Files
seaweedfs/weed/storage/erasure_coding/ec_shards_info.go
T
Chris Lu d4e39b499b EC placement: shared replica-placement resolver, snapshot + Place core, capacity fixes, tiering (#9621)
* Add shared super_block.ResolveReplicaPlacement; use it in ec_balance

* Add ecbalancer.FromActiveTopology snapshot constructor for EC encode/repair

* Add ecbalancer.Place greenfield/repair placement core (strict + durability-first)

* topology: add GetEffectiveAvailableEcShardSlots; FromActiveTopology uses shard-granular free slots

GetDisksWithEffectiveCapacity flattens reserved shard slots into volume slots via
integer truncation, so an in-flight EC task reserving a non-multiple-of-
DataShardsCount number of shards was lost from the snapshot and freeSlots was
over-reported. GetEffectiveAvailableEcShardSlots subtracts the full reservation
impact at shard granularity.

* ecbalancer.Place: reject nodes without a free disk of the requested type

FromActiveTopology keeps all disk types in the snapshot, so an SSD-only request
could be routed to a node with only HDD capacity (pickBestDiskOnNode then returns
disk 0 on the wrong tier). Filter rack/node selection to those with a free disk
of the requested type.

* ecbalancer.Place: enforce ReplicaPlacement DiffDataCenterCount (per-DC shard cap)

* ecbalancer: enforce DiffDataCenterCount in balance (cross-DC phase + cross-rack DC cap)

Adds a cross-DC corrective phase that drains data centers holding more than
DiffDataCenterCount shards of a volume, and a per-DC cap on cross-rack move
targets. Both are no-ops when DiffDataCenterCount is unset, so balance output is
unchanged for non-DC placements.

* topology: ratio-aware EC shard slots and provisional empty-disk slot

GetEffectiveAvailableEcShardSlots now takes the target collection's data-shard
count, so a 4+2 volume's larger shards are not over-counted at 10 per volume slot;
and it keeps the one provisional slot for freshly started empty servers that
report max=0, matching getEffectiveAvailableCapacityUnsafe. FromActiveTopology
threads the ratio through.

* ecbalancer.Place: explicit disk-type filter signal (fix HDD vs any ambiguity)

HardDriveType normalizes to "", which collided with "" meaning any disk. Add
Constraints.FilterDiskType and normalize both sides so a hdd request matches disks
reported as "" and never leaks to SSD, while filter=false still means any.

* ecbalancer: add clearShardAccounting for repair snapshot reconciliation

Clears one disk's copy of a shard from per-domain accounting and recomputes the
node-level union (preserving a kept copy on another disk of the same node), without
crediting capacity. Repair uses it to drop to-be-deleted copies before placing
missing shards.

* ecbalancer: don't cap cross-DC target racks when DiffRackCount is unset

len(racks)+1 wrongly limited each target rack (3 in a 2-rack cluster), so draining
a DC could stop short of the DiffDataCenterCount cap. Use MaxShardCount+1 as the
effectively-unlimited default.

* topology/ecbalancer: ratio-correct EC capacity accounting

Reservation shard slots (default ShardsPerVolumeSlot units) are now converted to
the target ratio before subtracting, and existing EC shards are charged by size
(targetDataShards/shardDataShards) so a 2+1 shard isn't counted as one 10+4 slot.
Per-shard ratio lookup is behind shardDataShards (OSS uses the standard ratio).

* ecbalancer.Place: candidate tiering and eligible-rack caps

Adds a per-disk eligibility/preference abstraction so Place supports:
- preferred-tag whole-plan retry (try disks carrying the earliest tags first,
  widen to all only if a tier cannot place every shard; reports
  SpilledOutsidePreferredTags),
- soft disk-type spill via DiskTypePolicy (Any/Prefer/Require): Prefer fills the
  preferred type then spills, reporting SpilledToOtherDiskType; Require filters,
- even per-rack caps that divide by racks holding an eligible disk, so a tiered
  cluster (e.g. SSDs in 2 of 4 racks) isn't capped impossibly low.
Disk tags carried via Node.AddDiskTags + FromActiveTopology.

