        PendingScannerHealKind::Object,
+        "bucket",
+        Some("object"),
+        Some("version-a"),
+        "sub_quorum_metadata",
+    );
+
+    assert_eq!(scanner.new_cache.info.pending_heals.len(), 1);
+    assert_eq!(scanner.new_cache.info.pending_heals[0].last_admission_reason, "sub_quorum_metadata");
+}
+
 #[test]
 fn test_pending_heal_retry_candidates_respect_cap_and_order() {
     let pending: Vec<PendingScannerHeal> = (0..(MAX_PENDING_SCANNER_HEAL_RETRIES_PER_BUCKET + 2))
@@ -1323,6 +1374,20 @@ fn metadata_for_object(bucket: &str, object: &str) -> Vec<u8> {
     meta.marshal_msg().expect("test metadata should marshal")
 }
 
+fn metadata_for_object_version(bucket: &str, object: &str, version_id: Option<Uuid>) -> Vec<u8> {
+    let mut file_info = FileInfo::new(object, 4, 2);
+    file_info.volume = bucket.to_string();
+    file_info.name = object.to_string();
+    file_info.version_id = version_id;
+    file_info.versioned = version_id.is_some();
+    file_info.mod_time = Some(OffsetDateTime::now_utc());
+    file_info.size = 1;
+
+    let mut meta = FileMeta::new();
+    meta.add_version(file_info).expect("test metadata version should be accepted");
+    meta.marshal_msg().expect("test metadata should marshal")
+}
+
 async fn write_test_object_metadata(root: &std::path::Path, bucket: &str, object: &str) {
     write_test_object_metadata_bytes(root, bucket, object, &metadata_for_object(bucket, object)).await;
 }
@@ -1726,12 +1791,21 @@ async fn test_scan_folder_exits_when_abandoned_child_listing_finishes() {
     let _guard = TestGuard::new(60, 100, &mut scanner, temp_dir.clone());
     let heal_starts = Arc::new(AtomicUsize::new(0));
     let heal_starts_clone = heal_starts.clone();
+    let healed_versions = Arc::new(Mutex::new(Vec::<Option<String>>::new()));
+    let healed_versions_clone = healed_versions.clone();
     let mut heal_rx =
         rustfs_common::heal_channel::init_heal_channel().expect("heal channel should initialize once for scanner tests");
     let _heal_responder = tokio::spawn(async move {
         while let Some(command) = heal_rx.recv().await {
-            if let rustfs_common::heal_channel::HealChannelCommand::Start { response_tx, .. } = command {
+            if let rustfs_common::heal_channel::HealChannelCommand::Start {
+                request, response_tx, ..
+            } = command
+            {
                 heal_starts_clone.fetch_add(1, Ordering::Relaxed);
+                healed_versions_clone
+                    .lock()
+                    .expect("heal version capture lock should not be poisoned")
+                    .push(request.object_version_id);
                 let _ = response_tx.send(Ok(HealAdmissionResult::Accepted));
             }
         }
@@ -1739,13 +1813,18 @@ async fn test_scan_folder_exits_when_abandoned_child_listing_finishes() {
 
     let bucket = "src-archive";
     let object = "snapshots/37b3f20d941e2f5e6d99114d9bb2f3e67a8a2e5c9c4c5a1b0d6e7f8091a2b3c4";
-    let metadata = metadata_for_object(bucket, object);
-    write_test_object_metadata_bytes(&temp_dir, bucket, object, &metadata).await;
+    let orphan_version = Uuid::from_u128(0x1934);
+    let shared_version = Uuid::from_u128(0x1935);
+    let orphan_metadata = metadata_for_object_version(bucket, object, Some(orphan_version));
+    let shared_metadata = metadata_for_object_version(bucket, object, Some(shared_version));
+    write_test_object_metadata_bytes(&temp_dir, bucket, object, &orphan_metadata).await;
+    let mut expected_metadata = vec![(temp_dir.join(bucket).join(object).join("xl.meta"), orphan_metadata.clone())];
 
     let mut disks = vec![scanner.local_disk.clone()];
     for disk_name in ["disk2", "disk3", "disk4"] {
         let disk_root = temp_dir.join(disk_name);
-        write_test_object_metadata_bytes(&disk_root, bucket, object, &metadata).await;
+        write_test_object_metadata_bytes(&disk_root, bucket, object, &shared_metadata).await;
+        expected_metadata.push((disk_root.join(bucket).join(object).join("xl.meta"), shared_metadata.clone()));
         let endpoint = Endpoint::try_from(disk_root.to_string_lossy().as_ref()).expect("failed to create extra disk endpoint");
         let disk = new_disk(
             &endpoint,
@@ -1794,8 +1873,29 @@ async fn test_scan_folder_exits_when_abandoned_child_listing_finishes() {
         .new_cache
         .checked_flatten(bucket)
         .expect("healed cache must contain canonical child links");
-    assert_eq!(root.objects, 1);
+    // The fixture intentionally exposes two divergent version histories, so
+    // the scanner keeps both logical versions visible while discovering heals.
+    assert_eq!(root.objects, 2);
     assert!(heal_starts.load(Ordering::Relaxed) > 0, "test must execute the heal child-link path");
+    let orphan_version_text = orphan_version.to_string();
+    assert!(
+        healed_versions
+            .lock()
+            .expect("heal version capture lock should not be poisoned")
+            .iter()
+            .any(|version| version.as_deref() == Some(orphan_version_text.as_str())),
+        "sub-quorum orphan version must be submitted as an exact heal candidate"
+    );
+    for (path, expected) in expected_metadata {
+        assert_eq!(
+            tokio::fs::read(&path)
+                .await
+                .expect("scanner discovery must not delete metadata"),
+            expected,
+            "scanner discovery must not modify candidate metadata: {}",
+            path.display()
+        );
+    }
 }
 
 #[tokio::test]

deny.toml

@@ -43,9 +43,6 @@ allow-git = [
     # RustFS fork carrying presigned expiry and constant-time authentication fixes.
     # owner: rustfs-maintainers  review: 2026-10
     "https://github.com/rustfs/s3s.git",
-    # MiMalloc fork pinned for hotpath allocation counting support.
-    # owner: houseme  review: 2026-10
-    "https://github.com/xonatius/mimalloc_rust.git",
 ]
 
 [bans]

docs/testing/ci-gates.md

@@ -0,0 +1,149 @@
+# CI gate matrix
+
+This file is the source of truth for which validation runs on each event, its
+configured wall-clock budget, and whether it can block a merge. Test taxonomy,
+naming, and nextest serialization rules remain in [README.md](README.md); e2e
+membership and counts remain in
+[e2e-suite-inventory.md](e2e-suite-inventory.md).
+
+The distinction between **required** and **report-only** is load-bearing:
+a failing job blocks a merge only when its exact check name is present in the
+live `main` ruleset. A workflow name, a `merge_group` trigger, or a red PR check
+does not make a job required by itself.
+
+## Required merge checks
+
+The live `main` ruleset (`6436880`) currently requires exactly these contexts:
+
+| Required context | Producer | Validation |
+|---|---|---|
+| `CLA Check` | `.github/workflows/cla.yml` | Contributor agreement |
+| `Quick Checks` | `.github/workflows/ci.yml` | Formatting and repository guard scripts |
+| `Test and Lint` | `.github/workflows/ci.yml` | Clippy, workspace nextest excluding `e2e_test`, doctests, and migration proofs |
+
+For pull requests limited to the paths excluded by the main CI workflow,
+`.github/workflows/ci-docs-only.yml` reports `Quick Checks` and
+`Test and Lint` under the same names. It runs the real quick checks and the
+planning-document guard; it does not claim that Rust compilation or runtime
+tests ran. Despite the workflow name, these paths also include selected deploy,
+workflow, and lock files.
+
+Verify the live rule rather than trusting this snapshot before changing merge
+policy:
+
+```bash
+gh api repos/rustfs/rustfs/rulesets/6436880 \
+  --jq '.rules[] | select(.type == "required_status_checks") | .parameters'
+```
+
+The ruleset currently has `strict_required_status_checks_policy=false`.
+`Continuous Integration` accepts `merge_group` events and runs `e2e-full` for
+them, but `End-to-End Tests (full merge gate)` is not currently a required
+context. Therefore the repository is prepared to test a merge-queue SHA, but
+the workflow alone does not prove that every merge passed that lane.
+
+## Pull request and merge matrix
+
+Budgets below are job `timeout-minutes`, not typical runtimes. “Report-only”
+means the result is visible and actionable but is not in the live required
+context list.
+
+| Event | Validation | Budget | Merge status | Reproduction |
+|---|---|---:|---|---|
+| PR, non-doc change | `Quick Checks` | 10 min | Required | `make pre-commit` (broader local umbrella) |
+| PR, non-doc change | `Test and Lint` | 90 min | Required | `cargo nextest run --profile ci --all --exclude e2e_test` |
+| PR, non-doc change | `Typos` | 10 min | Report-only | `typos` |
+| PR, non-doc change | `ILM Integration (serial)` | 90 min | Report-only | Use the exact command in `.github/workflows/ci.yml` |
+| PR, non-doc change | rio-v2 / swift / sftp test-and-lint variants | 90 min each | Report-only | `cargo nextest run` with the workflow's feature set |
+| PR, non-doc change | `Build RustFS Debug Binary` | 30 min | Report-only; prerequisite for black-box lanes | `cargo build -p rustfs --bins` |
+| PR, non-doc change | `io_uring Integration (real)` | 30 min | Report-only | `cargo test -p rustfs-ecstore --lib uring_ -- --test-threads=1 --nocapture` |
+| PR, non-doc change | `End-to-End Tests` (`e2e-smoke` plus `s3s-e2e`) | 30 min | Report-only | `cargo nextest run --profile e2e-smoke -p e2e_test`; then `./scripts/e2e-run.sh ./target/debug/rustfs <data-dir>` |
+| PR, non-doc change | `S3 Implemented Tests` | 60 min | Report-only | Build `rustfs`, then run `scripts/s3-tests/run.sh` with `DEPLOY_MODE=binary`, `TEST_MODE=single`, and `MAXFAIL=0` |
+| PR, non-doc change | `S3 Lifecycle Behavior Tests` | 30 min | Report-only | Use the accelerated scanner environment in `.github/workflows/ci.yml` with `scripts/s3-tests/run.sh` |
+| PR touching dependency or workflow inputs | Cargo Deny / Workflow Pin Report / Dependency Review | 20 / 5 / 30 min | Report-only | `cargo deny check`; `scripts/security/check_workflow_pins.sh` |
+| PR touching architecture rules or architecture docs | `Architecture Migration Rules` | 10 min | Report-only | `scripts/check_architecture_migration_rules.sh` |
+| PR touching Nix or workspace manifests | `Nix Build & Check` | 60 min | Report-only | `nix flake check` |
+| PR limited to main-CI-excluded paths | companion `Quick Checks` and `Test and Lint` | 10 min each | Required | `git diff --check`; `make doc-paths-check` when documentation paths changed |
+| `merge_group` | Standard CI plus `e2e-full` | 55 min for `e2e-full` | Standard required contexts only; `e2e-full` report-only | `cargo nextest run --profile e2e-full -p e2e_test` |
+| Push to `main` | Standard CI plus `e2e-full` | 55 min for `e2e-full` | Post-merge detection | Same as `merge_group` |
+| PR touching fuzz inputs or harness paths | Build plus five 60-second fuzz smoke targets | 60 min build; 30 min per target | Report-only | `MAX_TOTAL_TIME=60 ./scripts/fuzz/run.sh` |
+| PR touching selected ecstore disk/format paths | `Rename Safety` on Windows | 60 min | Report-only | Run the four `cargo test -p rustfs-ecstore --lib <filter>` commands in `windows-filesystem.yml` on Windows |
+
+The authoritative e2e filters live in `.config/nextest.toml`; extend a profile
+instead of adding a second ad-hoc selector. Before a profile runs,
+`scripts/check_test_wiring.py` compares its exact membership to the committed
+digest so a silent test drop fails closed.
+
+## Scheduled and manual validation
+
+Scheduled lanes are independent fault domains. They do not block a pull
+request, but their workflow-local gate can fail the run and scheduled failures
+are routed to the shared failure-issue action. The scheduled-validation
+watchdog and freshness workflow separately detect incomplete runs and missing
+schedules.
+
+| Cadence (UTC unless noted) | Workflow / validation | Budget | Verdict and artifacts | Reproduction |
+|---|---|---:|---|---|
+| Daily 02:17 | Fuzz: five nightly corpus targets | 60 min build; 60 min per target | Gate; corpus/crash artifacts, scheduled failure alert | `MAX_TOTAL_TIME=<seconds> ./scripts/fuzz/run.sh` |
+| Daily 03:17 | MinIO interop (EC + SSE read parity) | 40 min | Gate; scheduled failure alert | Dispatch `minio-interop.yml` or follow its pinned Docker fixture steps |
+| Daily 04:29 | Replication / cluster-fault / protocol e2e | 45 / 90 / 90 min | Three independent gates; JUnit, membership, and server logs | `cargo nextest run --profile e2e-repl-nightly -p e2e_test`; `--profile e2e-nightly`; `-j 1 --profile e2e-protocols` |
+| Daily 06:31 | Warp performance A/B | 180 min | Regression budget gate; A/B summaries and server logs | `bash scripts/run_hotpath_warp_abba.sh --help` |
+| Daily 00:07 Asia/Shanghai (16:07 UTC previous day) | Nightly GNU build and Vault lanes | 150 / 90 / 60 min | Build, live Vault, and HA failover gates | Use the commands and pinned Vault images in `nightly-gnu.yml` |
+| Daily 03:23 | Security Audit | 20 / 5 min, plus 30 min on PR dependency review | Cargo Deny and workflow-pin gates; scheduled failure alert | `cargo deny check`; `scripts/security/check_workflow_pins.sh` |
+| Daily 23:47 | Scheduled Validation Freshness | 10 min | Fails when a critical schedule was never created or is stale | Dispatch `scheduled-validation-freshness.yml` |
+| Sunday 00:11 | Full `Continuous Integration` matrix | Per-job budgets above | Weekly variant coverage, including dormant rio-v2 binary/e2e lanes | Dispatch `ci.yml` |
+| Sunday 01:13 | Seven-platform build matrix | 150 min per platform | Build/package integrity; scheduled failure alert | Dispatch `build.yml` with an exact platform set |
+| Sunday 02:19 | Ceph s3-tests full sweep: single and real four-node, four shards each | 180 min per shard | Compatibility gate; report, JUnit, exact node IDs, and server logs | `scripts/s3-tests/run.sh` against an existing single or distributed target |
+| Sunday 06:41 | Mint | 120 min | **Report-only by design**; per-suite PASS/FAIL/NA and raw `log.json` | Reproduce the pinned Docker sequence in `mint.yml` or dispatch it |
+| Sunday 07:43 | Workspace line coverage | 120 min | Report-only trend; lcov and JSON retained 90 days | `make coverage` |
+| Monthly, day 1 06:37 | Runner Hygiene | 15 min | Validates runner ephemerality; scheduled failure alert | Dispatch `runner-hygiene.yml` |
+
+Manual `workflow_dispatch` exists for the scheduled workflows above. Manual
+runs are debugging evidence and intentionally do not open scheduled-failure
+issues. A manual performance run may explicitly allow a known regression; that
+override must not be treated as an ordinary passing baseline.
+
+## Release validation
+
+Release validation is post-merge and tag-driven; it does not substitute for a
+pull-request gate.
+
+| Event | Validation | Budget | Result |
+|---|---|---:|---|
+| Push to `main` or weekly schedule | `Build and Release` platform matrix | 150 min per platform | Build artifacts for all selected targets; no release publication on a main push |
+| Valid release or preview tag | `Build and Release` plus asset checks | 150 min per platform | Draft release, checksummed assets, and publish step |
+| Successful non-preview release-tag build | Docker image build and image scan | 60 min build; 30 min scan | Multi-architecture images plus vulnerability report |
+| Successful release-tag build | DEB/RPM packaging | 30 min per architecture | Packages and checksum files uploaded to the release |
+| Successful non-preview release-tag build | Helm template test and package | 30 min build; 30 min publish | Versioned chart and repository index |
+
+Use an exact preview tag for end-to-end release rehearsal. Manual dispatches
+are backfill/debug paths and do not prove the automatic `workflow_run` chain.
+
+## Evidence requirements
+
+A green check is useful only when it proves the intended behavior ran:
+
+- Record the exact commit SHA and run URL.
+- Separate product failure from runner prerequisites, service readiness, and
+  cancellation. Repair the precondition, then rerun the exact workload.
+- Preserve membership manifests, JUnit, raw compatibility logs, seeds, and
+  server logs where the workflow provides them.
+- For a bug fix or a new fault checker, provide sensitivity evidence: the old
+  behavior or an intentional mutation must fail the new oracle, and the fixed
+  behavior must pass it.
+- Never promote a report-only lane to required from one green run. Require at
+  least 14 days and 30 representative pull requests with at least 99% complete
+  execution, then update the ruleset and this table together.
+
+## Change checklist
+
+Update this file in the same pull request when any of these change:
+
+- workflow triggers, job names, timeouts, or nextest profile ownership;
+- required status contexts or strict/merge-queue policy;
+- scheduled cadence, alert routing, artifact contract, or local reproduction;
+- report-only versus gating semantics.
+
+Do not copy per-module test counts here. Update
+[e2e-suite-inventory.md](e2e-suite-inventory.md) and its enforced membership
+digest instead.

