Files
seaweedfs/weed/admin/dash/ec_shard_management.go
T
Chris Lu 52d74df4d1 clients: stream the volume listings that ask for everything (#10679)
* master: stream volume listings

A listing of 800k volumes is 36MB on the wire but 305MB as messages, and the
master built all of it, then held it while grpc encoded it. Two of those at
once is most of a small master's heap, and the maintenance scanner asks every
30 minutes.

The topology goes out first, listing nothing, then its volumes in batches, so
the master holds a batch rather than a cluster: 341MB of live heap for one
listing becomes 4.4MB. It allocates much the same either way -- what changes is
how much of it has to be live at once, which is what sets the heap ceiling.

Batches are built under their disk's lock and sent outside it, so a slow reader
stalls the stream rather than the topology. They therefore do not share one
instant, which a single listing did not either: it takes each disk's lock in
turn, so a volume moving during either can be seen twice or not at all.

The client helper hides which kind of master answered: one too old for the
stream is asked the old way and its reply cut into the same batches. Either way
the topology handed over lists no volumes, so a caller cannot come to depend on
finding them there.

* admin: stream the listing the maintenance scan reads

It asks for every volume in the cluster every 30 minutes. Reassembling it
client-side keeps the scan identical -- ActiveTopology splits disks by the
disk ids on the volumes, so it needs them in the topology -- while the master
no longer builds the whole reply to send it.

* topology: report a disk id that does not depend on map order

A topology disk that fronts several physical disks took its reported id from
whichever volume the map yielded first, so two listings of an unchanged disk
could disagree. Take the smallest instead.

* topology: test that a streamed listing rebuilds to the whole one

The callers that stream now rebuild the listing from a topology sent without
volumes plus the batches after it, so that has to come out the same as being
sent it whole, at every batch size and under a filter.

* clients: stream the volume listings that ask for everything

The dashboard's list and export pages, the collection and ec shard pages, the
topology view, the worker metrics and two shell commands each asked the master
to build all 800k volumes into one reply. They read the same listing as before,
rebuilt on their side, so the master no longer holds it.

The three that already ask for one volume or one collection stay as they are:
their replies are small, and streaming one costs a round trip to say so.
2026-08-10 09:51:08 -07:00