* ecbalancer: export ClearShardAccounting for repair snapshot reconciliation

* ecbalancer: address review feedback (ratio rounding, bitmap walk, same-DC moves)

- topology/ecbalancer: round shard-reservation and existing-shard footprint up
  when converting to target-ratio shard slots, so a sub-slot reservation is not
  truncated to zero and free capacity is not overstated for low-data-shard
  layouts (targetDataShards < ds).
- erasure_coding: add ShardBits.All iterator and use it across the balancer,
  cross-DC phase, and placement scoring instead of scanning 0..MaxShardCount and
  probing Has on every id.
- ecbalancer: allow same-DC cross-rack moves when a DC already sits at its
  DiffDataCenterCount cap; a same-DC move leaves the DC total unchanged. Add a
  regression test that fails without the guard.
- ecbalancer cross-DC phase: pick targets via the eligible-aware
  pickNodeInRackEligible/pickBestDiskEligible helpers so the disk-type filter is
  honored and a 0 disk id is not mistaken for a valid selection.

* ecbalancer: test ecShardSlotsOnDisk fractional round-up

Cover the mixed-ratio path (targetDataShards < existing data shards) so a
shard's fractional footprint is never floored to zero and free capacity is not
overstated. Exercises the round-up via the targetDataShards parameter; OSS uses
the standard ratio at runtime while the enterprise build hits it with real
per-volume ratios.

* ecbalancer: assert node B rack in TestFromActiveTopology

* ecbalancer: split Destination into separate DataCenter and bare Rack

Replace the composite "dc:rack" Rack field on Destination with separate
DataCenter and bare Rack values, matching topology.DiskInfo and the worker-task
convention. Callers (and tests) read the data center directly instead of parsing
the composite with strings.SplitN.

* shell ec.balance: use utilization-based global balancing (parity with worker)

The shell's global rebalance phase balanced by raw shard count; switch it to
fractional fullness (shards/capacity), as the worker already does. On uniform
capacity the two agree; on heterogeneous capacity it fills nodes proportionally
instead of driving small-capacity nodes toward full.

Updates the heterogeneous-capacity regression test to assert even fullness
(~equal shards/capacity per node) rather than even shard count.

* ecbalancer: bounded-proportional per-DC shard spread

DiffDataCenterCount was enforced only as a ceiling (drain-to-cap), which could
leave a within-cap-but-lopsided DC distribution under a loose cap (e.g. 10/4 of 14
with cap=10). Now the cross-DC phase, the cross-rack DC guard, and Place all target
boundedMaxPerDC = min(DiffDataCenterCount, max(ceil(total/numDCs), parityShards)):
shards spread proportionally across DCs, but no tighter than the durability floor
(once each DC holds <= parityShards a DC loss is recoverable, so further spreading
only adds cross-DC/WAN traffic). No-op when DiffDataCenterCount is 0; identical to
before when the cap is the binding constraint.

* ecbalancer: drop DiffDataCenterCount enforcement for EC placement

The 1-byte volume ReplicaPlacement packs xyz into x*100+y*10+z<=255, so the DC
digit can only be 0-2 -- far too small to be a meaningful per-DC EC shard cap (a
cap of 1-2 would demand 7-14 DCs for a 10+4 volume). It's volume replica-placement,
not an EC spec. Removes the cross-DC balance phase, the DC guard in the cross-rack
phase, and the per-DC cap in Place (and the just-added bounded-proportional logic);
EC relies on the RP-independent rack/node even spread instead. Rack/node caps
(DiffRackCount/SameRackCount) are unchanged. Per-domain EC caps are left for a real
EC placement spec.

* ecbalancer: enforce per-disk durability cap; symmetric reserve/release

Place now refuses to put more than parityShards shards of a volume on a single
disk (pickBestDiskEligible skips a disk once it holds parityShards of the volume,
a hard cap not relaxed even in durability-first). Previously Place assigned by
free capacity, so a skewed near-full cluster could pile >parityShards onto one
disk -> losing it loses the volume; only distinct-disk count was checked. This
covers encode and repair (both route through Place); the caller skips/leaves the
volume rather than minting an unrecoverable layout.

Also makes reserveShard decrement freeSlots unconditionally, symmetric with
releaseShard's unconditional increment (the old guarded decrement could credit a
phantom slot on release if a shard were ever reserved onto a full disk).

* ecbalancer: add Topology.ReleaseVolumeShards (clear + credit) for greenfield encode

Releases all of a volume's shards from the snapshot and credits the freed disk
capacity, so a greenfield encode can plan as if stale EC shards from a prior failed
attempt are gone. Safe to credit because the encode task deletes stale shards
(cleanupStaleEcShards) before distributing the new ones. Distinct from
ClearShardAccounting (repair), which does not credit.

* ecbalancer: ReleaseVolumeShards credits node freeSlots, not just disks

releaseShard only increments per-disk freeSlots, but rack capacity is summed from
node freeSlots (buildRacks) and node freeSlots gates node eligibility. Crediting
only disks left a node/rack looking full after releasing stale shards, so a
greenfield encode still couldn't use the freed capacity. Now credits the node by
the total disk-slots freed.