rustfs/Cargo.toml

@@ -336,13 +336,13 @@ opentelemetry = { workspace = true }
 tracing-opentelemetry = { workspace = true }
 # Data structures
 hashbrown = { workspace = true, features = ["serde", "rayon"] }
-mimalloc = { workspace = true }
+rustfs-mimalloc = { workspace = true }
 
 [target.'cfg(target_os = "linux")'.dependencies]
 libsystemd.workspace = true
 
 [target.'cfg(not(target_os = "windows"))'.dependencies]
-libmimalloc-sys.workspace = true
+rustfs-mimalloc-sys.workspace = true
 
 [dev-dependencies]
 uuid = { workspace = true, features = ["v4", "v5", "fast-rng", "macro-diagnostics"] }

rustfs/src/allocator_reclaim.rs

@@ -369,14 +369,8 @@ pub fn allocator_reclaim_controller_snapshot(ctx: &CancellationToken) -> Allocat
 }
 
 #[cfg(not(target_os = "windows"))]
-#[allow(unsafe_code)]
 fn collect_allocator_memory(force: bool) -> Result<(), String> {
-    // SAFETY: `mi_collect` is provided by the active global allocator backend
-    // on this target family. It is explicitly intended to reclaim retained
-    // pages/segments and does not require additional invariants from the caller.
-    unsafe {
-        libmimalloc_sys::mi_collect(force);
-    }
+    rustfs_mimalloc::MiMalloc::collect(force);
     Ok(())
 }
 

rustfs/src/connect/heartbeat.rs

@@ -19,19 +19,23 @@ use std::sync::atomic::{AtomicU64, Ordering};
 use std::time::Duration;
 
 use chrono::{DateTime, SecondsFormat, Utc};
+use reqwest::{Client, StatusCode, Url, header};
+use rustls::RootCertStore;
+use rustls::pki_types::{CertificateDer, pem::PemObject as _};
 use serde::{Deserialize, Serialize};
 use uuid::Uuid;
+use zeroize::Zeroizing;
 
 use super::config::HeartbeatConfig;
-use super::credential_store::CredentialStoreError;
+use super::credential_store::{CredentialStoreError, DeviceCredential};
 use super::identity::IdentityError;
 use super::identity_store::StoreError;
-use super::registration::CredentialValidationError;
-use super::telemetry::{TelemetryDelivery, TelemetryError, TelemetryTransport, is_exact_utc_seconds};
+use super::registration::{CredentialValidationError, validate_stored_credential};
 
 const PROTOCOL_VERSION: &str = "v1";
 const AGENT_VERSION: &str = concat!("rustfs-agent/", env!("CARGO_PKG_VERSION"));
 const MAX_SEQUENCE: u64 = 9_007_199_254_740_991;
+const MAX_RESPONSE_BYTES: usize = 64 * 1024;
 #[cfg(unix)]
 const FILE_MODE: u32 = 0o600;
 static STAGING_SEQUENCE: AtomicU64 = AtomicU64::new(0);
@@ -118,39 +122,133 @@ pub(crate) enum Delivery {
 }
 
 pub(crate) struct HeartbeatSender {
-    transport: TelemetryTransport,
+    endpoint: Url,
+    root_store: RootCertStore,
+    roots: Vec<CertificateDer<'static>>,
+    config: HeartbeatConfig,
 }
 
 impl HeartbeatSender {
     pub(crate) fn new(config: HeartbeatConfig) -> Result<Self, HeartbeatError> {
+        let mut endpoint = Url::parse(&config.endpoint).map_err(|_| HeartbeatError::Endpoint)?;
+        if endpoint.scheme() != "https"
+            || endpoint.cannot_be_a_base()
+            || !endpoint.username().is_empty()
+            || endpoint.password().is_some()
+            || endpoint.query().is_some()
+            || endpoint.fragment().is_some()
+        {
+            return Err(HeartbeatError::Endpoint);
+        }
+        if !endpoint.path().ends_with('/') {
+            endpoint.set_path(&format!("{}/", endpoint.path()));
+        }
+        let roots = CertificateDer::pem_slice_iter(&config.root_ca_pem)
+            .collect::<Result<Vec<_>, _>>()
+            .map_err(|_| HeartbeatError::RootCertificate)?;
+        if roots.is_empty() {
+            return Err(HeartbeatError::RootCertificate);
+        }
+        let mut root_store = RootCertStore::empty();
+        let (accepted, rejected) = root_store.add_parsable_certificates(roots.clone());
+        if accepted != roots.len() || rejected != 0 {
+            return Err(HeartbeatError::RootCertificate);
+        }
         let schedule = config.schedule;
-        if schedule.cadence.is_zero() || schedule.jitter > schedule.cadence {
+        if schedule.cadence.is_zero()
+            || schedule.timeout.is_zero()
+            || schedule.timeout > Duration::from_secs(5)
+            || schedule.initial_backoff.is_zero()
+            || schedule.max_backoff < schedule.initial_backoff
+            || schedule.max_backoff > Duration::from_secs(5 * 60)
+            || schedule.jitter > schedule.cadence
+        {
             return Err(HeartbeatError::Schedule);
         }
         Ok(Self {
-            transport: TelemetryTransport::new(config)?,
+            endpoint,
+            root_store,
+            roots,
+            config,
         })
     }
 
     pub(crate) async fn send(&self, heartbeat: &PendingHeartbeat) -> Result<Delivery, HeartbeatError> {
-        match self.transport.post("heartbeats", heartbeat).await? {
-            TelemetryDelivery::Accepted { body, .. } => {
-                let accepted: HeartbeatResponse = serde_json::from_slice(&body).map_err(|_| HeartbeatError::Response)?;
-                if accepted.accepted_version != PROTOCOL_VERSION
-                    || accepted.capability_hints.len() > 32
-                    || accepted.capability_hints.iter().any(|hint| hint.len() > 32)
-                    || !is_exact_utc_seconds(&accepted.server_time)
-                {
-                    return Err(HeartbeatError::Response);
-                }
-                Ok(Delivery::Accepted {
-                    server_time: accepted.server_time,
-                })
+        let (cluster_uid, client) = {
+            let _lock = self.config.credential_store.lock().await?;
+            let credential = self.config.credential_store.load()?.ok_or(HeartbeatError::NotRegistered)?;
+            let identity = self.config.identity_store.load()?.ok_or(HeartbeatError::IdentityMissing)?;
+            validate_stored_credential(&credential, &identity, &self.root_store, &self.roots)?;
+            let now = Utc::now().timestamp();
+            if now < credential.not_before_unix || now >= credential.not_after_unix {
+                return Err(HeartbeatError::CredentialExpired);
             }
-            TelemetryDelivery::Retry { retry_after } => Ok(Delivery::Retry { retry_after }),
-            TelemetryDelivery::AuthenticationStopped { status, reason } => Ok(Delivery::AuthenticationStopped { status, reason }),
-            TelemetryDelivery::Rejected { status, reason } => Ok(Delivery::Rejected { status, reason }),
+            let cluster_uid = cluster_uid(&credential)?.to_owned();
+            let client = self.client(&credential, &identity.to_pkcs8_pem()?)?;
+            (cluster_uid, client)
+        };
+        let url = self.endpoint.join(&format!("clusters/{cluster_uid}/heartbeats"))?;
+        let response = match client.post(url).json(heartbeat).send().await {
+            Ok(response) => response,
+            Err(error) if error.is_timeout() || error.is_connect() || error.is_request() => {
+                return Ok(Delivery::Retry { retry_after: None });
+            }
+            Err(error) => return Err(error.into()),
+        };
+        let status = response.status();
+        if status == StatusCode::TOO_MANY_REQUESTS {
+            return Ok(Delivery::Retry {
+                retry_after: retry_after(response.headers(), Utc::now(), self.config.schedule.max_backoff),
+            });
         }
+        if status == StatusCode::REQUEST_TIMEOUT || status.is_server_error() {
+            return Ok(Delivery::Retry { retry_after: None });
+        }
+        if matches!(status, StatusCode::UNAUTHORIZED | StatusCode::FORBIDDEN) {
+            return Ok(Delivery::AuthenticationStopped {
+                status: status.as_u16(),
+                reason: response_reason(response).await,
+            });
+        }
+        if status != StatusCode::OK {
+            return Ok(Delivery::Rejected {
+                status: status.as_u16(),
+                reason: response_reason(response).await,
+            });
+        }
+        let accepted: HeartbeatResponse =
+            serde_json::from_slice(&bounded_body(response).await?).map_err(|_| HeartbeatError::Response)?;
+        if accepted.accepted_version != PROTOCOL_VERSION
+            || accepted.capability_hints.len() > 32
+            || accepted.capability_hints.iter().any(|hint| hint.len() > 32)
+            || !is_exact_utc_seconds(&accepted.server_time)
+        {
+            return Err(HeartbeatError::Response);
+        }
+        Ok(Delivery::Accepted {
+            server_time: accepted.server_time,
+        })
+    }
+
+    fn client(&self, credential: &DeviceCredential, key: &Zeroizing<String>) -> Result<Client, HeartbeatError> {
+        let mut pem = Zeroizing::new(Vec::with_capacity(credential.certificate_chain.len() + key.len() + 1));
+        pem.extend_from_slice(credential.certificate_chain.as_bytes());
+        pem.push(b'\n');
+        pem.extend_from_slice(key.as_bytes());
+        let identity = reqwest::Identity::from_pem(&pem).map_err(|_| HeartbeatError::IdentityCertificate)?;
+        let roots = self
+            .roots
+            .iter()
+            .map(|root| reqwest::Certificate::from_der(root.as_ref()))
+            .collect::<Result<Vec<_>, _>>()?;
+        Client::builder()
+            .https_only(true)
+            .redirect(reqwest::redirect::Policy::none())
+            .timeout(self.config.schedule.timeout)
+            .tls_certs_only(roots)
+            .identity(identity)
+            .build()
+            .map_err(Into::into)
     }
 }
 
