gsadmin/rustfs
1
0
Fork 0
mirror of https://github.com/rustfs/rustfs.git synced 2026-08-26 05:56:50 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix(scanner): fence system metadata publication (#6444)

Browse Source 
* feat(scanner): fence usage publication during data movement

* fix(scanner): detect movement refresh state changes

* fix(scanner): fence publication during data movement

* fix(scanner): close movement epoch publication races

* fix(scanner): fence movement-sensitive publication paths

* fix(scanner): fence cache and heal recovery paths

* fix(scanner): carry publication epoch through scan cycle

* fix(scanner): recheck remote cache epoch after save

* fix(scanner): recheck local cache epoch before publish

* fix(scanner): fence data usage writers and baseline

* fix(scanner): expose decommission activity to publication fence

* fix(scanner): release publication gate before reads

* fix(scanner): complete publication fence integration

* fix(scanner): avoid empty usage baseline publication

* chore(scanner): gate test-only helpers

* fix: use decommission canceler in reload test

---------

Co-authored-by: houseme <housemecn@gmail.com>
This commit is contained in:
cxymds
2026-08-23 17:28:21 +08:00
committed by GitHub
parent 9cda615519
commit e196a134cc
26 changed files with 2465 additions and 370 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/ecstore/src/set_disk/mod.rs
+20
View File
@@ -3614,6 +3614,26 @@ impl SetDisks {
&self.ctx
}
/// Admit one short scanner cache publication under this set's instance
/// movement fence. The caller must hold the returned guard through its
/// final conditional cache write; no scan-round work belongs under it.
pub async fn scanner_data_usage_publication_admission_guard(&self) -> Option<(tokio::sync::OwnedRwLockReadGuard<()>, u64)> {
let operation_gate = self.ctx.data_movement_operation_gate();
let operation_guard = operation_gate.read_owned().await;
if self.ctx.scanner_publication_state_allowed() {
let epoch = self.ctx.data_movement_operation_epoch();
return Some((operation_guard, epoch));
}
// The owner deliberately marks the cached state UNKNOWN after every
// movement epoch advance. Do not strand remote scanner writers in that
// state: release this guard before asking the storage owner to refresh
// its durable movement snapshot, since the owner uses the same gate.
drop(operation_guard);
let owner = runtime_sources::object_store_handle().filter(|owner| Arc::ptr_eq(&owner.ctx, &self.ctx))?;
owner.scanner_data_usage_publication_admission_guard().await
}
/// Whether both sets' namespace-lock implementations cover the same object key.
pub(crate) async fn shares_namespace_lock_domain(&self, other: &Self) -> bool {
match (self.ctx.is_dist_erasure().await, other.ctx.is_dist_erasure().await) {