683 lines
18 KiB
Go

package dash
import (
"context"
"fmt"
"sort"
"time"
"github.com/seaweedfs/seaweedfs/weed/pb"
"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
"github.com/seaweedfs/seaweedfs/weed/storage/erasure_coding"
)
// matchesCollection checks if a volume/EC volume collection matches the filter collection.
// Handles the special case where empty collection ("") represents the "default" collection.
func matchesCollection(volumeCollection, filterCollection string) bool {
// Handle special case where "default" filter matches empty collection
if filterCollection == "default" && volumeCollection == "" {
return true
}
// Direct string match for named collections
return volumeCollection == filterCollection
}
// GetClusterEcShards retrieves cluster EC shards data with pagination, sorting, and filtering
func (s *AdminServer) GetClusterEcShards(page int, pageSize int, sortBy string, sortOrder string, collection string) (*ClusterEcShardsData, error) {
// Set defaults
if page < 1 {
page = 1
}
if pageSize < 1 || pageSize > 1000 {
pageSize = 100
}
if sortBy == "" {
sortBy = "volume_id"
}
if sortOrder == "" {
sortOrder = "asc"
}
var ecShards []EcShardWithInfo
volumeShardsMap := make(map[uint32]map[int]bool) // volumeId -> set of shards present
volumesWithAllShards := 0
volumesWithMissingShards := 0
// Get detailed EC shard information via gRPC
err := s.WithMasterClient(func(client master_pb.SeaweedClient) error {
resp, err := pb.CollectVolumeList(context.Background(), client, &master_pb.VolumeListRequest{})
if err != nil {
return err
}
if resp.TopologyInfo != nil {
for _, dc := range resp.TopologyInfo.DataCenterInfos {
for _, rack := range dc.RackInfos {
for _, node := range rack.DataNodeInfos {
for _, diskInfo := range node.DiskInfos {
// Process EC shard information
for _, ecShardInfo := range diskInfo.EcShardInfos {
volumeId := ecShardInfo.Id
// Initialize volume shards map if needed
if volumeShardsMap[volumeId] == nil {
volumeShardsMap[volumeId] = make(map[int]bool)
}
// Create individual shard entries for each shard this server has
shardBits := ecShardInfo.EcIndexBits
for shardId := 0; shardId < erasure_coding.MaxShardCount; shardId++ {
if (shardBits & (1 << uint(shardId))) != 0 {
// Mark this shard as present for this volume
volumeShardsMap[volumeId][shardId] = true
ecShard := EcShardWithInfo{
VolumeID: volumeId,
ShardID: uint32(shardId),
Collection: ecShardInfo.Collection,
Size: 0, // EC shards don't have individual size in the API response
Server: node.Id,
DataCenter: dc.Id,
Rack: rack.Id,
DiskType: diskInfo.Type,
ModifiedTime: 0, // Not available in current API
EcIndexBits: ecShardInfo.EcIndexBits,
ShardCount: getShardCount(ecShardInfo.EcIndexBits),
}
ecShards = append(ecShards, ecShard)
}
}
}
}
}
}
}
}
return nil
})
if err != nil {
return nil, err
}
// Calculate volume-level completeness (across all servers)
volumeCompleteness := make(map[uint32]bool)
volumeMissingShards := make(map[uint32][]int)
for volumeId, shardsPresent := range volumeShardsMap {
var missingShards []int
shardCount := len(shardsPresent)
// Find which shards are missing for this volume across ALL servers
// Uses default 10+4 (14 total shards)
for shardId := 0; shardId < erasure_coding.TotalShardsCount; shardId++ {
if !shardsPresent[shardId] {
missingShards = append(missingShards, shardId)
}
}
isComplete := (shardCount == erasure_coding.TotalShardsCount)
volumeCompleteness[volumeId] = isComplete
volumeMissingShards[volumeId] = missingShards
if isComplete {
volumesWithAllShards++
} else {
volumesWithMissingShards++
}
}
// Update completeness info for each shard based on volume-level completeness
for i := range ecShards {
volumeId := ecShards[i].VolumeID
ecShards[i].IsComplete = volumeCompleteness[volumeId]
ecShards[i].MissingShards = volumeMissingShards[volumeId]
}
// Filter by collection if specified
if collection != "" {
var filteredShards []EcShardWithInfo
for _, shard := range ecShards {
if shard.Collection == collection {
filteredShards = append(filteredShards, shard)
}
}
ecShards = filteredShards
}
// Sort the results