@@ -280,6 +378,73 @@ impl HeartbeatStateStore {
     }
 }
 
+fn cluster_uid(credential: &DeviceCredential) -> Result<&str, HeartbeatError> {
+    let mut parts = credential.name.split('/');
+    let valid = parts.next() == Some("organizations");
+    let organization_uid = parts.next();
+    let valid = valid && parts.next() == Some("clusters");
+    let cluster_uid = parts.next();
+    let valid = valid && parts.next() == Some("clusterDevices");
+    let device_uid = parts.next();
+    if !valid
+        || organization_uid.is_none_or(str::is_empty)
+        || cluster_uid.is_none_or(str::is_empty)
+        || device_uid != Some(credential.uid.as_str())
+        || parts.next().is_some()
+    {
+        return Err(HeartbeatError::CredentialName);
+    }
+    cluster_uid.ok_or(HeartbeatError::CredentialName)
+}
+
+fn retry_after(headers: &header::HeaderMap, now: DateTime<Utc>, maximum: Duration) -> Option<Duration> {
+    let value = headers.get(header::RETRY_AFTER)?.to_str().ok()?;
+    let delay = value.parse::<u64>().ok().map(Duration::from_secs).or_else(|| {
+        DateTime::parse_from_rfc2822(value)
+            .ok()
+            .and_then(|at| (at.with_timezone(&Utc) - now).to_std().ok())
+    })?;
+    Some(delay.min(maximum))
+}
+
+fn is_exact_utc_seconds(value: &str) -> bool {
+    DateTime::parse_from_rfc3339(value).is_ok_and(|time| {
+        time.offset().local_minus_utc() == 0
+            && value.ends_with('Z')
+            && time.with_timezone(&Utc).to_rfc3339_opts(SecondsFormat::Secs, true) == value
+    })
+}
+
+async fn response_reason(response: reqwest::Response) -> Option<String> {
+    #[derive(Deserialize)]
+    struct Envelope {
+        #[serde(default)]
+        details: Vec<Detail>,
+    }
+    #[derive(Deserialize)]
+    struct Detail {
+        #[serde(default)]
+        reason: String,
+    }
+
+    serde_json::from_slice::<Envelope>(&bounded_body(response).await.ok()?)
+        .ok()?
+        .details
+        .into_iter()
+        .find_map(|detail| (!detail.reason.is_empty()).then_some(detail.reason))
+}
+
+async fn bounded_body(mut response: reqwest::Response) -> Result<Vec<u8>, HeartbeatError> {
+    let mut body = Vec::new();
+    while let Some(chunk) = response.chunk().await? {
+        if body.len().saturating_add(chunk.len()) > MAX_RESPONSE_BYTES {
+            return Err(HeartbeatError::ResponseTooLarge);
+        }
+        body.extend_from_slice(&chunk);
+    }
+    Ok(body)
+}
+
 fn parent(path: &Path) -> Result<&Path, HeartbeatError> {
     path.parent()
         .ok_or_else(|| state_io(path, io::Error::new(io::ErrorKind::InvalidInput, "state path has no parent")))
@@ -418,25 +583,3 @@ pub enum HeartbeatError {
     #[error(transparent)]
     CredentialValidation(#[from] CredentialValidationError),
 }
-
-impl From<TelemetryError> for HeartbeatError {
-    fn from(error: TelemetryError) -> Self {
-        match error {
-            TelemetryError::Endpoint => Self::Endpoint,
-            TelemetryError::RootCertificate => Self::RootCertificate,
-            TelemetryError::Schedule => Self::Schedule,
-            TelemetryError::NotRegistered => Self::NotRegistered,
-            TelemetryError::IdentityMissing => Self::IdentityMissing,
-            TelemetryError::IdentityCertificate => Self::IdentityCertificate,
-            TelemetryError::CredentialName => Self::CredentialName,
-            TelemetryError::CredentialExpired => Self::CredentialExpired,
-            TelemetryError::ResponseTooLarge => Self::ResponseTooLarge,
-            TelemetryError::Url(error) => Self::Url(error),
-            TelemetryError::Transport(error) => Self::Transport(error),
-            TelemetryError::Identity(error) => Self::Identity(error),
-            TelemetryError::IdentityStore(error) => Self::IdentityStore(error),
-            TelemetryError::CredentialStore(error) => Self::CredentialStore(error),
-            TelemetryError::CredentialValidation(error) => Self::CredentialValidation(error),
-        }
-    }
-}

rustfs/src/connect/inventory.rs

@@ -1,594 +0,0 @@
-// Copyright 2024 RustFS Team
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-use std::collections::BTreeSet;
-use std::fs;
-use std::io::{self, Write as _};
-use std::path::{Path, PathBuf};
-use std::sync::atomic::{AtomicU64, Ordering};
-use std::time::Duration;
-
-use serde::{Deserialize, Serialize};
-use sha2::{Digest as _, Sha256};
-use uuid::Uuid;
-
-use super::config::HeartbeatConfig;
-use super::telemetry::{TelemetryDelivery, TelemetryError, TelemetryTransport, is_exact_utc_seconds};
-
-const PROTOCOL_VERSION: &str = "v1";
-const RUSTFS_VERSION: &str = concat!(
-    env!("CARGO_PKG_VERSION_MAJOR"),
-    ".",
-    env!("CARGO_PKG_VERSION_MINOR"),
-    ".",
-    env!("CARGO_PKG_VERSION_PATCH")
-);
-const HASH_PREFIX: &[u8] = b"rustfs-connect/agent/v1/inventory-snapshot\n";
-const MAX_SEQUENCE: u64 = 9_007_199_254_740_991;
-const MAX_SAFE_INTEGER: u64 = 9_007_199_254_740_991;
-#[cfg(unix)]
-const FILE_MODE: u32 = 0o600;
-static STAGING_SEQUENCE: AtomicU64 = AtomicU64::new(0);
-
-#[derive(Clone, Copy, Debug, PartialEq, Eq)]
-pub struct InventorySchedule {
-    pub cadence: Duration,
-    pub jitter: Duration,
-}
-
-impl Default for InventorySchedule {
-    fn default() -> Self {
-        Self {
-            cadence: Duration::from_secs(6 * 60 * 60),
-            jitter: Duration::from_secs(30 * 60),
-        }
-    }
-}
-
-#[derive(Clone, Debug, PartialEq, Eq)]
-pub enum InventoryStatus {
-    Starting,
-    Unchanged { content_hash: String },
-    Online { content_hash: String, received_at: String },
-    BackingOff { delay: Duration },
-    AuthenticationStopped { status: u16, reason: Option<String> },
-    Failed { reason: String },
-    Stopped,
-}
-
-#[derive(Clone, Copy, Debug, PartialEq, Eq, PartialOrd, Ord, Serialize, Deserialize)]
-pub enum InventoryFlag {
-    #[serde(rename = "capacity.critical")]
-    CapacityCritical,
-    #[serde(rename = "capacity.warning")]
-    CapacityWarning,
-    #[serde(rename = "clock.skew")]
-    ClockSkew,
-    #[serde(rename = "cluster.degraded")]
-    ClusterDegraded,
-    #[serde(rename = "cluster.healing")]
-    ClusterHealing,
-    #[serde(rename = "cluster.readonly")]
-    ClusterReadonly,
-    #[serde(rename = "drive.offline")]
-    DriveOffline,
-    #[serde(rename = "node.offline")]
-    NodeOffline,
-}
-
-#[derive(Clone, Copy, Debug, PartialEq, Eq, Serialize, Deserialize)]
-#[serde(rename_all = "lowercase")]
-pub enum OperatingSystemFamily {
-    Linux,
-    Darwin,
-    Windows,
-    Freebsd,
-    Other,
-}
-
-#[derive(Clone, Copy, Debug, PartialEq, Eq, Serialize, Deserialize)]
-#[serde(deny_unknown_fields, rename_all = "camelCase")]
-pub struct InventoryOsVersion {
-    family: OperatingSystemFamily,
-    major: u16,
-    minor: u16,
-}
-
-impl InventoryOsVersion {
-    pub fn new(family: OperatingSystemFamily, major: u16, minor: u16) -> Result<Self, InventoryError> {
-        if major > 9999 || minor > 9999 {
-            return Err(InventoryError::OsVersion);
-        }
-        Ok(Self { family, major, minor })
-    }
-}
-
-#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize)]
-#[serde(deny_unknown_fields, rename_all = "camelCase")]
-pub struct InventorySnapshot {
-    rustfs_version: String,
-    os_version: Option<InventoryOsVersion>,
-    node_count: u16,
-    drive_count: u32,
-    capacity_total_bytes: u64,
-    capacity_used_bytes: u64,
-    coarse_flags: Vec<InventoryFlag>,
-}
-
-impl InventorySnapshot {
-    pub fn current(
-        node_count: usize,
-        drive_count: usize,
-        capacity_total_bytes: u64,
-        capacity_free_bytes: u64,
-        coarse_flags: impl IntoIterator<Item = InventoryFlag>,
-    ) -> Result<Self, InventoryError> {
-        let capacity_used_bytes = capacity_total_bytes
-            .checked_sub(capacity_free_bytes)
-            .ok_or(InventoryError::Capacity)?;
-        Self::new(
-            RUSTFS_VERSION,
-            None,
-            node_count,
-            drive_count,
-            capacity_total_bytes,
-            capacity_used_bytes,
-            coarse_flags,
-        )
-    }
-
-    pub fn new(
-        rustfs_version: impl Into<String>,
-        os_version: Option<InventoryOsVersion>,
-        node_count: usize,
-        drive_count: usize,
-        capacity_total_bytes: u64,
-        capacity_used_bytes: u64,
-        coarse_flags: impl IntoIterator<Item = InventoryFlag>,
-    ) -> Result<Self, InventoryError> {
-        let snapshot = Self {
-            rustfs_version: rustfs_version.into(),
-            os_version,
-            node_count: u16::try_from(node_count).map_err(|_| InventoryError::NodeCount)?,
-            drive_count: u32::try_from(drive_count).map_err(|_| InventoryError::DriveCount)?,
-            capacity_total_bytes,
-            capacity_used_bytes,
-            coarse_flags: coarse_flags.into_iter().collect::<BTreeSet<_>>().into_iter().collect(),
-        };
-        snapshot.validate()?;
-        Ok(snapshot)
-    }
-
-    pub fn content_hash(&self) -> Result<String, InventoryError> {
-        #[derive(Serialize)]
-        #[serde(rename_all = "camelCase")]
-        struct Canonical<'a> {
-            capacity_total_bytes: u64,
-            capacity_used_bytes: u64,
-            coarse_flags: &'a [InventoryFlag],
-            drive_count: u32,
-            node_count: u16,
-            os_version: Option<InventoryOsVersion>,
-            rustfs_version: &'a str,
-        }
-
-        let canonical = serde_json::to_vec(&Canonical {
-            capacity_total_bytes: self.capacity_total_bytes,
-            capacity_used_bytes: self.capacity_used_bytes,
-            coarse_flags: &self.coarse_flags,
-            drive_count: self.drive_count,
-            node_count: self.node_count,
-            os_version: self.os_version,
-            rustfs_version: &self.rustfs_version,
-        })?;
-        let mut digest = Sha256::new();
-        digest.update(HASH_PREFIX);
-        digest.update(canonical);
-        Ok(hex_simd::encode_to_string(digest.finalize(), hex_simd::AsciiCase::Lower))
-    }
-
-    fn validate(&self) -> Result<(), InventoryError> {
-        if !valid_version(&self.rustfs_version) {
-            return Err(InventoryError::RustfsVersion);
-        }
-        if self.node_count == 0 || self.node_count > 4096 {
-            return Err(InventoryError::NodeCount);
-        }
-        if self.drive_count > 1_048_576 {
-            return Err(InventoryError::DriveCount);
-        }
-        if self.capacity_total_bytes > MAX_SAFE_INTEGER || self.capacity_used_bytes > self.capacity_total_bytes {
-            return Err(InventoryError::Capacity);
-        }
-        Ok(())
-    }
-}
-
-fn valid_version(version: &str) -> bool {
-    let components = version.split('.').collect::<Vec<_>>();
-    components.len() == 3
-        && components.iter().all(|component| {
-            !component.is_empty()
-                && component.len() <= 4
-                && (component == &"0" || !component.starts_with('0'))
-                && component.parse::<u16>().is_ok_and(|value| value <= 9999)
-        })
-}
-
-#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize)]
-#[serde(deny_unknown_fields, rename_all = "camelCase")]
-pub(crate) struct PendingInventory {
-    protocol_version: String,
-    request_id: String,
-    sequence: u64,
-    #[serde(flatten)]
-    snapshot: InventorySnapshot,
-}
-
-impl PendingInventory {
-    fn new(snapshot: InventorySnapshot, sequence: u64) -> Self {
-        Self {
-            protocol_version: PROTOCOL_VERSION.to_owned(),
-            request_id: Uuid::new_v4().to_string(),
-            sequence,
-            snapshot,
-        }
-    }
-
-    fn is_valid(&self) -> bool {
-        self.protocol_version == PROTOCOL_VERSION
-            && self.sequence <= MAX_SEQUENCE
-            && self.snapshot.validate().is_ok()
-            && Uuid::parse_str(&self.request_id)
-                .is_ok_and(|request_id| request_id.get_version_num() == 4 && request_id.to_string() == self.request_id)
-    }
-
-    fn content_hash(&self) -> Result<String, InventoryError> {
-        self.snapshot.content_hash()
-    }
-}
-
-pub(crate) enum InventoryDelivery {
-    Accepted { content_hash: String, received_at: String },
-    Retry { retry_after: Option<Duration> },
-    AuthenticationStopped { status: u16, reason: Option<String> },
-    Rejected { status: u16, reason: Option<String> },
-}
-
-pub(crate) struct InventorySender {
-    transport: TelemetryTransport,
-}
-
-impl InventorySender {
-    pub(crate) fn new(config: HeartbeatConfig) -> Result<Self, InventoryError> {
-        Ok(Self {
-            transport: TelemetryTransport::new(config)?,
-        })
-    }
-
-    pub(crate) async fn send(&self, inventory: &PendingInventory) -> Result<InventoryDelivery, InventoryError> {
-        match self.transport.post("inventorySnapshots", inventory).await? {
-            TelemetryDelivery::Accepted { cluster_name, body } => {
-                #[derive(Deserialize)]
-                #[serde(rename_all = "camelCase")]
-                struct InventoryResponse {
-                    name: String,
-                    uid: String,
-                    content_hash: String,
-                    received_at: String,
-                }
-
-                let accepted: InventoryResponse = serde_json::from_slice(&body).map_err(|_| InventoryError::Response)?;
-                let uid = Uuid::parse_str(&accepted.uid).map_err(|_| InventoryError::Response)?;
-                let content_hash = inventory.content_hash()?;
-                if uid.get_version_num() != 7
-                    || uid.to_string() != accepted.uid
-                    || accepted.name != format!("{cluster_name}/inventorySnapshots/{}", accepted.uid)
-                    || accepted.content_hash != content_hash
-                    || !is_exact_utc_seconds(&accepted.received_at)
-                {
-                    return Err(InventoryError::Response);
-                }
-                Ok(InventoryDelivery::Accepted {
-                    content_hash,
-                    received_at: accepted.received_at,
-                })
-            }
-            TelemetryDelivery::Retry { retry_after } => Ok(InventoryDelivery::Retry { retry_after }),
-            TelemetryDelivery::AuthenticationStopped { status, reason } => {
-                Ok(InventoryDelivery::AuthenticationStopped { status, reason })
-            }
-            TelemetryDelivery::Rejected { status, reason } => Ok(InventoryDelivery::Rejected { status, reason }),
-        }
-    }
-}
-
-#[derive(Clone)]
-pub(crate) struct InventoryStateStore {
-    path: PathBuf,
-}
-
-#[derive(Default, Serialize, Deserialize)]
-#[serde(deny_unknown_fields, rename_all = "camelCase")]
-struct InventoryState {
-    next_sequence: u64,
-    pending: Option<PendingInventory>,
-    last_accepted_content_hash: Option<String>,
-}
-
-impl InventoryStateStore {
-    pub(crate) fn from_heartbeat_path(path: &Path) -> Result<Self, InventoryError> {
-        let root = path.parent().and_then(Path::parent).ok_or(InventoryError::StatePath)?;
-        Ok(Self {
-            path: root.join("inventory/state.json"),
-        })
-    }
-
-    pub(crate) fn try_runtime_lock(&self) -> Result<fs::File, InventoryError> {
-        let directory = parent(&self.path)?;
-        fs::create_dir_all(directory).map_err(|source| state_io(directory, source))?;
-        let name = filename(&self.path)?;
-        let path = directory.join(format!(".{name}.lock"));
-        let mut options = fs::OpenOptions::new();
-        options.create(true).truncate(false).read(true).write(true);
-        #[cfg(unix)]
-        {
-            use std::os::unix::fs::OpenOptionsExt as _;
-            options.mode(FILE_MODE);
-        }
-        let lock = options.open(&path).map_err(|source| state_io(&path, source))?;
-        check_mode(&path)?;
-        lock.try_lock().map_err(|_| InventoryError::AlreadyRunning)?;
-        Ok(lock)
-    }
-
-    pub(crate) async fn pending(&self) -> Result<Option<PendingInventory>, InventoryError> {
-        let store = self.clone();
-        tokio::task::spawn_blocking(move || {
-            let state = store.read()?;
-            if state.pending.is_none() && state.next_sequence > MAX_SEQUENCE {
-                return Err(InventoryError::SequenceExhausted);
-            }
-            Ok(state.pending)
-        })
-        .await
-        .map_err(|source| state_io(&self.path, io::Error::other(source)))?
-    }
-
-    pub(crate) async fn prepare(&self, snapshot: InventorySnapshot) -> Result<Option<PendingInventory>, InventoryError> {
-        let store = self.clone();
-        tokio::task::spawn_blocking(move || store.prepare_sync(snapshot))
-            .await
-            .map_err(|source| state_io(&self.path, io::Error::other(source)))?
-    }
-
-    pub(crate) async fn mark_accepted(&self, accepted: &PendingInventory) -> Result<(), InventoryError> {
-        let store = self.clone();
-        let accepted = accepted.clone();
-        tokio::task::spawn_blocking(move || store.mark_accepted_sync(&accepted))
-            .await
-            .map_err(|source| state_io(&self.path, io::Error::other(source)))?
-    }
-
-    fn prepare_sync(&self, snapshot: InventorySnapshot) -> Result<Option<PendingInventory>, InventoryError> {
-        let mut state = self.read()?;
-        if state.pending.is_some() {
-            return Ok(state.pending);
-        }
-        let content_hash = snapshot.content_hash()?;
-        if state.last_accepted_content_hash.as_deref() == Some(&content_hash) {
-            return Ok(None);
-        }
-        if state.next_sequence > MAX_SEQUENCE {
-            return Err(InventoryError::SequenceExhausted);
-        }
-        let pending = PendingInventory::new(snapshot, state.next_sequence);
-        state.pending = Some(pending.clone());
-        self.write(&state)?;
-        Ok(Some(pending))
-    }
-
-    fn mark_accepted_sync(&self, accepted: &PendingInventory) -> Result<(), InventoryError> {
-        let mut state = self.read()?;
-        if state.pending.as_ref() != Some(accepted) {
-            return Err(InventoryError::StateConflict);
-        }
-        state.next_sequence = accepted.sequence.checked_add(1).ok_or(InventoryError::SequenceExhausted)?;
-        state.last_accepted_content_hash = Some(accepted.content_hash()?);
-        state.pending = None;
-        self.write(&state)
-    }
-
-    fn read(&self) -> Result<InventoryState, InventoryError> {
-        let bytes = match fs::read(&self.path) {
-            Ok(bytes) => bytes,
-            Err(source) if source.kind() == io::ErrorKind::NotFound => return Ok(InventoryState::default()),
-            Err(source) => return Err(state_io(&self.path, source)),
-        };
-        check_mode(&self.path)?;
-        let state: InventoryState = serde_json::from_slice(&bytes).map_err(|source| InventoryError::StateInvalid {
-            path: self.path.clone(),
-            source,
-        })?;
-        let last_hash_valid = state.last_accepted_content_hash.as_deref().is_none_or(valid_content_hash);
-        let pending_valid = state.pending.as_ref().is_none_or(|pending| {
-            pending.sequence == state.next_sequence
-                && pending.is_valid()
-                && pending
-                    .content_hash()
-                    .is_ok_and(|hash| state.last_accepted_content_hash.as_deref() != Some(&hash))
-        });
-        if state.next_sequence > MAX_SEQUENCE + 1 || !last_hash_valid || !pending_valid {
-            return Err(InventoryError::StateCorrupt { path: self.path.clone() });
-        }
-        Ok(state)
-    }
-
-    fn write(&self, state: &InventoryState) -> Result<(), InventoryError> {
-        let bytes = serde_json::to_vec(state).map_err(|source| InventoryError::StateInvalid {
-            path: self.path.clone(),
-            source,
-        })?;
-        let directory = parent(&self.path)?;
-        fs::create_dir_all(directory).map_err(|source| state_io(directory, source))?;
-        let temp = stage(directory, &self.path, &bytes)?;
-        let result = fs::rename(&temp, &self.path)
-            .map_err(|source| state_io(&self.path, source))
-            .and_then(|()| fsync_dir(directory).map_err(|source| state_io(directory, source)));
-        if result.is_err() {
-            let _ = fs::remove_file(temp);
-        }
-        result
-    }
-}
-
-fn valid_content_hash(value: &str) -> bool {
-    value.len() == 64
-        && value
-            .bytes()
-            .all(|byte| byte.is_ascii_digit() || (b'a'..=b'f').contains(&byte))
-}
-
-fn parent(path: &Path) -> Result<&Path, InventoryError> {
-    path.parent()
-        .ok_or_else(|| state_io(path, io::Error::new(io::ErrorKind::InvalidInput, "state path has no parent")))
-}
-
-fn filename(path: &Path) -> Result<&str, InventoryError> {
-    path.file_name()
-        .and_then(|name| name.to_str())
-        .ok_or_else(|| state_io(path, io::Error::new(io::ErrorKind::InvalidInput, "state filename is invalid")))
-}
-
-fn stage(directory: &Path, destination: &Path, bytes: &[u8]) -> Result<PathBuf, InventoryError> {
-    let name = filename(destination)?;
-    loop {
-        let path = directory.join(format!(
-            ".{name}.{}.{}.tmp",
-            std::process::id(),
-            STAGING_SEQUENCE.fetch_add(1, Ordering::Relaxed)
-        ));
-        let mut options = fs::OpenOptions::new();
-        options.write(true).create_new(true);
-        #[cfg(unix)]
-        {
-            use std::os::unix::fs::OpenOptionsExt as _;
-            options.mode(FILE_MODE);
-        }
-        let mut file = match options.open(&path) {
-            Ok(file) => file,
-            Err(source) if source.kind() == io::ErrorKind::AlreadyExists => continue,
-            Err(source) => return Err(state_io(&path, source)),
-        };
-        if let Err(source) = file.write_all(bytes).and_then(|()| file.sync_all()) {
-            let _ = fs::remove_file(&path);
-            return Err(state_io(&path, source));
-        }
-        return Ok(path);
-    }
-}
-
-fn state_io(path: &Path, source: io::Error) -> InventoryError {
-    InventoryError::StateIo {
-        path: path.to_path_buf(),
-        source,
-    }
-}
-
-#[cfg(unix)]
-fn check_mode(path: &Path) -> Result<(), InventoryError> {
-    use std::os::unix::fs::PermissionsExt as _;
-
-    let mode = fs::metadata(path)
-        .map_err(|source| state_io(path, source))?
-        .permissions()
-        .mode()
-        & 0o7777;
-    if mode != FILE_MODE {
-        return Err(InventoryError::StatePermissions {
-            path: path.to_path_buf(),
-            mode,
-            expected: FILE_MODE,
-        });
-    }
-    Ok(())
-}
-
-#[cfg(not(unix))]
-fn check_mode(_path: &Path) -> Result<(), InventoryError> {
-    Ok(())
-}
-
-fn fsync_dir(directory: &Path) -> io::Result<()> {
-    #[cfg(unix)]
-    fs::File::open(directory)?.sync_all()?;
-    #[cfg(not(unix))]
-    let _ = directory;
-    Ok(())
-}
-
-#[derive(Debug, thiserror::Error)]
-pub enum InventoryError {
-    #[error("the RustFS inventory version is outside protocol bounds")]
-    RustfsVersion,
-    #[error("the RustFS inventory operating-system version is outside protocol bounds")]
-    OsVersion,
-    #[error("the RustFS inventory node count is outside protocol bounds")]
-    NodeCount,
-    #[error("the RustFS inventory drive count is outside protocol bounds")]
-    DriveCount,
-    #[error("the RustFS inventory capacity is outside protocol bounds")]
-    Capacity,
-    #[error("the RustFS inventory snapshot is incomplete: observed {observed} of {expected} configured drives")]
-    SnapshotIncomplete { expected: usize, observed: usize },
-    #[error("the Connect inventory schedule is invalid")]
-    Schedule,
-    #[error("the Connect inventory sequence is exhausted")]
-    SequenceExhausted,
-    #[error("a Connect inventory runtime already owns this state")]
-    AlreadyRunning,
-    #[error("the persisted Connect inventory changed while delivery was in flight")]
-    StateConflict,
-    #[error("the Connect inventory state path is invalid")]
-    StatePath,
-    #[error("Connect inventory state I/O failed at {path}: {source}")]
-    StateIo {
-        path: PathBuf,
-        #[source]
-        source: io::Error,
-    },
-    #[error("Connect inventory state at {path} is invalid: {source}")]
-    StateInvalid {
-        path: PathBuf,
-        #[source]
-        source: serde_json::Error,
-    },
-    #[error("Connect inventory state at {path} violates the protocol invariants")]
-    StateCorrupt { path: PathBuf },
-    #[cfg(unix)]
-    #[error("Connect inventory state at {path} has mode {mode:o}, expected {expected:o}")]
-    StatePermissions { path: PathBuf, mode: u32, expected: u32 },
-    #[error("Connect returned an invalid inventory response")]
-    Response,
-    #[error(transparent)]
-    Json(#[from] serde_json::Error),
-    #[error("Connect inventory delivery failed: {0}")]
-    Telemetry(String),
-}
-
-impl From<TelemetryError> for InventoryError {
-    fn from(error: TelemetryError) -> Self {
-        Self::Telemetry(error.to_string())
-    }
-}

