Files
seaweedfs/weed/topology/topology_ec.go
T
Chris Lu da243b9423 fix(ec): group orphan-source completeness by encode generation (topology encode_ts_ns) (#9952)
* feat(ec): carry the encode generation through the topology heartbeat

Add encode_ts_ns (field 14) to VolumeEcShardInformationMessage and
populate it from each EC volume's .vif identity. The volume server emits
it on the full and incremental heartbeats; the master stores it on
EcVolumeInfo and re-emits it via GetTopologyInfo, so the admin/worker
layer can see which encode run produced each shard set. Field 14 avoids
the enterprise fork's reserved 10-13. Mirror the proto field and both
heartbeat emit sites in the Rust volume server.

* fix(ec): group orphan-source shard completeness by encode generation

countExistingEcShardsForVolume ORed EcIndexBits across every disk, so two
interrupted encode runs whose shard sets overlap unioned into a
false-complete set -- triggering the orphaned-source delete while no
single generation was actually complete. Group shards by encode_ts_ns and
return the largest single generation's count, so the trigger fires only
when one run holds the full set. Shards from pre-upgrade servers
(encode_ts_ns==0) form their own bucket.

The heartbeat carries one encode_ts_ns per (volume, disk), so this
separates generations on different disks; same-disk mixing is prevented
upstream by the pre-encode artifact wipe and the cross-run read guard.

* fix(ec): guard against a nil Ec shard info entry in the generation count

Defensive: a manually-constructed or corrupted topology could carry a nil
entry in EcShardInfos. Skip it rather than dereference.

* fix(ec): carry the encode generation on the EC shard unmount delta

The mount delta sets EncodeTsNs; the unmount deletion delta left it 0.
Populate it from the Ec volume before unloading so both incremental
deltas are consistent (the Rust volume server already does this via its
snapshot diff).
2026-06-14 00:14:12 -07:00

