fix(scanner): publish partial usage observations

This commit is contained in:
马登山
2026-08-23 09:43:41 +08:00
parent 20d1266496
commit ec3b7a7dc6
10 changed files with 659 additions and 43 deletions
+108 -1
View File
@@ -236,6 +236,15 @@ pub struct DataUsageInfo {
/// without relying on synchronized clocks.
#[serde(default, skip_serializing_if = "Option::is_none")]
pub usage_snapshot_authoritative_baseline: Option<DataUsageSnapshotIdentity>,
/// Per-set freshness for an observational aggregate. A set entry is
/// never sufficient to make the aggregate authoritative; it only records
/// which last-known-good generation contributed to the view.
#[serde(default, skip_serializing_if = "Vec::is_empty")]
pub usage_snapshot_set_states: Vec<DataUsageSnapshotSetState>,
/// An observational view may contain only the sets that completed this
/// cycle (or retained a compatible last-known-good cache).
#[serde(default)]
pub usage_snapshot_partial: bool,
/// Deprecated kept here for backward compatibility reasons
pub bucket_sizes: HashMap<String, u64>,
/// Per-disk snapshot information when available
@@ -252,6 +261,22 @@ pub struct DataUsageSnapshotIdentity {
pub scanner_epoch: Option<u64>,
}
#[derive(Clone, Debug, Default, Serialize, Deserialize, PartialEq, Eq)]
pub struct DataUsageSnapshotSetState {
pub pool_index: u64,
pub set_index: u64,
#[serde(default)]
pub scanner_cycle: Option<u64>,
#[serde(default)]
pub scanner_epoch: Option<u64>,
#[serde(default, skip_serializing_if = "Option::is_none")]
pub scan_plan_digest: Option<[u8; 32]>,
#[serde(default)]
pub complete: bool,
#[serde(default)]
pub tombstone: bool,
}
impl DataUsageInfo {
pub fn snapshot_identity(&self) -> DataUsageSnapshotIdentity {
DataUsageSnapshotIdentity {
@@ -291,7 +316,7 @@ pub fn data_usage_snapshot_is_newer(candidate: &DataUsageInfo, baseline: &DataUs
/// rollback delete/recreate fences the previous bucket incarnation too.
pub fn observed_data_usage_is_newer(observed: &DataUsageInfo, authoritative: &DataUsageInfo) -> bool {
observed.usage_snapshot_converged == Some(false)
&& observed.is_complete_bucket_usage_snapshot()
&& (observed.is_complete_bucket_usage_snapshot() || observed.is_valid_partial_snapshot())
&& observed.usage_snapshot_authoritative_baseline.as_ref() == Some(&authoritative.snapshot_identity())
&& data_usage_snapshot_is_newer(observed, authoritative)
}
@@ -1436,6 +1461,39 @@ impl DataUsageInfo {
&& u64::try_from(self.buckets_usage.len()).ok() == Some(self.buckets_count)
}
/// Validate provenance before an observational view can be selected for
/// admin display. Partial data is accepted only with unique set states,
/// a plan digest for every state, and at least one usable generation.
pub fn is_valid_partial_snapshot(&self) -> bool {
if !self.usage_snapshot_partial
|| self.usage_snapshot_converged != Some(false)
|| self.last_update.is_none()
|| self.scanner_cycle.is_none()
|| self.scanner_epoch.is_none()
|| self.usage_snapshot_set_states.is_empty()
|| u64::try_from(self.buckets_usage.len()).ok() != Some(self.buckets_count)
{
return false;
}
let mut previous = None;
let mut plan_digest = None;
let mut has_source = false;
for state in &self.usage_snapshot_set_states {
if state.scan_plan_digest.is_none()
|| plan_digest.is_some_and(|digest| Some(digest) != state.scan_plan_digest)
|| state.scanner_cycle.is_some() != state.scanner_epoch.is_some()
|| previous.is_some_and(|(pool, set)| (pool, set) >= (state.pool_index, state.set_index))
{
return false;
}
previous = Some((state.pool_index, state.set_index));
plan_digest = state.scan_plan_digest;
has_source |= state.scanner_cycle.is_some() && !state.tombstone;
}
has_source
}
/// Add object metadata to data usage statistics
pub fn add_object(&mut self, object_path: &str, meta_object: &rustfs_filemeta::MetaObject) {
// This method is kept for backward compatibility
@@ -2263,6 +2321,55 @@ mod tests {
assert!(!observed_data_usage_is_newer(&candidate(2, 9, Some(false), true), &authoritative));
assert!(!observed_data_usage_is_newer(&candidate(2, 11, Some(true), true), &authoritative));
assert!(!observed_data_usage_is_newer(&candidate(2, 11, Some(false), false), &authoritative));
let mut partial = candidate(2, 11, Some(false), false);
partial.usage_snapshot_partial = true;
partial.usage_snapshot_set_states = vec![DataUsageSnapshotSetState {
pool_index: 0,
set_index: 0,
scanner_cycle: Some(10),
scanner_epoch: Some(2),
scan_plan_digest: Some([1; 32]),
complete: false,
tombstone: false,
}];
assert!(observed_data_usage_is_newer(&partial, &authoritative));
}
#[test]
fn mixed_topology_snapshot_is_rejected() {
let mut partial = DataUsageInfo {
last_update: Some(SystemTime::UNIX_EPOCH + Duration::from_secs(2)),
scanner_cycle: Some(11),
scanner_epoch: Some(2),
buckets_count: 0,
usage_snapshot_converged: Some(false),
usage_snapshot_partial: true,
usage_snapshot_set_states: vec![
DataUsageSnapshotSetState {
pool_index: 0,
set_index: 0,
scanner_cycle: Some(11),
scanner_epoch: Some(2),
scan_plan_digest: Some([1; 32]),
complete: true,
tombstone: false,
},
DataUsageSnapshotSetState {
pool_index: 1,
set_index: 0,
scanner_cycle: Some(10),
scanner_epoch: Some(2),
scan_plan_digest: Some([2; 32]),
complete: false,
tombstone: false,
},
],
..Default::default()
};
assert!(!partial.is_valid_partial_snapshot());
partial.usage_snapshot_set_states[1].scan_plan_digest = Some([1; 32]);
assert!(partial.is_valid_partial_snapshot());
}
#[test]