rustfs/src/connect/mod.rs

@@ -31,11 +31,9 @@ pub mod credential_store;
 pub mod heartbeat;
 pub mod identity;
 pub mod identity_store;
-pub mod inventory;
 pub mod offline;
 pub mod registration;
 pub mod runtime;
-mod telemetry;
 
 pub use client::{ClientError, ConnectClient, ConnectConfig};
 pub use config::{HeartbeatConfig, HeartbeatConfigError, HeartbeatSchedule};
@@ -43,10 +41,6 @@ pub use credential_store::{CredentialStore, DeviceCredential};
 pub use heartbeat::{CoarseNodeSummary, HeartbeatError, HeartbeatStatus};
 pub use identity::{DeviceIdentity, IdentityError, RegistrationProof, RegistrationTranscript};
 pub use identity_store::{IdentityStore, StoreError};
-pub use inventory::{
-    InventoryError, InventoryFlag, InventoryOsVersion, InventorySchedule, InventorySnapshot, InventoryStatus,
-    OperatingSystemFamily,
-};
 pub use offline::{EnrollmentError, OfflineEnrollment, OfflineKeyStore, VerifiedChallenge};
 pub use registration::{RegistrationToken, TokenError};
-pub use runtime::{HeartbeatRuntime, InventoryRuntime, spawn_heartbeat_runtime, spawn_inventory_runtime};
+pub use runtime::{HeartbeatRuntime, spawn_heartbeat_runtime};

rustfs/src/connect/runtime.rs

@@ -23,16 +23,11 @@ use tokio_util::sync::CancellationToken;
 
 use super::config::HeartbeatConfig;
 use super::heartbeat::{CoarseNodeSummary, Delivery, HeartbeatError, HeartbeatSender, HeartbeatStateStore, HeartbeatStatus};
-use super::inventory::{
-    InventoryDelivery, InventoryError, InventorySchedule, InventorySender, InventorySnapshot, InventoryStateStore,
-    InventoryStatus,
-};
 
 pub struct HeartbeatRuntime {
     shutdown: CancellationToken,
     status: watch::Receiver<HeartbeatStatus>,
     task: Option<JoinHandle<()>>,
-    inventory: Option<InventoryRuntime>,
 }
 
 impl HeartbeatRuntime {
@@ -40,19 +35,11 @@ impl HeartbeatRuntime {
         self.status.clone()
     }
 
-    pub(crate) fn with_inventory(mut self, inventory: Option<InventoryRuntime>) -> Self {
-        self.inventory = inventory;
-        self
-    }
-
     pub async fn shutdown(mut self) {
         self.shutdown.cancel();
         if let Some(task) = self.task.take() {
             let _ = task.await;
         }
-        if let Some(inventory) = self.inventory.take() {
-            inventory.shutdown().await;
-        }
     }
 }
 
