mirrors/seaweedfs
1
0
Fork 0
mirror of https://github.com/seaweedfs/seaweedfs.git synced 2026-10-21 14:02:35 +00:00
Code Issues Packages Projects Releases Wiki Activity

master: quarantine all copies of a duplicated volume id

Browse Source 
Near-simultaneous registrations let each side quarantine only the other,
so arrival order decided which copy stayed writable. A confirmed
cross-layout collision means either copy could be the impostor, so flag
every colliding copy read-only regardless of registration order.
This commit is contained in:
Chris Lu
2026-10-19 03:10:22 -07:00
parent 1f76b74ef9
commit bdfd31d799
2 changed files with 22 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files
weed/topology/duplicate_volume_id_test.go
+8
View File
@@ -32,6 +32,14 @@ func TestDuplicateVolumeIdAcrossCollections(t *testing.T) {
if !vi.ReadOnly {
t.Fatal("duplicate newcomer should be marked read-only")
}
vi, err = dn1.GetVolumesById(7)
if err != nil {
t.Fatalf("volume info not stored on dn1: %v", err)
}
if !vi.ReadOnly {
t.Fatal("either copy could be the impostor; the earlier copy must go read-only too")
}
}
func TestSameLayoutReplicaIsNotDuplicate(t *testing.T) {
weed/topology/topology.go
+14 -8
View File
@@ -643,27 +643,33 @@ func (t *Topology) RegisterVolumeLayout(v storage.VolumeInfo, dn *DataNode) {
// flagDuplicateVolumeId reports and quarantines v.Id when another data node
// already serves it under a different layout key. It runs after registration
// so a legitimate same-node move (disk type or collection change) settles
// first and does not trip it.
// first and does not trip it. Every copy across the conflicting layouts is
// quarantined: either could be the impostor, and arrival order must not
// decide which copy stays writable.
func (t *Topology) flagDuplicateVolumeId(v storage.VolumeInfo, dn *DataNode) {
dupCollection, nodes := t.findDuplicateVolumeLayout(v, dn)
dupCollection, dupLayout, nodes := t.findDuplicateVolumeLayout(v, dn)
if len(nodes) == 0 {
return
}
stats.MasterDuplicateVolumeIds.WithLabelValues(v.Collection, v.Id.String()).Inc()
glog.Errorf("duplicate volume id %d: %s reports it under collection %q but %v also serves it under collection %q; marking %s's copy read-only", v.Id, dn.Id(), v.Collection, nodes, dupCollection, dn.Id())
glog.Errorf("duplicate volume id %d: %s reports it under collection %q but %v also serves it under collection %q; marking all copies read-only", v.Id, dn.Id(), v.Collection, nodes, dupCollection)
t.quarantinedVolumeIds.Store(quarantinedVolume{dn.Id(), v.Id}, struct{}{})
if vl := t.FindVolumeLayout(v.Collection, v.ReplicaPlacement, v.Ttl, types.ToDiskType(v.DiskType)); vl != nil {
vl.SetVolumeReadOnly(dn, v.Id)
}
for _, n := range nodes {
t.quarantinedVolumeIds.Store(quarantinedVolume{n.Id(), v.Id}, struct{}{})
dupLayout.SetVolumeReadOnly(n, v.Id)
}
}
// findDuplicateVolumeLayout reports the collection and the other data nodes
// findDuplicateVolumeLayout reports the layout and the other data nodes
// already holding v.Id under a different layout key. Volume ids are global,
// so the same id held by two nodes under two layouts means the allocator
// reused an id.
func (t *Topology) findDuplicateVolumeLayout(v storage.VolumeInfo, dn *DataNode) (string, []*DataNode) {
func (t *Topology) findDuplicateVolumeLayout(v storage.VolumeInfo, dn *DataNode) (string, *VolumeLayout, []*DataNode) {
if v.ReplicaPlacement == nil {
return "", nil
return "", nil, nil
}
diskType := types.ToDiskType(v.DiskType)
for _, item := range t.collectionMap.Items() {
@@ -683,11 +689,11 @@ func (t *Topology) findDuplicateVolumeLayout(v storage.VolumeInfo, dn *DataNode)
}
}
if len(others) > 0 {
return c.Name, others
return c.Name, vl, others
}
}
}
return "", nil
return "", nil, nil
}
func (t *Topology) UnRegisterVolumeLayout(v storage.VolumeInfo, dn *DataNode) {