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/scanner/src/remote_scanner/stream.rs
+26 -1
View File
@@ -23,6 +23,7 @@ use crate::storage_api::owner::NS_SCANNER_PROTOCOL_VERSION;
use crate::{
DATA_USAGE_CACHE_NAME, DataUsageCache, DataUsageCachePrepareOutcome, DataUsageCacheSource, DataUsageEntryInfo,
DataUsageScanPlanDigest, Disk, ScannerError, StorageError, resolve_scanner_object_store_handle,
scanner_publication_admission_for_epoch, scanner_publication_epoch,
};
use hmac::{Hmac, KeyInit, Mac};
use rustfs_common::heal_channel::HealScanMode;
@@ -686,6 +687,9 @@ async fn scan_and_persist_local_bucket(
source.pool_index, source.set_index
))
})?;
let expected_publication_epoch = scanner_publication_epoch(set.clone()).await.ok_or_else(|| {
RemoteScannerServerError::worker("remote namespace scanner cache publication is blocked by data movement")
})?;
let cache_name = path_join_buf(&[&bucket, DATA_USAGE_CACHE_NAME]);
let guard = acquire_scanner_cache_locks(set.as_ref(), &cache_name, source)
.await
@@ -708,6 +712,14 @@ async fn scan_and_persist_local_bucket(
"remote namespace scanner cache lock was lost before reusing the current snapshot",
));
}
if scanner_publication_admission_for_epoch(set.clone(), expected_publication_epoch)
.await
.is_none()
{
return Err(RemoteScannerServerError::retry_bucket(
"remote namespace scanner cache publication epoch changed before reusing the current snapshot",
));
}
return Ok(RemoteScannerFrameResult::Complete(Box::new(RemoteScannerComplete {
source,
scan_plan_digest,
@@ -796,9 +808,22 @@ async fn scan_and_persist_local_bucket(
.await
.map_err(|err| RemoteScannerServerError::worker(format!("remote namespace scanner leader fence changed: {err}")))?;
let done_save = Metrics::time(Metric::SaveUsage);
let save_result = cache.save_with_revisions(set, &cache_name, &revisions).await;
// Each physical main/backup PUT must still prove the epoch captured before
// the scan. A movement transition that starts and ends during the scan
// therefore cannot admit the stale cache under the new epoch.
let save_result = cache
.save_with_revisions_for_epoch(set.clone(), &cache_name, &revisions, expected_publication_epoch)
.await;
done_save();
save_result.map_err(|err| RemoteScannerServerError::worker(format!("remote namespace scanner cache save failed: {err}")))?;
if scanner_publication_admission_for_epoch(set, expected_publication_epoch)
.await
.is_none()
{
return Err(RemoteScannerServerError::retry_bucket(
"remote namespace scanner cache publication epoch changed after persistence",
));
}
validate_remote_scanner_request_fence_with_store(next_cycle, leader_epoch, store)
.await
.map_err(|err| RemoteScannerServerError::worker(format!("remote namespace scanner leader fence changed: {err}")))?;