@@ -62,31 +49,6 @@ impl Drop for HeartbeatRuntime {
     }
 }
 
-pub struct InventoryRuntime {
-    shutdown: CancellationToken,
-    status: watch::Receiver<InventoryStatus>,
-    task: Option<JoinHandle<()>>,
-}
-
-impl InventoryRuntime {
-    pub fn status(&self) -> watch::Receiver<InventoryStatus> {
-        self.status.clone()
-    }
-
-    pub async fn shutdown(mut self) {
-        self.shutdown.cancel();
-        if let Some(task) = self.task.take() {
-            let _ = task.await;
-        }
-    }
-}
-
-impl Drop for InventoryRuntime {
-    fn drop(&mut self) {
-        self.shutdown.cancel();
-    }
-}
-
 pub fn spawn_heartbeat_runtime<F>(
     config: Option<HeartbeatConfig>,
     parent_shutdown: &CancellationToken,
@@ -160,126 +122,6 @@ where
         shutdown,
         status: status_rx,
         task: Some(task),
-        inventory: None,
-    }))
-}
-
-pub fn spawn_inventory_runtime<F, Fut>(
-    config: Option<HeartbeatConfig>,
-    schedule: InventorySchedule,
-    parent_shutdown: &CancellationToken,
-    sample: F,
-) -> Result<Option<InventoryRuntime>, InventoryError>
-where
-    F: Fn() -> Fut + Send + Sync + 'static,
-    Fut: Future<Output = Result<InventorySnapshot, InventoryError>> + Send + 'static,
-{
-    let Some(config) = config else {
-        return Ok(None);
-    };
-    if schedule.cadence.is_zero() || schedule.jitter > schedule.cadence {
-        return Err(InventoryError::Schedule);
-    }
-    let retry_schedule = config.schedule;
-    let store = InventoryStateStore::from_heartbeat_path(&config.state_path)?;
-    let lock = store.try_runtime_lock()?;
-    let sender = InventorySender::new(config)?;
-    let shutdown = parent_shutdown.child_token();
-    let task_shutdown = shutdown.clone();
-    let (status_tx, status_rx) = watch::channel(InventoryStatus::Starting);
-    let task = tokio::spawn(async move {
-        let _lock = lock;
-        let mut backoff = retry_schedule.initial_backoff;
-        loop {
-            if task_shutdown.is_cancelled() {
-                break;
-            }
-            let pending = match store.pending().await {
-                Ok(Some(pending)) => pending,
-                Ok(None) => {
-                    let snapshot = match cancellable(&task_shutdown, sample()).await {
-                        Some(Ok(snapshot)) => snapshot,
-                        Some(Err(InventoryError::SnapshotIncomplete { .. })) => {
-                            let delay = backoff;
-                            backoff = backoff.saturating_mul(2).min(retry_schedule.max_backoff);
-                            let _ = status_tx.send(InventoryStatus::BackingOff { delay });
-                            if sleep_or_cancel(&task_shutdown, delay).await {
-                                break;
-                            }
-                            continue;
-                        }
-                        Some(Err(error)) => return failed_inventory(&status_tx, error),
-                        None => break,
-                    };
-                    let content_hash = match snapshot.content_hash() {
-                        Ok(content_hash) => content_hash,
-                        Err(error) => return failed_inventory(&status_tx, error),
-                    };
-                    match store.prepare(snapshot).await {
-                        Ok(Some(pending)) => pending,
-                        Ok(None) => {
-                            backoff = retry_schedule.initial_backoff;
-                            let _ = status_tx.send(InventoryStatus::Unchanged { content_hash });
-                            if sleep_or_cancel(&task_shutdown, schedule.cadence.saturating_add(jitter(schedule.jitter))).await {
-                                break;
-                            }
-                            continue;
-                        }
-                        Err(error) => return failed_inventory(&status_tx, error),
-                    }
-                }
-                Err(error) => return failed_inventory(&status_tx, error),
-            };
-            let delivery = match cancellable(&task_shutdown, sender.send(&pending)).await {
-                Some(Ok(delivery)) => delivery,
-                Some(Err(error)) => return failed_inventory(&status_tx, error),
-                None => break,
-            };
-            let delay = match delivery {
-                InventoryDelivery::Accepted {
-                    content_hash,
-                    received_at,
-                } => {
-                    if let Err(error) = store.mark_accepted(&pending).await {
-                        return failed_inventory(&status_tx, error);
-                    }
-                    backoff = retry_schedule.initial_backoff;
-                    let _ = status_tx.send(InventoryStatus::Online {
-                        content_hash,
-                        received_at,
-                    });
-                    schedule.cadence.saturating_add(jitter(schedule.jitter))
-                }
-                InventoryDelivery::Retry { retry_after } => {
-                    let delay = retry_after
-                        .unwrap_or(backoff)
-                        .clamp(retry_schedule.initial_backoff, retry_schedule.max_backoff);
-                    backoff = backoff.saturating_mul(2).min(retry_schedule.max_backoff);
-                    let _ = status_tx.send(InventoryStatus::BackingOff { delay });
-                    delay
-                }
-                InventoryDelivery::AuthenticationStopped { status, reason } => {
-                    let _ = status_tx.send(InventoryStatus::AuthenticationStopped { status, reason });
-                    return;
-                }
-                InventoryDelivery::Rejected { status, reason } => {
-                    let suffix = reason.map_or_else(String::new, |reason| format!("; reason={reason}"));
-                    let _ = status_tx.send(InventoryStatus::Failed {
-                        reason: format!("Connect rejected inventory with HTTP {status}{suffix}"),
-                    });
-                    return;
-                }
-            };
-            if sleep_or_cancel(&task_shutdown, delay).await {
-                break;
-            }
-        }
-        let _ = status_tx.send(InventoryStatus::Stopped);
-    });
-    Ok(Some(InventoryRuntime {
-        shutdown,
-        status: status_rx,
-        task: Some(task),
     }))
 }
 
@@ -289,12 +131,6 @@ fn failed(status: &watch::Sender<HeartbeatStatus>, error: HeartbeatError) {
     });
 }
 
-fn failed_inventory(status: &watch::Sender<InventoryStatus>, error: InventoryError) {
-    let _ = status.send(InventoryStatus::Failed {
-        reason: error.to_string(),
-    });
-}
-
 fn jitter(maximum: Duration) -> Duration {
     if maximum.is_zero() {
         Duration::ZERO

rustfs/src/connect/telemetry.rs

@@ -1,265 +0,0 @@
-// Copyright 2024 RustFS Team
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-use std::time::Duration;
-
-use chrono::{DateTime, SecondsFormat, Utc};
-use reqwest::{Client, StatusCode, Url, header};
-use rustls::RootCertStore;
-use rustls::pki_types::{CertificateDer, pem::PemObject as _};
-use serde::{Deserialize, Serialize};
-use zeroize::Zeroizing;
-
-use super::config::HeartbeatConfig;
-use super::credential_store::{CredentialStoreError, DeviceCredential};
-use super::identity::IdentityError;
-use super::identity_store::StoreError;
-use super::registration::{CredentialValidationError, validate_stored_credential};
-
-const MAX_RESPONSE_BYTES: usize = 64 * 1024;
-
-pub(crate) enum TelemetryDelivery {
-    Accepted { cluster_name: String, body: Vec<u8> },
-    Retry { retry_after: Option<Duration> },
-    AuthenticationStopped { status: u16, reason: Option<String> },
-    Rejected { status: u16, reason: Option<String> },
-}
-
-pub(crate) struct TelemetryTransport {
-    endpoint: Url,
-    root_store: RootCertStore,
-    roots: Vec<CertificateDer<'static>>,
-    config: HeartbeatConfig,
-}
-
-impl TelemetryTransport {
-    pub(crate) fn new(config: HeartbeatConfig) -> Result<Self, TelemetryError> {
-        let mut endpoint = Url::parse(&config.endpoint).map_err(|_| TelemetryError::Endpoint)?;
-        if endpoint.scheme() != "https"
-            || endpoint.cannot_be_a_base()
-            || !endpoint.username().is_empty()
-            || endpoint.password().is_some()
-            || endpoint.query().is_some()
-            || endpoint.fragment().is_some()
-        {
-            return Err(TelemetryError::Endpoint);
-        }
-        if !endpoint.path().ends_with('/') {
-            endpoint.set_path(&format!("{}/", endpoint.path()));
-        }
-        let roots = CertificateDer::pem_slice_iter(&config.root_ca_pem)
-            .collect::<Result<Vec<_>, _>>()
-            .map_err(|_| TelemetryError::RootCertificate)?;
-        if roots.is_empty() {
-            return Err(TelemetryError::RootCertificate);
-        }
-        let mut root_store = RootCertStore::empty();
-        let (accepted, rejected) = root_store.add_parsable_certificates(roots.clone());
-        if accepted != roots.len() || rejected != 0 {
-            return Err(TelemetryError::RootCertificate);
-        }
-        let schedule = config.schedule;
-        if schedule.timeout.is_zero()
-            || schedule.timeout > Duration::from_secs(5)
-            || schedule.initial_backoff.is_zero()
-            || schedule.max_backoff < schedule.initial_backoff
-            || schedule.max_backoff > Duration::from_secs(5 * 60)
-        {
-            return Err(TelemetryError::Schedule);
-        }
-        Ok(Self {
-            endpoint,
-            root_store,
-            roots,
-            config,
-        })
-    }
-
-    pub(crate) async fn post<T: Serialize>(&self, collection: &str, value: &T) -> Result<TelemetryDelivery, TelemetryError> {
-        let (cluster_name, cluster_uid, client) = self.authenticated_client().await?;
-        let url = self.endpoint.join(&format!("clusters/{cluster_uid}/{collection}"))?;
-        let response = match client.post(url).json(value).send().await {
-            Ok(response) => response,
-            Err(error) if error.is_timeout() || error.is_connect() || error.is_request() => {
-                return Ok(TelemetryDelivery::Retry { retry_after: None });
-            }
-            Err(error) => return Err(error.into()),
-        };
-        let status = response.status();
-        if status == StatusCode::TOO_MANY_REQUESTS {
-            return Ok(TelemetryDelivery::Retry {
-                retry_after: retry_after(response.headers(), Utc::now(), self.config.schedule.max_backoff),
-            });
-        }
-        if status == StatusCode::REQUEST_TIMEOUT || status.is_server_error() {
-            return Ok(TelemetryDelivery::Retry { retry_after: None });
-        }
-        if matches!(status, StatusCode::UNAUTHORIZED | StatusCode::FORBIDDEN) {
-            return Ok(TelemetryDelivery::AuthenticationStopped {
-                status: status.as_u16(),
-                reason: response_reason(response).await,
-            });
-        }
-        if status != StatusCode::OK {
-            return Ok(TelemetryDelivery::Rejected {
-                status: status.as_u16(),
-                reason: response_reason(response).await,
-            });
-        }
-        Ok(TelemetryDelivery::Accepted {
-            cluster_name,
-            body: bounded_body(response).await?,
-        })
-    }
-
-    async fn authenticated_client(&self) -> Result<(String, String, Client), TelemetryError> {
-        let _lock = self.config.credential_store.lock().await?;
-        let credential = self.config.credential_store.load()?.ok_or(TelemetryError::NotRegistered)?;
-        let identity = self.config.identity_store.load()?.ok_or(TelemetryError::IdentityMissing)?;
-        validate_stored_credential(&credential, &identity, &self.root_store, &self.roots)?;
-        let now = Utc::now().timestamp();
-        if now < credential.not_before_unix || now >= credential.not_after_unix {
-            return Err(TelemetryError::CredentialExpired);
-        }
-        let (organization_uid, cluster_uid) = credential_parent(&credential)?;
-        let cluster_name = format!("organizations/{organization_uid}/clusters/{cluster_uid}");
-        let client = self.client(&credential, &identity.to_pkcs8_pem()?)?;
-        Ok((cluster_name, cluster_uid.to_owned(), client))
-    }
-
-    fn client(&self, credential: &DeviceCredential, key: &Zeroizing<String>) -> Result<Client, TelemetryError> {
-        let mut pem = Zeroizing::new(Vec::with_capacity(credential.certificate_chain.len() + key.len() + 1));
-        pem.extend_from_slice(credential.certificate_chain.as_bytes());
-        pem.push(b'\n');
-        pem.extend_from_slice(key.as_bytes());
-        let identity = reqwest::Identity::from_pem(&pem).map_err(|_| TelemetryError::IdentityCertificate)?;
-        let roots = self
-            .roots
-            .iter()
-            .map(|root| reqwest::Certificate::from_der(root.as_ref()))
-            .collect::<Result<Vec<_>, _>>()?;
-        Client::builder()
-            .https_only(true)
-            .redirect(reqwest::redirect::Policy::none())
-            .timeout(self.config.schedule.timeout)
-            .tls_certs_only(roots)
-            .identity(identity)
-            .build()
-            .map_err(Into::into)
-    }
-}
-
-fn credential_parent(credential: &DeviceCredential) -> Result<(&str, &str), TelemetryError> {
-    let mut parts = credential.name.split('/');
-    let valid = parts.next() == Some("organizations");
-    let organization_uid = parts.next();
-    let valid = valid && parts.next() == Some("clusters");
-    let cluster_uid = parts.next();
-    let valid = valid && parts.next() == Some("clusterDevices");
-    let device_uid = parts.next();
-    if !valid
-        || organization_uid.is_none_or(str::is_empty)
-        || cluster_uid.is_none_or(str::is_empty)
-        || device_uid != Some(credential.uid.as_str())
-        || parts.next().is_some()
-    {
-        return Err(TelemetryError::CredentialName);
-    }
-    Ok((
-        organization_uid.ok_or(TelemetryError::CredentialName)?,
-        cluster_uid.ok_or(TelemetryError::CredentialName)?,
-    ))
-}
-
-fn retry_after(headers: &header::HeaderMap, now: DateTime<Utc>, maximum: Duration) -> Option<Duration> {
-    let value = headers.get(header::RETRY_AFTER)?.to_str().ok()?;
-    let delay = value.parse::<u64>().ok().map(Duration::from_secs).or_else(|| {
-        DateTime::parse_from_rfc2822(value)
-            .ok()
-            .and_then(|at| (at.with_timezone(&Utc) - now).to_std().ok())
-    })?;
-    Some(delay.min(maximum))
-}
-
-pub(crate) fn is_exact_utc_seconds(value: &str) -> bool {
-    DateTime::parse_from_rfc3339(value).is_ok_and(|time| {
-        time.offset().local_minus_utc() == 0
-            && value.ends_with('Z')
-            && time.with_timezone(&Utc).to_rfc3339_opts(SecondsFormat::Secs, true) == value
-    })
-}
-
-async fn response_reason(response: reqwest::Response) -> Option<String> {
-    #[derive(Deserialize)]
-    struct Envelope {
-        #[serde(default)]
-        details: Vec<Detail>,
-    }
-    #[derive(Deserialize)]
-    struct Detail {
-        #[serde(default)]
-        reason: String,
-    }
-
-    serde_json::from_slice::<Envelope>(&bounded_body(response).await.ok()?)
-        .ok()?
-        .details
-        .into_iter()
-        .find_map(|detail| (!detail.reason.is_empty()).then_some(detail.reason))
-}
-
-async fn bounded_body(mut response: reqwest::Response) -> Result<Vec<u8>, TelemetryError> {
-    let mut body = Vec::new();
-    while let Some(chunk) = response.chunk().await? {
-        if body.len().saturating_add(chunk.len()) > MAX_RESPONSE_BYTES {
-            return Err(TelemetryError::ResponseTooLarge);
-        }
-        body.extend_from_slice(&chunk);
-    }
-    Ok(body)
-}
-
-#[derive(Debug, thiserror::Error)]
-pub(crate) enum TelemetryError {
-    #[error("Connect telemetry endpoint must be an HTTPS base URL without credentials, query, or fragment")]
-    Endpoint,
-    #[error("Connect telemetry root CA configuration is invalid")]
-    RootCertificate,
-    #[error("Connect telemetry retry schedule is invalid")]
-    Schedule,
-    #[error("RustFS is not registered with Connect")]
-    NotRegistered,
-    #[error("the Connect device private key is missing")]
-    IdentityMissing,
-    #[error("the stored Connect certificate and device private key cannot form a TLS identity")]
-    IdentityCertificate,
-    #[error("the stored Connect credential name is invalid")]
-    CredentialName,
-    #[error("the stored Connect device certificate is not currently valid")]
-    CredentialExpired,
-    #[error("Connect telemetry response exceeded 64 KiB")]
-    ResponseTooLarge,
-    #[error(transparent)]
-    Url(#[from] url::ParseError),
-    #[error(transparent)]
-    Transport(#[from] reqwest::Error),
-    #[error(transparent)]
-    Identity(#[from] IdentityError),
-    #[error(transparent)]
-    IdentityStore(#[from] StoreError),
-    #[error(transparent)]
-    CredentialStore(#[from] CredentialStoreError),
-    #[error(transparent)]
-    CredentialValidation(#[from] CredentialValidationError),
-}