212 lines
6.1 KiB
Go

package topology
import (
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
"github.com/seaweedfs/seaweedfs/weed/storage/erasure_coding"
"github.com/seaweedfs/seaweedfs/weed/storage/needle"
)
type EcShardLocations struct {
Collection string
// Use MaxShardCount (32) to support custom EC ratios
Locations [erasure_coding.MaxShardCount][]*DataNode
}
func (t *Topology) SyncDataNodeEcShards(shardInfos []*master_pb.VolumeEcShardInformationMessage, dn *DataNode) (newShards, deletedShards []*erasure_coding.EcVolumeInfo) {
// convert into in memory struct storage.VolumeInfo
var shards []*erasure_coding.EcVolumeInfo
for _, shardInfo := range shardInfos {
// Create EcVolumeInfo directly with optimized format
ecVolumeInfo := &erasure_coding.EcVolumeInfo{
VolumeId: needle.VolumeId(shardInfo.Id),
Collection: shardInfo.Collection,
ShardsInfo: erasure_coding.ShardsInfoFromVolumeEcShardInformationMessage(shardInfo),
DiskType: shardInfo.DiskType,
DiskId: shardInfo.DiskId,
ExpireAtSec: shardInfo.ExpireAtSec,
FileCount: shardInfo.FileCount,
DeleteCount: shardInfo.DeleteCount,
EncodeTsNs: shardInfo.EncodeTsNs,
}
shards = append(shards, ecVolumeInfo)
}
// find out the delta volumes
newShards, deletedShards = dn.UpdateEcShards(shards)
for _, v := range newShards {
t.RegisterEcShards(v, dn)
}
for _, v := range deletedShards {
t.UnRegisterEcShards(v, dn)
}
return
}
func (t *Topology) IncrementalSyncDataNodeEcShards(newEcShards, deletedEcShards []*master_pb.VolumeEcShardInformationMessage, dn *DataNode) {
// convert into in memory struct storage.VolumeInfo
var newShards, deletedShards []*erasure_coding.EcVolumeInfo
for _, shardInfo := range newEcShards {
// Create EcVolumeInfo directly with optimized format
ecVolumeInfo := &erasure_coding.EcVolumeInfo{
VolumeId: needle.VolumeId(shardInfo.Id),
Collection: shardInfo.Collection,
ShardsInfo: erasure_coding.ShardsInfoFromVolumeEcShardInformationMessage(shardInfo),
DiskType: shardInfo.DiskType,
DiskId: shardInfo.DiskId,
ExpireAtSec: shardInfo.ExpireAtSec,
FileCount: shardInfo.FileCount,
DeleteCount: shardInfo.DeleteCount,
EncodeTsNs: shardInfo.EncodeTsNs,
}
newShards = append(newShards, ecVolumeInfo)
}
for _, shardInfo := range deletedEcShards {
// Create EcVolumeInfo directly with optimized format
ecVolumeInfo := &erasure_coding.EcVolumeInfo{
VolumeId: needle.VolumeId(shardInfo.Id),
Collection: shardInfo.Collection,
ShardsInfo: erasure_coding.ShardsInfoFromVolumeEcShardInformationMessage(shardInfo),
DiskType: shardInfo.DiskType,
DiskId: shardInfo.DiskId,
ExpireAtSec: shardInfo.ExpireAtSec,
FileCount: shardInfo.FileCount,
DeleteCount: shardInfo.DeleteCount,
EncodeTsNs: shardInfo.EncodeTsNs,
}
deletedShards = append(deletedShards, ecVolumeInfo)
}
dn.DeltaUpdateEcShards(newShards, deletedShards)
for _, v := range newShards {
t.RegisterEcShards(v, dn)
}
for _, v := range deletedShards {
t.UnRegisterEcShards(v, dn)
}
}
func NewEcShardLocations(collection string) *EcShardLocations {
return &EcShardLocations{
Collection: collection,
}
}
func (loc *EcShardLocations) AddShard(shardId erasure_coding.ShardId, dn *DataNode) (added bool) {
// Defensive bounds check to prevent panic with out-of-range shard IDs
if int(shardId) >= erasure_coding.MaxShardCount {
return false
}
dataNodes := loc.Locations[shardId]
for _, n := range dataNodes {
if n.Id() == dn.Id() {
return false
}
}
loc.Locations[shardId] = append(dataNodes, dn)
return true
}
func (loc *EcShardLocations) DeleteShard(shardId erasure_coding.ShardId, dn *DataNode) (deleted bool) {
// Defensive bounds check to prevent panic with out-of-range shard IDs
if int(shardId) >= erasure_coding.MaxShardCount {
return false
}
dataNodes := loc.Locations[shardId]
foundIndex := -1
for index, n := range dataNodes {
if n.Id() == dn.Id() {
foundIndex = index
}
}
if foundIndex < 0 {
return false
}
loc.Locations[shardId] = append(dataNodes[:foundIndex], dataNodes[foundIndex+1:]...)
return true
}
func (t *Topology) RegisterEcShards(ecvi *erasure_coding.EcVolumeInfo, dn *DataNode) {
// EC-only volumes (source volume deleted after encoding) must bump
// maxVolumeId too, or a heartbeat-rebuilt master could re-issue their id.
t.UpAdjustMaxVolumeId(ecvi.VolumeId)
t.ecShardMapLock.Lock()
defer t.ecShardMapLock.Unlock()
locations, found := t.ecShardMap[ecvi.VolumeId]
if !found {
locations = NewEcShardLocations(ecvi.Collection)
t.ecShardMap[ecvi.VolumeId] = locations
}
for _, shardId := range ecvi.ShardsInfo.Ids() {
locations.AddShard(shardId, dn)
}
}
func (t *Topology) UnRegisterEcShards(ecvi *erasure_coding.EcVolumeInfo, dn *DataNode) {
glog.Infof("removing ec shard info:%+v", ecvi)
t.ecShardMapLock.Lock()
defer t.ecShardMapLock.Unlock()
locations, found := t.ecShardMap[ecvi.VolumeId]
if !found {
return
}
for _, shardId := range ecvi.ShardsInfo.Ids() {
locations.DeleteShard(shardId, dn)
}
}
func (t *Topology) LookupEcShards(vid needle.VolumeId) (locations *EcShardLocations, found bool) {
t.ecShardMapLock.RLock()
defer t.ecShardMapLock.RUnlock()
locations, found = t.ecShardMap[vid]
return
}
func (t *Topology) ListEcServersByCollection(collection string) (dataNodes []pb.ServerAddress) {
t.ecShardMapLock.RLock()
defer t.ecShardMapLock.RUnlock()
dateNodeMap := make(map[pb.ServerAddress]bool)
for _, ecVolumeLocation := range t.ecShardMap {
if ecVolumeLocation.Collection == collection {
for _, locations := range ecVolumeLocation.Locations {
for _, loc := range locations {
dateNodeMap[loc.ServerAddress()] = true
}
}
}
}
for k, _ := range dateNodeMap {
dataNodes = append(dataNodes, k)
}
return
}
func (t *Topology) DeleteEcCollection(collection string) {
t.ecShardMapLock.Lock()
defer t.ecShardMapLock.Unlock()
var vids []needle.VolumeId
for vid, ecVolumeLocation := range t.ecShardMap {
if ecVolumeLocation.Collection == collection {
vids = append(vids, vid)
}
}
for _, vid := range vids {
delete(t.ecShardMap, vid)
}
}