sortEcShards(ecShards, sortBy, sortOrder)
// Calculate statistics for conditional display
dataCenters := make(map[string]bool)
racks := make(map[string]bool)
collections := make(map[string]bool)
for _, shard := range ecShards {
dataCenters[shard.DataCenter] = true
racks[shard.Rack] = true
if shard.Collection != "" {
collections[shard.Collection] = true
}
}
// Pagination
totalShards := len(ecShards)
totalPages := (totalShards + pageSize - 1) / pageSize
startIndex := (page - 1) * pageSize
endIndex := startIndex + pageSize
if endIndex > totalShards {
endIndex = totalShards
}
if startIndex >= totalShards {
startIndex = 0
endIndex = 0
}
paginatedShards := ecShards[startIndex:endIndex]
// Build response
data := &ClusterEcShardsData{
EcShards: paginatedShards,
TotalShards: totalShards,
TotalVolumes: len(volumeShardsMap),
LastUpdated: time.Now(),
// Pagination
CurrentPage: page,
TotalPages: totalPages,
PageSize: pageSize,
// Sorting
SortBy: sortBy,
SortOrder: sortOrder,
// Statistics
DataCenterCount: len(dataCenters),
RackCount: len(racks),
CollectionCount: len(collections),
// Conditional display flags
ShowDataCenterColumn: len(dataCenters) > 1,
ShowRackColumn: len(racks) > 1,
ShowCollectionColumn: len(collections) > 1 || collection != "",
// Filtering
FilterCollection: collection,
// EC specific statistics
ShardsPerVolume: make(map[uint32]int), // This will be recalculated below
VolumesWithAllShards: volumesWithAllShards,
VolumesWithMissingShards: volumesWithMissingShards,
}
// Recalculate ShardsPerVolume for the response
for volumeId, shardsPresent := range volumeShardsMap {
data.ShardsPerVolume[volumeId] = len(shardsPresent)
}
// Set single values when only one exists
if len(dataCenters) == 1 {
for dc := range dataCenters {
data.SingleDataCenter = dc
break
}
}
if len(racks) == 1 {
for rack := range racks {
data.SingleRack = rack
break
}
}
if len(collections) == 1 {
for col := range collections {
data.SingleCollection = col
break
}
}
return data, nil
}
// GetClusterEcVolumes retrieves cluster EC volumes data grouped by volume ID with shard locations
func (s *AdminServer) GetClusterEcVolumes(page int, pageSize int, sortBy string, sortOrder string, collection string) (*ClusterEcVolumesData, error) {
// Set defaults
if page < 1 {
page = 1
}
if pageSize < 1 || pageSize > 1000 {
pageSize = 100
}
if sortBy == "" {
sortBy = "volume_id"
}
if sortOrder == "" {
sortOrder = "asc"
}
volumeData := make(map[uint32]*EcVolumeWithShards)
totalShards := 0
// Get detailed EC shard information via gRPC
err := s.WithMasterClient(func(client master_pb.SeaweedClient) error {
resp, err := pb.CollectVolumeList(context.Background(), client, &master_pb.VolumeListRequest{})
if err != nil {
return err
}
if resp.TopologyInfo != nil {
for _, dc := range resp.TopologyInfo.DataCenterInfos {
for _, rack := range dc.RackInfos {
for _, node := range rack.DataNodeInfos {
for _, diskInfo := range node.DiskInfos {
// Process EC shard information
for _, ecShardInfo := range diskInfo.EcShardInfos {
volumeId := ecShardInfo.Id
// Initialize volume data if needed
if volumeData[volumeId] == nil {
volumeData[volumeId] = &EcVolumeWithShards{
VolumeID: volumeId,
Collection: ecShardInfo.Collection,
TotalShards: 0,
IsComplete: false,
MissingShards: []int{},
ShardLocations: make(map[int]string),
ShardSizes: make(map[int]int64),
DataCenters: []string{},
Servers: []string{},
Racks: []string{},
}
}
volume := volumeData[volumeId]
// Track data centers and servers
dcExists := false
for _, existingDc := range volume.DataCenters {
if existingDc == dc.Id {
dcExists = true
break
}
}
if !dcExists {
volume.DataCenters = append(volume.DataCenters, dc.Id)
}
serverExists := false
for _, existingServer := range volume.Servers {
if existingServer == node.Id {
serverExists = true
break
}
}
if !serverExists {
volume.Servers = append(volume.Servers, node.Id)
}
// Track racks
rackExists := false
for _, existingRack := range volume.Racks {
if existingRack == rack.Id {
rackExists = true
break
}
}
if !rackExists {
volume.Racks = append(volume.Racks, rack.Id)
}
// Process each shard this server has for this volume