rustfs/src/main.rs

@@ -26,22 +26,22 @@ struct MiMallocAllocator;
 unsafe impl GlobalAlloc for MiMallocAllocator {
     unsafe fn alloc(&self, layout: Layout) -> *mut u8 {
         // SAFETY: the caller upholds GlobalAlloc's contract for layout.
-        unsafe { mimalloc::MiMalloc.alloc(layout) }
+        unsafe { rustfs_mimalloc::MiMalloc.alloc(layout) }
     }
 
     unsafe fn alloc_zeroed(&self, layout: Layout) -> *mut u8 {
         // SAFETY: the caller upholds GlobalAlloc's contract for layout.
-        unsafe { mimalloc::MiMalloc.alloc_zeroed(layout) }
+        unsafe { rustfs_mimalloc::MiMalloc.alloc_zeroed(layout) }
     }
 
     unsafe fn dealloc(&self, ptr: *mut u8, layout: Layout) {
         // SAFETY: ptr and layout came from this allocator and are forwarded unchanged.
-        unsafe { mimalloc::MiMalloc.dealloc(ptr, layout) }
+        unsafe { rustfs_mimalloc::MiMalloc.dealloc(ptr, layout) }
     }
 
     unsafe fn realloc(&self, ptr: *mut u8, layout: Layout, new_size: usize) -> *mut u8 {
         // SAFETY: ptr and layout came from this allocator and are forwarded unchanged.
-        unsafe { mimalloc::MiMalloc.realloc(ptr, layout, new_size) }
+        unsafe { rustfs_mimalloc::MiMalloc.realloc(ptr, layout, new_size) }
     }
 }
 
@@ -51,7 +51,7 @@ static GLOBAL: hotpath::CountingAllocator<MiMallocAllocator> = hotpath::Counting
 
 #[cfg(not(all(feature = "hotpath", feature = "hotpath-alloc")))]
 #[global_allocator]
-static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+static GLOBAL: rustfs_mimalloc::MiMalloc = rustfs_mimalloc::MiMalloc;
 
 fn main() {
     let _hotpath_guard = hotpath::HotpathGuardBuilder::new("main").build();
@@ -71,8 +71,9 @@ mod tests {
         allocation.extend_from_slice(&[7_u8; 64]);
 
         assert_eq!(allocation.len(), 64);
+        let heap = rustfs_mimalloc::heap::Heap::main();
         // SAFETY: the live Vec pointer is valid to inspect for heap ownership.
-        assert!(unsafe { libmimalloc_sys::mi_is_in_heap_region(allocation.as_ptr().cast()) });
+        assert!(unsafe { heap.contains(allocation.as_ptr()) });
     }
 
     #[test]
@@ -85,12 +86,13 @@ mod tests {
         let layout = Layout::from_size_align(32, 8).expect("valid test allocation layout");
         let grown_layout = Layout::from_size_align(64, 8).expect("valid grown test allocation layout");
         let allocator = super::MiMallocAllocator;
+        let heap = rustfs_mimalloc::heap::Heap::main();
 
         // SAFETY: The pointer is checked for null before use and later released
         // through the same allocator with the corresponding layout.
         let ptr = unsafe { allocator.alloc_zeroed(layout) };
         assert!(!ptr.is_null());
-        assert!(unsafe { libmimalloc_sys::mi_is_in_heap_region(ptr.cast()) });
+        assert!(unsafe { heap.contains(ptr) });
         assert!(unsafe { std::slice::from_raw_parts(ptr, 32).iter().all(|byte| *byte == 0) });
 
         // SAFETY: `ptr` was allocated by `allocator` with `layout`; on failure
@@ -102,7 +104,7 @@ mod tests {
             panic!("mimalloc realloc failed in allocator smoke test");
         }
 
-        assert!(unsafe { libmimalloc_sys::mi_is_in_heap_region(grown_ptr.cast()) });
+        assert!(unsafe { heap.contains(grown_ptr) });
         // SAFETY: `grown_ptr` was reallocated by `allocator` and is released
         // with the matching grown layout.
         unsafe { allocator.dealloc(grown_ptr, grown_layout) };

rustfs/src/memory_observability.rs

@@ -17,10 +17,7 @@ use rustfs_io_metrics::{
     record_cpu_usage, record_memory_usage, record_process_memory_split,
 };
 use serde::Serialize;
-#[cfg(any(test, not(target_os = "windows")))]
 use serde_json::Value;
-#[cfg(not(target_os = "windows"))]
-use std::ffi::CStr;
 use std::path::Path;
 use std::sync::{Arc, Mutex, OnceLock};
 use std::time::Duration;
@@ -231,7 +228,18 @@ fn read_cgroup_memory_snapshot() -> Option<CgroupMemorySnapshot> {
     read_cgroup_v2().or_else(read_cgroup_v1)
 }
 
-#[cfg(any(test, not(target_os = "windows")))]
+fn read_allocator_memory_snapshot() -> Option<AllocatorMemorySnapshot> {
+    let json = rustfs_mimalloc::MiMalloc::stats_json();
+    if json.is_empty() {
+        return None;
+    }
+    let observation = parse_mimalloc_stats_json(&json)?;
+    Some(AllocatorMemorySnapshot {
+        backend: crate::allocator_reclaim::allocator_backend(),
+        observation,
+    })
+}
+
 fn numeric_json_value(value: &Value) -> Option<u64> {
     match value {
         Value::Number(number) => number
@@ -242,7 +250,6 @@ fn numeric_json_value(value: &Value) -> Option<u64> {
     }
 }
 
-#[cfg(any(test, not(target_os = "windows")))]
 fn numeric_json_field(value: &Value, field: &str) -> Option<u64> {
     match value {
         Value::Object(fields) => fields
@@ -254,7 +261,6 @@ fn numeric_json_field(value: &Value, field: &str) -> Option<u64> {
     }
 }
 
-#[cfg(any(test, not(target_os = "windows")))]
 fn mimalloc_stat_field(value: &Value, metric: &str, field: &str) -> Option<u64> {
     match value {
         Value::Object(fields) => {
@@ -271,12 +277,10 @@ fn mimalloc_stat_field(value: &Value, metric: &str, field: &str) -> Option<u64>
     }
 }
 
-#[cfg(any(test, not(target_os = "windows")))]
 fn mimalloc_stat_current(value: &Value, metric: &str) -> Option<u64> {
     mimalloc_stat_field(value, metric, "current")
 }
 
-#[cfg(any(test, not(target_os = "windows")))]
 fn mimalloc_stat_sum(value: &Value, metrics: &[&str], field: &str) -> Option<u64> {
     metrics
         .iter()
@@ -285,7 +289,6 @@ fn mimalloc_stat_sum(value: &Value, metrics: &[&str], field: &str) -> Option<u64
         .filter(|value| *value > 0)
 }
 
-#[cfg(any(test, not(target_os = "windows")))]
 fn parse_mimalloc_stats_json(stats_json: &str) -> Option<AllocatorMemoryObservation> {
     let value = serde_json::from_str::<Value>(stats_json).ok()?;
     let malloc_metrics = ["malloc_normal", "malloc_huge"];
@@ -312,33 +315,6 @@ fn parse_mimalloc_stats_json(stats_json: &str) -> Option<AllocatorMemoryObservat
     }
 }
 
-#[cfg(not(target_os = "windows"))]
-#[allow(unsafe_code)]
-fn read_allocator_memory_snapshot() -> Option<AllocatorMemorySnapshot> {
-    // SAFETY: `mi_stats_get_json` returns a null-terminated JSON buffer owned by
-    // mimalloc when called with a null input buffer. The mimalloc API requires
-    // freeing that buffer with `mi_free`; parsing finishes before the buffer is freed.
-    let observation = unsafe {
-        let stats_ptr = libmimalloc_sys::mi_stats_get_json(0, std::ptr::null_mut());
-        if stats_ptr.is_null() {
-            return None;
-        }
-
-        let observation = CStr::from_ptr(stats_ptr).to_str().ok().and_then(parse_mimalloc_stats_json);
-        libmimalloc_sys::mi_free(stats_ptr.cast());
-        observation?
-    };
-    Some(AllocatorMemorySnapshot {
-        backend: crate::allocator_reclaim::allocator_backend(),
-        observation,
-    })
-}
-
-#[cfg(target_os = "windows")]
-fn read_allocator_memory_snapshot() -> Option<AllocatorMemorySnapshot> {
-    None
-}
-
 fn configured_memory_observability_interval_secs() -> u64 {
     rustfs_utils::get_env_u64(ENV_MEMORY_OBSERVABILITY_INTERVAL_SECS, DEFAULT_MEMORY_OBSERVABILITY_INTERVAL_SECS).max(1)
 }
@@ -566,6 +542,13 @@ mod tests {
         assert_eq!(parse_mimalloc_stats_json(r#"{ "allocator": "unknown" }"#), None);
     }
 
+    #[test]
+    fn read_allocator_memory_snapshot_uses_mimalloc_stats_json() {
+        let snapshot = super::read_allocator_memory_snapshot();
+        #[cfg(not(target_os = "windows"))]
+        assert!(snapshot.is_some(), "allocator snapshot should be available on non-Windows");
+    }
+
     #[test]
     fn memory_observability_snapshot_reports_disabled_when_metrics_are_disabled() {
         let snapshot = build_memory_observability_status_snapshot(false, 15, false);

rustfs/src/startup_services.rs

@@ -13,13 +13,10 @@
 // limitations under the License.
 
 use crate::site_replication_reconcile::spawn_site_replication_reconcile_task;
-use crate::storage_api::startup::services::{ECStore, EndpointServerPools, ServerContextSlot, StorageAdminApi};
+use crate::storage_api::startup::services::{ECStore, EndpointServerPools, ServerContextSlot};
 use crate::{
     config::Config,
-    connect::{
-        CoarseNodeSummary, HeartbeatConfig, HeartbeatRuntime, InventoryError, InventoryFlag, InventoryRuntime, InventorySchedule,
-        InventorySnapshot, spawn_heartbeat_runtime, spawn_inventory_runtime,
-    },
+    connect::{CoarseNodeSummary, HeartbeatConfig, HeartbeatRuntime, spawn_heartbeat_runtime},
     init::{init_buffer_profile_system, init_kms_system},
     server::ServiceStateManager,
     startup_audit::init_audit_runtime,
@@ -80,9 +77,6 @@ pub(crate) async fn init_startup_runtime_services(
     let optional_runtimes = init_optional_runtime_services().await?;
     let heartbeat_config = HeartbeatConfig::from_env().map_err(std::io::Error::other)?;
     let heartbeat_nodes = heartbeat_config.as_ref().map(|_| endpoint_pools.get_nodes().len());
-    let inventory_drives = heartbeat_config
-        .as_ref()
-        .map(|_| endpoint_pools.as_ref().iter().map(|pool| pool.endpoints.as_ref().len()).sum());
 
     init_buffer_profile_system(config);
     init_deadlock_detector_runtime();
@@ -102,9 +96,7 @@ pub(crate) async fn init_startup_runtime_services(
     init_notification_runtime(endpoint_pools, buckets).await?;
     let enable_scanner = init_background_service_runtime(store.clone()).await?;
     init_observability_runtime(store.clone(), ctx.clone()).await;
-    let heartbeat = start_heartbeat_runtime(heartbeat_config.clone(), heartbeat_nodes, &ctx)?;
-    let inventory = start_inventory_runtime(heartbeat_config, heartbeat_nodes, inventory_drives, store, &ctx)?;
-    let heartbeat = heartbeat.map(|heartbeat| heartbeat.with_inventory(inventory));
+    let heartbeat = start_heartbeat_runtime(heartbeat_config, heartbeat_nodes, &ctx)?;
 
     Ok(StartupServiceRuntime {
         optional_runtimes,
@@ -128,69 +120,3 @@ fn start_heartbeat_runtime(
         .ok_or_else(|| std::io::Error::other("Connect heartbeat node count is outside protocol bounds"))?;
     spawn_heartbeat_runtime(Some(config), shutdown, move || summary).map_err(std::io::Error::other)
 }
-
-fn start_inventory_runtime(
-    config: Option<HeartbeatConfig>,
-    node_count: Option<usize>,
-    expected_drive_count: Option<usize>,
-    store: Arc<ECStore>,
-    shutdown: &CancellationToken,
-) -> Result<Option<InventoryRuntime>> {
-    let Some(config) = config else {
-        return Ok(None);
-    };
-    let node_count = node_count.unwrap_or_default();
-    let expected_drive_count = expected_drive_count.unwrap_or_default();
-    spawn_inventory_runtime(Some(config), InventorySchedule::default(), shutdown, move || {
-        let store = store.clone();
-        async move {
-            let info = StorageAdminApi::storage_info(store.as_ref()).await;
-            inventory_snapshot(node_count, expected_drive_count, info)
-        }
-    })
-    .map_err(std::io::Error::other)
-}
-
-fn inventory_snapshot(
-    node_count: usize,
-    expected_drive_count: usize,
-    info: rustfs_madmin::StorageInfo,
-) -> std::result::Result<InventorySnapshot, InventoryError> {
-    if info.disks.len() != expected_drive_count {
-        return Err(InventoryError::SnapshotIncomplete {
-            expected: expected_drive_count,
-            observed: info.disks.len(),
-        });
-    }
-    let total = crate::app::storage_api::capacity::get_total_usable_capacity(&info.disks, &info) as u64;
-    let free = crate::app::storage_api::capacity::get_total_usable_capacity_free(&info.disks, &info) as u64;
-    let mut flags = Vec::with_capacity(3);
-    if info.disks.iter().any(|disk| disk.state == rustfs_madmin::ITEM_OFFLINE) {
-        flags.extend([InventoryFlag::ClusterDegraded, InventoryFlag::DriveOffline]);
-    }
-    if info.disks.iter().any(|disk| disk.healing) {
-        flags.push(InventoryFlag::ClusterHealing);
-    }
-    InventorySnapshot::current(node_count, info.disks.len(), total, free, flags)
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn inventory_rejects_a_partial_startup_storage_snapshot() {
-        let info = rustfs_madmin::StorageInfo {
-            disks: vec![rustfs_madmin::Disk::default()],
-            ..Default::default()
-        };
-
-        assert!(matches!(
-            inventory_snapshot(2, 2, info),
-            Err(InventoryError::SnapshotIncomplete {
-                expected: 2,
-                observed: 1
-            })
-        ));
-    }
-}

rustfs/src/storage_api.rs

@@ -279,7 +279,6 @@ pub(crate) mod startup {
     }
 
     pub(crate) mod services {
-        pub(crate) use super::super::storage_contracts::StorageAdminApi;
         pub(crate) use crate::storage::storage_api::{ECStore, EndpointServerPools, ServerContextSlot};
     }
 