shardBits := ecShardInfo.EcIndexBits
shardSizes := ecShardInfo.ShardSizes
sizeIndex := 0
for shardId := 0; shardId < erasure_coding.MaxShardCount; shardId++ {
if (shardBits & (1 << uint(shardId))) != 0 {
// Record shard location
volume.ShardLocations[shardId] = node.Id
if sizeIndex < len(shardSizes) {
size := shardSizes[sizeIndex]
if size >= 0 {
volume.ShardSizes[shardId] = size
}
}
sizeIndex++
totalShards++
}
}
}
}
}
}
}
}
return nil
})
if err != nil {
return nil, err
}
// Calculate completeness for each volume
completeVolumes := 0
incompleteVolumes := 0
for _, volume := range volumeData {
volume.TotalShards = len(volume.ShardLocations)
// Find missing shards (default 10+4 = 14 total shards)
var missingShards []int
for shardId := 0; shardId < erasure_coding.TotalShardsCount; shardId++ {
if _, exists := volume.ShardLocations[shardId]; !exists {
missingShards = append(missingShards, shardId)
}
}
volume.MissingShards = missingShards
volume.IsComplete = (len(missingShards) == 0)
if volume.IsComplete {
completeVolumes++
} else {
incompleteVolumes++
}
}
// Convert map to slice
var ecVolumes []EcVolumeWithShards
for _, volume := range volumeData {
// Filter by collection if specified
if collection == "" || matchesCollection(volume.Collection, collection) {
ecVolumes = append(ecVolumes, *volume)
}
}
// Sort the results
sortEcVolumes(ecVolumes, sortBy, sortOrder)
// Calculate statistics for conditional display
dataCenters := make(map[string]bool)
collections := make(map[string]bool)
for _, volume := range ecVolumes {
for _, dc := range volume.DataCenters {
dataCenters[dc] = true
}
if volume.Collection != "" {
collections[volume.Collection] = true
}
}
// Pagination
totalVolumes := len(ecVolumes)
totalPages := (totalVolumes + pageSize - 1) / pageSize
startIndex := (page - 1) * pageSize
endIndex := startIndex + pageSize
if endIndex > totalVolumes {
endIndex = totalVolumes
}
if startIndex >= totalVolumes {
startIndex = 0
endIndex = 0
}
paginatedVolumes := ecVolumes[startIndex:endIndex]
// Build response
data := &ClusterEcVolumesData{
EcVolumes: paginatedVolumes,
TotalVolumes: totalVolumes,
LastUpdated: time.Now(),
// Pagination
Page: page,
PageSize: pageSize,
TotalPages: totalPages,
// Sorting
SortBy: sortBy,
SortOrder: sortOrder,
// Filtering
Collection: collection,
// Conditional display flags
ShowDataCenterColumn: len(dataCenters) > 1,
ShowRackColumn: false, // We don't track racks in this view for simplicity
ShowCollectionColumn: len(collections) > 1 || collection != "",
// Statistics
CompleteVolumes: completeVolumes,
IncompleteVolumes: incompleteVolumes,
TotalShards: totalShards,
}
return data, nil
}
// sortEcVolumes sorts EC volumes based on the specified field and order
func sortEcVolumes(volumes []EcVolumeWithShards, sortBy string, sortOrder string) {
sort.Slice(volumes, func(i, j int) bool {
var less bool
switch sortBy {
case "volume_id":
less = volumes[i].VolumeID < volumes[j].VolumeID
case "collection":
if volumes[i].Collection == volumes[j].Collection {
less = volumes[i].VolumeID < volumes[j].VolumeID
} else {
less = volumes[i].Collection < volumes[j].Collection
}
case "total_shards":
if volumes[i].TotalShards == volumes[j].TotalShards {
less = volumes[i].VolumeID < volumes[j].VolumeID
} else {
less = volumes[i].TotalShards < volumes[j].TotalShards
}
case "completeness":
// Complete volumes first, then by volume ID
if volumes[i].IsComplete == volumes[j].IsComplete {
less = volumes[i].VolumeID < volumes[j].VolumeID
} else {
less = volumes[i].IsComplete && !volumes[j].IsComplete
}
default:
less = volumes[i].VolumeID < volumes[j].VolumeID
}
if sortOrder == "desc" {
return !less
}
return less
})
}
// getShardCount returns the number of shards represented by the bitmap
func getShardCount(ecIndexBits uint32) int {
count := 0
for i := 0; i < erasure_coding.MaxShardCount; i++ {
if (ecIndexBits & (1 << uint(i))) != 0 {
count++
}
}
return count
}
// sortEcShards sorts EC shards based on the specified field and order
func sortEcShards(shards []EcShardWithInfo, sortBy string, sortOrder string) {
sort.Slice(shards, func(i, j int) bool {