rustfs/tests/connect_inventory.rs

@@ -1,669 +0,0 @@
-// Copyright 2024 RustFS Team
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-use std::collections::VecDeque;
-use std::fs;
-use std::sync::atomic::{AtomicUsize, Ordering};
-use std::sync::{Arc, Mutex};
-use std::time::Duration;
-
-use bytes::Bytes;
-use http_body_util::{BodyExt as _, Full};
-use hyper::service::service_fn;
-use hyper::{Request, Response, StatusCode};
-use hyper_util::rt::TokioIo;
-use rcgen::{
-    BasicConstraints, CertificateParams, DistinguishedName, DnType, ExtendedKeyUsagePurpose, IsCa, Issuer, KeyPair,
-    KeyUsagePurpose, SanType,
-};
-use rustfs::connect::{
-    CredentialStore, DeviceCredential, HeartbeatConfig, HeartbeatSchedule, IdentityStore, InventoryFlag, InventoryOsVersion,
-    InventorySchedule, InventorySnapshot, InventoryStatus, OperatingSystemFamily, spawn_inventory_runtime,
-};
-use rustls::RootCertStore;
-use rustls::pki_types::{CertificateDer, PrivateKeyDer, PrivatePkcs8KeyDer};
-use rustls::server::WebPkiClientVerifier;
-use serde_json::{Value, json};
-use time::OffsetDateTime;
-use tokio::net::TcpListener;
-use tokio::sync::watch;
-use tokio_rustls::TlsAcceptor;
-use tokio_util::sync::CancellationToken;
-
-const ORGANIZATION_UID: &str = "0198f4b0-1a00-7c10-8d21-2e3f4a5b6c70";
-const CLUSTER_UID: &str = "0198f4b0-2b00-7d20-9e31-3f4a5b6c7d81";
-const DEVICE_UID: &str = "0198f4b0-3c00-7e30-8f41-4a5b6c7d8e92";
-const SNAPSHOT_UID: &str = "0198f4b0-4d00-7f40-9051-5b6c7d8e9fa3";
-
-struct TestPki {
-    root_params: CertificateParams,
-    root_key: KeyPair,
-    root_der: CertificateDer<'static>,
-    root_pem: String,
-    server_der: CertificateDer<'static>,
-    server_key: PrivatePkcs8KeyDer<'static>,
-}
-
-impl TestPki {
-    fn new() -> Self {
-        let now = OffsetDateTime::now_utc();
-        let root_key = KeyPair::generate().expect("generate root key");
-        let mut root_params = CertificateParams::default();
-        root_params.is_ca = IsCa::Ca(BasicConstraints::Unconstrained);
-        root_params.not_before = now - time::Duration::days(30);
-        root_params.not_after = now + time::Duration::days(30);
-        root_params.key_usages = vec![KeyUsagePurpose::KeyCertSign, KeyUsagePurpose::DigitalSignature];
-        let root = root_params.self_signed(&root_key).expect("sign root");
-
-        let server_key = KeyPair::generate().expect("generate server key");
-        let mut server_params = CertificateParams::default();
-        server_params.not_before = now - time::Duration::hours(1);
-        server_params.not_after = now + time::Duration::days(2);
-        server_params
-            .subject_alt_names
-            .push(SanType::DnsName("localhost".try_into().expect("valid DNS name")));
-        server_params.extended_key_usages = vec![ExtendedKeyUsagePurpose::ServerAuth];
-        let server = server_params
-            .signed_by(&server_key, &Issuer::from_params(&root_params, &root_key))
-            .expect("sign server certificate");
-        Self {
-            root_params,
-            root_key,
-            root_der: root.der().clone(),
-            root_pem: root.pem(),
-            server_der: server.der().clone(),
-            server_key: PrivatePkcs8KeyDer::from(server_key.serialize_der()),
-        }
-    }
-
-    fn server_config(&self) -> rustls::ServerConfig {
-        let mut roots = RootCertStore::empty();
-        roots.add(self.root_der.clone()).expect("add client root");
-        let verifier = WebPkiClientVerifier::builder(Arc::new(roots))
-            .build()
-            .expect("client verifier");
-        rustls::ServerConfig::builder()
-            .with_client_cert_verifier(verifier)
-            .with_single_cert(vec![self.server_der.clone()], PrivateKeyDer::Pkcs8(self.server_key.clone_key()))
-            .expect("server TLS")
-    }
-
-    fn stores(&self, temp: &tempfile::TempDir) -> (IdentityStore, CredentialStore) {
-        let identity_store = IdentityStore::new(temp.path().join("identity"));
-        let identity = identity_store.load_or_create().expect("create identity");
-        let private_key = PrivatePkcs8KeyDer::from(identity.to_pkcs8_der().expect("serialize key").to_vec());
-        let device_key = KeyPair::from_pkcs8_der_and_sign_algo(&private_key, &rcgen::PKCS_ECDSA_P256_SHA256).expect("device key");
-        let now = OffsetDateTime::now_utc();
-        let mut params = CertificateParams::default();
-        params.not_before = now - time::Duration::hours(1);
-        params.not_after = now + time::Duration::hours(23);
-        params.serial_number = Some(vec![1; 16].into());
-        params.key_usages = vec![KeyUsagePurpose::DigitalSignature];
-        params.extended_key_usages = vec![ExtendedKeyUsagePurpose::ClientAuth];
-        params.distinguished_name = DistinguishedName::new();
-        params.distinguished_name.push(DnType::CommonName, DEVICE_UID);
-        params.subject_alt_names.push(SanType::URI(
-            format!("urn:rustfs:connect:device:{DEVICE_UID}")
-                .try_into()
-                .expect("device URI"),
-        ));
-        let certificate = params
-            .signed_by(&device_key, &Issuer::from_params(&self.root_params, &self.root_key))
-            .expect("device certificate");
-        let cluster = format!("organizations/{ORGANIZATION_UID}/clusters/{CLUSTER_UID}");
-        let credential = DeviceCredential {
-            name: format!("{cluster}/clusterDevices/{DEVICE_UID}"),
-            uid: DEVICE_UID.to_owned(),
-            protocol_version: "v1".to_owned(),
-            key_id: format!("x509-{}", "01".repeat(16)),
-            certificate_serial: "01".repeat(16),
-            certificate: certificate.pem(),
-            certificate_chain: certificate.pem(),
-            not_before_unix: (now - time::Duration::hours(1)).unix_timestamp(),
-            not_after_unix: (now + time::Duration::hours(23)).unix_timestamp(),
-        };
-        let directory = temp.path().join("credential");
-        fs::create_dir_all(&directory).expect("credential directory");
-        let path = directory.join("device.crt.json");
-        fs::write(&path, serde_json::to_vec(&credential).expect("credential JSON")).expect("write credential");
-        private_mode(&path);
-        (identity_store, CredentialStore::new(directory))
-    }
-}
-
-#[derive(Clone)]
-struct Reply {
-    status: StatusCode,
-    body: Value,
-    retry_after: Option<&'static str>,
-}
-
-impl Reply {
-    fn ok(content_hash: &str) -> Self {
-        Self {
-            status: StatusCode::OK,
-            body: json!({
-                "name": format!("organizations/{ORGANIZATION_UID}/clusters/{CLUSTER_UID}/inventorySnapshots/{SNAPSHOT_UID}"),
-                "uid": SNAPSHOT_UID,
-                "contentHash": content_hash,
-                "receivedAt": "2026-08-22T01:02:03Z",
-                "futureField": true
-            }),
-            retry_after: None,
-        }
-    }
-
-    fn error(status: StatusCode, reason: &str) -> Self {
-        Self {
-            status,
-            body: json!({"details": [{"reason": reason}]}),
-            retry_after: None,
-        }
-    }
-}
-
-struct TestServer {
-    endpoint: String,
-    seen: Arc<Mutex<Vec<Value>>>,
-    task: tokio::task::JoinHandle<()>,
-}
-
-impl Drop for TestServer {
-    fn drop(&mut self) {
-        self.task.abort();
-    }
-}
-
-async fn server(pki: &TestPki, replies: Vec<Reply>) -> TestServer {
-    let listener = TcpListener::bind("127.0.0.1:0").await.expect("bind server");
-    let address = listener.local_addr().expect("server address");
-    let acceptor = TlsAcceptor::from(Arc::new(pki.server_config()));
-    let replies = Arc::new(Mutex::new(VecDeque::from(replies)));
-    let seen = Arc::new(Mutex::new(Vec::new()));
-    let captured = seen.clone();
-    let task = tokio::spawn(async move {
-        while let Ok((stream, _)) = listener.accept().await {
-            let acceptor = acceptor.clone();
-            let replies = replies.clone();
-            let seen = captured.clone();
-            tokio::spawn(async move {
-                let Ok(stream) = acceptor.accept(stream).await else { return };
-                let service = service_fn(move |request: Request<hyper::body::Incoming>| {
-                    let replies = replies.clone();
-                    let seen = seen.clone();
-                    async move {
-                        assert_eq!(request.uri().path(), format!("/agent/clusters/{CLUSTER_UID}/inventorySnapshots"));
-                        let body = request.into_body().collect().await.expect("request body").to_bytes();
-                        seen.lock()
-                            .expect("seen lock")
-                            .push(serde_json::from_slice(&body).expect("request JSON"));
-                        let reply = replies
-                            .lock()
-                            .expect("reply lock")
-                            .pop_front()
-                            .unwrap_or_else(|| Reply::error(StatusCode::SERVICE_UNAVAILABLE, "UNAVAILABLE"));
-                        let mut builder = Response::builder()
-                            .status(reply.status)
-                            .header("content-type", "application/json");
-                        if let Some(value) = reply.retry_after {
-                            builder = builder.header("retry-after", value);
-                        }
-                        Ok::<_, hyper::Error>(
-                            builder
-                                .body(Full::new(Bytes::from(serde_json::to_vec(&reply.body).expect("reply JSON"))))
-                                .expect("reply"),
-                        )
-                    }
-                });
-                let _ = hyper::server::conn::http1::Builder::new()
-                    .serve_connection(TokioIo::new(stream), service)
-                    .await;
-            });
-        }
-    });
-    TestServer {
-        endpoint: format!("https://localhost:{}/agent/", address.port()),
-        seen,
-        task,
-    }
-}
-
-fn config(temp: &tempfile::TempDir, pki: &TestPki, server: &TestServer) -> HeartbeatConfig {
-    let (identity_store, credential_store) = pki.stores(temp);
-    HeartbeatConfig {
-        endpoint: server.endpoint.clone(),
-        root_ca_pem: pki.root_pem.as_bytes().to_vec(),
-        identity_store,
-        credential_store,
-        state_path: temp.path().join("private-config-secret/heartbeat/state.json"),
-        schedule: HeartbeatSchedule {
-            cadence: Duration::from_secs(30),
-            jitter: Duration::ZERO,
-            timeout: Duration::from_millis(200),
-            initial_backoff: Duration::from_millis(20),
-            max_backoff: Duration::from_millis(80),
-        },
-    }
-}
-
-fn schedule() -> InventorySchedule {
-    InventorySchedule {
-        cadence: Duration::from_secs(60),
-        jitter: Duration::ZERO,
-    }
-}
-
-fn snapshot() -> InventorySnapshot {
-    InventorySnapshot::new(
-        "1.4.2",
-        Some(InventoryOsVersion::new(OperatingSystemFamily::Linux, 6, 8).expect("valid operating-system version")),
-        8,
-        96,
-        1_099_511_627_776,
-        412_316_860_416,
-        [InventoryFlag::ClusterDegraded, InventoryFlag::DriveOffline],
-    )
-    .expect("valid inventory")
-}
-
-fn collect_strings(value: &Value, strings: &mut Vec<String>) {
-    match value {
-        Value::String(value) => strings.push(value.clone()),
-        Value::Array(values) => values.iter().for_each(|value| collect_strings(value, strings)),
-        Value::Object(values) => values.values().for_each(|value| collect_strings(value, strings)),
-        _ => {}
-    }
-}
-
-async fn wait_for(
-    status: &mut watch::Receiver<InventoryStatus>,
-    predicate: impl Fn(&InventoryStatus) -> bool,
-) -> InventoryStatus {
-    tokio::time::timeout(Duration::from_secs(3), async {
-        loop {
-            let current = status.borrow_and_update().clone();
-            if predicate(&current) {
-                return current;
-            }
-            status.changed().await.expect("status channel");
-        }
-    })
-    .await
-    .expect("inventory status timeout")
-}
-
-#[test]
-fn connect_inventory_frozen_vector_has_the_exact_canonical_hash_and_no_open_ended_fields() {
-    let fixtures: Value = serde_json::from_str(include_str!("../../protocol/agent/v1/fixtures/inventory/valid-vectors.json"))
-        .expect("valid fixture JSON");
-    let expected = &fixtures["vectors"][0]["expected"];
-    let snapshot = snapshot();
-
-    assert_eq!(snapshot.content_hash().expect("content hash"), expected["contentHash"]);
-    assert_eq!(InventorySchedule::default().cadence, Duration::from_secs(6 * 60 * 60));
-    assert_eq!(InventorySchedule::default().jitter, Duration::from_secs(30 * 60));
-    let encoded = serde_json::to_value(snapshot).expect("snapshot JSON");
-    assert_eq!(
-        encoded,
-        json!({
-            "rustfsVersion": "1.4.2",
-            "osVersion": {"family": "linux", "major": 6, "minor": 8},
-            "nodeCount": 8,
-            "driveCount": 96,
-            "capacityTotalBytes": 1099511627776_u64,
-            "capacityUsedBytes": 412316860416_u64,
-            "coarseFlags": ["cluster.degraded", "drive.offline"]
-        })
-    );
-
-    let fixtures: Value =
-        serde_json::from_str(include_str!("../../protocol/agent/v1/fixtures/inventory/secret-like-vectors.json"))
-            .expect("valid secret-like fixture JSON");
-    let known_fields = [
-        "protocolVersion",
-        "rustfsVersion",
-        "osVersion",
-        "nodeCount",
-        "driveCount",
-        "capacityTotalBytes",
-        "capacityUsedBytes",
-        "coarseFlags",
-    ];
-    let known_flags = ["cluster.degraded", "drive.offline"];
-    let mut excluded = Vec::new();
-    for vector in fixtures["vectors"].as_array().expect("fixture vectors") {
-        let input = vector["input"].as_object().expect("fixture input");
-        for (name, value) in input {
-            if !known_fields.contains(&name.as_str()) {
-                collect_strings(value, &mut excluded);
-            }
-        }
-        for (name, value) in input["osVersion"].as_object().expect("fixture OS version") {
-            if !["family", "major", "minor"].contains(&name.as_str()) {
-                collect_strings(value, &mut excluded);
-            }
-        }
-        for flag in input["coarseFlags"].as_array().expect("fixture coarse flags") {
-            let flag = flag.as_str().expect("fixture coarse flag");
-            if !known_flags.contains(&flag) {
-                excluded.push(flag.to_owned());
-            }
-        }
-    }
-    let encoded = serde_json::to_string(&encoded).expect("encoded snapshot");
-    for value in excluded {
-        assert!(!encoded.contains(&value), "snapshot exposed fixture value {value}");
-    }
-}
-
-#[test]
-fn connect_inventory_bounds_fail_instead_of_truncating_or_inventing_values() {
-    assert!(matches!(
-        InventorySnapshot::current(0, 0, 0, 0, []),
-        Err(rustfs::connect::InventoryError::NodeCount)
-    ));
-    assert!(matches!(
-        InventorySnapshot::current(1, 1_048_577, 0, 0, []),
-        Err(rustfs::connect::InventoryError::DriveCount)
-    ));
-    assert!(matches!(
-        InventorySnapshot::current(1, 0, 9_007_199_254_740_992, 0, []),
-        Err(rustfs::connect::InventoryError::Capacity)
-    ));
-    assert!(matches!(
-        InventorySnapshot::current(1, 0, 10, 11, []),
-        Err(rustfs::connect::InventoryError::Capacity)
-    ));
-    assert!(matches!(
-        InventorySnapshot::new("1.0.0-private.1", None, 1, 0, 0, 0, []),
-        Err(rustfs::connect::InventoryError::RustfsVersion)
-    ));
-}
-
-#[tokio::test]
-async fn connect_inventory_restart_replays_the_pending_request_and_then_skips_unchanged_inventory() {
-    let pki = TestPki::new();
-    let content_hash = snapshot().content_hash().expect("content hash");
-    let first_server = server(&pki, vec![Reply::error(StatusCode::SERVICE_UNAVAILABLE, "UNAVAILABLE")]).await;
-    let temp = tempfile::tempdir().expect("tempdir");
-    let shutdown = CancellationToken::new();
-    let samples = Arc::new(AtomicUsize::new(0));
-    let sampled = samples.clone();
-    let runtime = spawn_inventory_runtime(Some(config(&temp, &pki, &first_server)), schedule(), &shutdown, move || {
-        sampled.fetch_add(1, Ordering::Relaxed);
-        std::future::ready(Ok(snapshot()))
-    })
-    .expect("start inventory")
-    .expect("configured inventory");
-    let mut status = runtime.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::BackingOff { .. })).await,
-        InventoryStatus::BackingOff { delay } if delay == Duration::from_millis(20)
-    ));
-    assert_eq!(samples.load(Ordering::Relaxed), 1);
-    let original = first_server.seen.lock().expect("seen lock")[0].clone();
-    runtime.shutdown().await;
-
-    let mut limited = Reply::error(StatusCode::TOO_MANY_REQUESTS, "RATE_LIMITED");
-    limited.retry_after = Some("0");
-    let restart_server = server(&pki, vec![limited, Reply::ok(&content_hash)]).await;
-    let restart_config = config(&temp, &pki, &restart_server);
-    let restart_samples = Arc::new(AtomicUsize::new(0));
-    let sampled = restart_samples.clone();
-    let restart = spawn_inventory_runtime(Some(restart_config.clone()), schedule(), &shutdown, move || {
-        sampled.fetch_add(1, Ordering::Relaxed);
-        std::future::ready(Ok(snapshot()))
-    })
-    .expect("restart inventory")
-    .expect("configured inventory");
-    let mut status = restart.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::Online { .. })).await,
-        InventoryStatus::Online { content_hash: accepted, received_at }
-            if accepted == content_hash && received_at == "2026-08-22T01:02:03Z"
-    ));
-    assert_eq!(restart_samples.load(Ordering::Relaxed), 0);
-    let delivered = restart_server.seen.lock().expect("seen lock").clone();
-    assert_eq!(delivered, vec![original.clone(), original.clone()]);
-    assert_eq!(original["sequence"], 0);
-    let encoded = serde_json::to_string(&original).expect("request JSON");
-    for forbidden in [
-        "private-config-secret",
-        "BEGIN CERTIFICATE",
-        "AKIAIOSFODNN7EXAMPLE",
-        "bucket",
-        "object",
-        "path",
-    ] {
-        assert!(!encoded.contains(forbidden), "request exposed {forbidden}");
-    }
-    assert_eq!(original.as_object().expect("request object").len(), 10);
-    restart.shutdown().await;
-
-    let unchanged_samples = Arc::new(AtomicUsize::new(0));
-    let sampled = unchanged_samples.clone();
-    let unchanged = spawn_inventory_runtime(Some(restart_config), schedule(), &shutdown, move || {
-        sampled.fetch_add(1, Ordering::Relaxed);
-        std::future::ready(Ok(snapshot()))
-    })
-    .expect("restart inventory")
-    .expect("configured inventory");
-    let mut unchanged_status = unchanged.status();
-    assert!(matches!(
-        wait_for(&mut unchanged_status, |status| matches!(status, InventoryStatus::Unchanged { .. })).await,
-        InventoryStatus::Unchanged { content_hash: unchanged } if unchanged == content_hash
-    ));
-    assert_eq!(unchanged_samples.load(Ordering::Relaxed), 1);
-    assert_eq!(restart_server.seen.lock().expect("seen lock").len(), 2);
-    unchanged.shutdown().await;
-}
-
-#[tokio::test]
-async fn connect_inventory_disconnect_retries_without_resampling() {
-    let pki = TestPki::new();
-    let unavailable = server(&pki, Vec::new()).await;
-    let temp = tempfile::tempdir().expect("tempdir");
-    let config = config(&temp, &pki, &unavailable);
-    drop(unavailable);
-    let shutdown = CancellationToken::new();
-    let samples = Arc::new(AtomicUsize::new(0));
-    let sampled = samples.clone();
-    let runtime = spawn_inventory_runtime(Some(config), schedule(), &shutdown, move || {
-        sampled.fetch_add(1, Ordering::Relaxed);
-        std::future::ready(Ok(snapshot()))
-    })
-    .expect("start inventory")
-    .expect("configured inventory");
-    let mut status = runtime.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| {
-            matches!(status, InventoryStatus::BackingOff { delay } if *delay == Duration::from_millis(40))
-        })
-        .await,
-        InventoryStatus::BackingOff { delay } if delay == Duration::from_millis(40)
-    ));
-    assert_eq!(samples.load(Ordering::Relaxed), 1);
-    runtime.shutdown().await;
-}
-
-#[tokio::test]
-async fn connect_inventory_retries_an_incomplete_sample_before_delivery() {
-    let pki = TestPki::new();
-    let content_hash = snapshot().content_hash().expect("content hash");
-    let server = server(&pki, vec![Reply::ok(&content_hash)]).await;
-    let temp = tempfile::tempdir().expect("tempdir");
-    let shutdown = CancellationToken::new();
-    let samples = Arc::new(AtomicUsize::new(0));
-    let sampled = samples.clone();
-    let runtime = spawn_inventory_runtime(Some(config(&temp, &pki, &server)), schedule(), &shutdown, move || {
-        let attempt = sampled.fetch_add(1, Ordering::Relaxed);
-        std::future::ready(if attempt == 0 {
-            Err(rustfs::connect::InventoryError::SnapshotIncomplete {
-                expected: 96,
-                observed: 12,
-            })
-        } else {
-            Ok(snapshot())
-        })
-    })
-    .expect("start inventory")
-    .expect("configured inventory");
-    let mut status = runtime.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::Online { .. })).await,
-        InventoryStatus::Online { content_hash: accepted, .. } if accepted == content_hash
-    ));
-    assert_eq!(samples.load(Ordering::Relaxed), 2);
-    assert_eq!(server.seen.lock().expect("seen lock").len(), 1);
-    runtime.shutdown().await;
-}
-
-#[tokio::test]
-async fn connect_inventory_unchanged_sample_resets_incomplete_backoff() {
-    let pki = TestPki::new();
-    let content_hash = snapshot().content_hash().expect("content hash");
-    let server = server(&pki, vec![Reply::ok(&content_hash)]).await;
-    let temp = tempfile::tempdir().expect("tempdir");
-    let shutdown = CancellationToken::new();
-    let config = config(&temp, &pki, &server);
-    let seed = spawn_inventory_runtime(Some(config.clone()), schedule(), &shutdown, || std::future::ready(Ok(snapshot())))
-        .expect("start inventory")
-        .expect("configured inventory");
-    let mut seed_status = seed.status();
-
-    assert!(matches!(
-        wait_for(&mut seed_status, |status| matches!(status, InventoryStatus::Online { .. })).await,
-        InventoryStatus::Online { content_hash: accepted, .. } if accepted == content_hash
-    ));
-    seed.shutdown().await;
-
-    let samples = Arc::new(AtomicUsize::new(0));
-    let sampled = samples.clone();
-    let runtime = spawn_inventory_runtime(
-        Some(config),
-        InventorySchedule {
-            cadence: Duration::from_millis(100),
-            jitter: Duration::ZERO,
-        },
-        &shutdown,
-        move || {
-            let attempt = sampled.fetch_add(1, Ordering::Relaxed);
-            std::future::ready(if matches!(attempt, 0 | 1 | 3) {
-                Err(rustfs::connect::InventoryError::SnapshotIncomplete {
-                    expected: 96,
-                    observed: 12,
-                })
-            } else {
-                Ok(snapshot())
-            })
-        },
-    )
-    .expect("restart inventory")
-    .expect("configured inventory");
-    let mut status = runtime.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| {
-            matches!(status, InventoryStatus::BackingOff { delay } if *delay == Duration::from_millis(20))
-        })
-        .await,
-        InventoryStatus::BackingOff { delay } if delay == Duration::from_millis(20)
-    ));
-    assert!(matches!(
-        wait_for(&mut status, |status| {
-            matches!(status, InventoryStatus::BackingOff { delay } if *delay == Duration::from_millis(40))
-        })
-        .await,
-        InventoryStatus::BackingOff { delay } if delay == Duration::from_millis(40)
-    ));
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::Unchanged { .. })).await,
-        InventoryStatus::Unchanged { content_hash: unchanged } if unchanged == content_hash
-    ));
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::BackingOff { .. })).await,
-        InventoryStatus::BackingOff { delay } if delay == Duration::from_millis(20)
-    ));
-    assert_eq!(samples.load(Ordering::Relaxed), 4);
-    assert_eq!(server.seen.lock().expect("seen lock").len(), 1);
-    runtime.shutdown().await;
-}
-
-#[tokio::test]
-async fn connect_inventory_revoked_device_stops_without_retrying() {
-    let pki = TestPki::new();
-    let server = server(&pki, vec![Reply::error(StatusCode::UNAUTHORIZED, "DEVICE_REVOKED")]).await;
-    let temp = tempfile::tempdir().expect("tempdir");
-    let shutdown = CancellationToken::new();
-    let runtime = spawn_inventory_runtime(Some(config(&temp, &pki, &server)), schedule(), &shutdown, || {
-        std::future::ready(Ok(snapshot()))
-    })
-    .expect("start inventory")
-    .expect("configured inventory");
-    let mut status = runtime.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::AuthenticationStopped { .. })).await,
-        InventoryStatus::AuthenticationStopped { status: 401, reason: Some(reason) } if reason == "DEVICE_REVOKED"
-    ));
-    assert_eq!(server.seen.lock().expect("seen lock").len(), 1);
-    runtime.shutdown().await;
-}
-
-#[tokio::test]
-async fn connect_inventory_sequence_overflow_fails_before_sampling_or_network_delivery() {
-    let pki = TestPki::new();
-    let server = server(&pki, Vec::new()).await;
-    let temp = tempfile::tempdir().expect("tempdir");
-    let config = config(&temp, &pki, &server);
-    let state = temp.path().join("private-config-secret/inventory/state.json");
-    fs::create_dir_all(state.parent().expect("state directory")).expect("create state directory");
-    fs::write(
-        &state,
-        br#"{"nextSequence":9007199254740992,"pending":null,"lastAcceptedContentHash":null}"#,
-    )
-    .expect("write state");
-    private_mode(&state);
-    let samples = Arc::new(AtomicUsize::new(0));
-    let sampled = samples.clone();
-    let shutdown = CancellationToken::new();
-    let runtime = spawn_inventory_runtime(Some(config), schedule(), &shutdown, move || {
-        sampled.fetch_add(1, Ordering::Relaxed);
-        std::future::ready(Ok(snapshot()))
-    })
-    .expect("start inventory")
-    .expect("configured inventory");
-    let mut status = runtime.status();
-
-    assert!(matches!(
-        wait_for(&mut status, |status| matches!(status, InventoryStatus::Failed { .. })).await,
-        InventoryStatus::Failed { reason } if reason.contains("sequence is exhausted")
-    ));
-    assert_eq!(samples.load(Ordering::Relaxed), 0);
-    assert!(server.seen.lock().expect("seen lock").is_empty());
-    runtime.shutdown().await;
-}
-
-#[cfg(unix)]
-fn private_mode(path: &std::path::Path) {
-    use std::os::unix::fs::PermissionsExt as _;
-    fs::set_permissions(path, fs::Permissions::from_mode(0o600)).expect("private permissions");
-}
-
-#[cfg(not(unix))]
-fn private_mode(_path: &std::path::Path) {}