var less bool
switch sortBy {
case "shard_id":
less = shards[i].ShardID < shards[j].ShardID
case "server":
if shards[i].Server == shards[j].Server {
less = shards[i].ShardID < shards[j].ShardID // Secondary sort by shard ID
} else {
less = shards[i].Server < shards[j].Server
}
case "data_center":
if shards[i].DataCenter == shards[j].DataCenter {
less = shards[i].ShardID < shards[j].ShardID // Secondary sort by shard ID
} else {
less = shards[i].DataCenter < shards[j].DataCenter
}
case "rack":
if shards[i].Rack == shards[j].Rack {
less = shards[i].ShardID < shards[j].ShardID // Secondary sort by shard ID
} else {
less = shards[i].Rack < shards[j].Rack
}
default:
less = shards[i].ShardID < shards[j].ShardID
}
if sortOrder == "desc" {
return !less
}
return less
})
}
// GetEcVolumeDetails retrieves detailed information about a specific EC volume
func (s *AdminServer) GetEcVolumeDetails(volumeID uint32, sortBy string, sortOrder string) (*EcVolumeDetailsData, error) {
// Set defaults
if sortBy == "" {
sortBy = "shard_id"
}
if sortOrder == "" {
sortOrder = "asc"
}
var shards []EcShardWithInfo
var collection string
dataCenters := make(map[string]bool)
servers := make(map[string]bool)
// Get detailed EC shard information for the specific volume via gRPC
err := s.WithMasterClient(func(client master_pb.SeaweedClient) error {
resp, err := client.VolumeList(context.Background(), &master_pb.VolumeListRequest{VolumeId: volumeID})
if err != nil {
return err
}
if resp.TopologyInfo != nil {
for _, dc := range resp.TopologyInfo.DataCenterInfos {
for _, rack := range dc.RackInfos {
for _, node := range rack.DataNodeInfos {
for _, diskInfo := range node.DiskInfos {
// Process EC shard information for this specific volume
for _, ecShardInfo := range diskInfo.EcShardInfos {
// An older master ignores the filter.
if ecShardInfo.Id == volumeID {
collection = ecShardInfo.Collection
dataCenters[dc.Id] = true
servers[node.Id] = true
// Create individual shard entries for each shard this server has
shardBits := ecShardInfo.EcIndexBits
shardSizes := ecShardInfo.ShardSizes
sizeIndex := 0
for shardId := 0; shardId < erasure_coding.MaxShardCount; shardId++ {
if (shardBits & (1 << uint(shardId))) != 0 {
var shardSize uint64
if sizeIndex < len(shardSizes) {
size := shardSizes[sizeIndex]
if size >= 0 {
shardSize = uint64(size)
}
}
sizeIndex++
ecShard := EcShardWithInfo{
VolumeID: ecShardInfo.Id,
ShardID: uint32(shardId),
Collection: ecShardInfo.Collection,
Size: shardSize,
Server: node.Id,
DataCenter: dc.Id,
Rack: rack.Id,
DiskType: diskInfo.Type,
ModifiedTime: 0, // Not available in current API
EcIndexBits: ecShardInfo.EcIndexBits,
ShardCount: getShardCount(ecShardInfo.EcIndexBits),
}
shards = append(shards, ecShard)
}
}
}
}
}
}
}
}
}
return nil
})
if err != nil {
return nil, err
}
if len(shards) == 0 {
return nil, fmt.Errorf("EC volume %d not found", volumeID)
}
// Calculate completeness based on unique shard IDs
foundShards := make(map[int]bool)
for _, shard := range shards {
foundShards[int(shard.ShardID)] = true
}
totalUniqueShards := len(foundShards)
// Check completeness using default 10+4 (14 total shards)
isComplete := (totalUniqueShards == erasure_coding.TotalShardsCount)
// Calculate missing shards
var missingShards []int
for i := 0; i < erasure_coding.TotalShardsCount; i++ {
if !foundShards[i] {
missingShards = append(missingShards, i)
}
}
// Update completeness info for each shard
for i := range shards {
shards[i].IsComplete = isComplete
shards[i].MissingShards = missingShards
}
// Sort shards based on parameters
sortEcShards(shards, sortBy, sortOrder)
// Convert maps to slices
var dcList []string
for dc := range dataCenters {
dcList = append(dcList, dc)
}
var serverList []string
for server := range servers {
serverList = append(serverList, server)
}
data := &EcVolumeDetailsData{
VolumeID: volumeID,
Collection: collection,
Shards: shards,
TotalShards: totalUniqueShards,
IsComplete: isComplete,
MissingShards: missingShards,
DataCenters: dcList,
Servers: serverList,
LastUpdated: time.Now(),
SortBy: sortBy,
SortOrder: sortOrder,
}
return data, nil
}