* ecbalancer: correct PlacementMode docs (encode uses durability-first)

PlaceStrict was labeled '(encode)' but encode uses PlaceDurabilityFirst. Clarify
that durability-first is used by both encode and repair, reports relaxations in
PlaceResult.Relaxed, and never relaxes the per-disk durability cap.

* ecbalancer: treat SameRackCount as a direct per-node shard cap

The 3rd ReplicaPlacement digit now caps shards per node at exactly the digit
value, matching how DiffRackCount (2nd digit) caps per rack, instead of allowing
digit+1 per node. This makes the per-rack and per-node caps consistent and
matches the documented "digits cap EC shards per rack and per node" semantics;
e.g. 011 now means at most one shard per rack and one per node.
2026-05-22 20:22:09 -07:00

406 lines
9.8 KiB
Go

package erasure_coding
import (
"fmt"
"iter"
"math/bits"
"sort"
"strings"
"sync"
"github.com/dustin/go-humanize"
"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
)
// ShardBits is a bitmap representing which shards are present (bit 0 = shard 0, etc.)
type ShardBits uint32
// Has checks if a shard ID is present in the bitmap
func (sb ShardBits) Has(id ShardId) bool {
return id < MaxShardCount && sb&(1<<id) != 0
}
// Set sets a shard ID in the bitmap
func (sb ShardBits) Set(id ShardId) ShardBits {
if id >= MaxShardCount {
return sb
}
return sb | (1 << id)
}
// Clear clears a shard ID from the bitmap
func (sb ShardBits) Clear(id ShardId) ShardBits {
if id >= MaxShardCount {
return sb
}
return sb &^ (1 << id)
}
// Count returns the number of set bits using popcount
func (sb ShardBits) Count() int {
return bits.OnesCount32(uint32(sb))
}
// All iterates the shard ids present in the bitmap, in ascending order. It walks
// only the set bits (trailing-zero scan), so cost scales with the number of
// shards present rather than the full id range. Prefer this over scanning
// 0..MaxShardCount and calling Has on each id.
func (sb ShardBits) All() iter.Seq[ShardId] {
return func(yield func(ShardId) bool) {
for b := uint32(sb); b != 0; b &= b - 1 {
if !yield(ShardId(bits.TrailingZeros32(b))) {
return
}
}
}
}
// ShardsInfo encapsulates information for EC shards with memory-efficient storage
type ShardsInfo struct {
mu sync.RWMutex
shards []ShardInfo // Sorted by Id
shardBits ShardBits
}
func NewShardsInfo() *ShardsInfo {
return &ShardsInfo{
shards: make([]ShardInfo, 0, TotalShardsCount),
}
}
// Initializes a ShardsInfo from a VolumeEcShardInformationMessage proto.
func ShardsInfoFromVolumeEcShardInformationMessage(vi *master_pb.VolumeEcShardInformationMessage) *ShardsInfo {
res := NewShardsInfo()
if vi == nil {
return res
}
var id ShardId
var j int
// Build shards directly to avoid locking in Set() since res is not yet shared
newShards := make([]ShardInfo, 0, 8)
for bitmap := vi.EcIndexBits; bitmap != 0; bitmap >>= 1 {
if bitmap&1 != 0 {
var size ShardSize
if j < len(vi.ShardSizes) {
size = ShardSize(vi.ShardSizes[j])
}
j++
newShards = append(newShards, NewShardInfo(id, size))
}
id++
}
res.shards = newShards
res.shardBits = ShardBits(vi.EcIndexBits)
return res
}
// Returns a count of shards from a VolumeEcShardInformationMessage proto.
func GetShardCount(vi *master_pb.VolumeEcShardInformationMessage) int {
if vi == nil {
return 0
}
return ShardBits(vi.EcIndexBits).Count()
}
// EcShardsTotalSize returns the sum of all shard sizes (data + parity) in
// the message. Walks vi.ShardSizes directly rather than materializing a
// ShardsInfo, which is significantly cheaper for callers that only need the
// aggregate size.
func EcShardsTotalSize(vi *master_pb.VolumeEcShardInformationMessage) int64 {
if vi == nil {
return 0
}
var total int64
for _, s := range vi.ShardSizes {
total += s
}
return total
}
// EcShardsDataSize returns the sum of sizes for data shards only (parity
// shards excluded). Data shards are those with id < dataShards; all higher
// shard ids are treated as parity. Passing dataShards <= 0 falls back to
// the upstream default of DataShardsCount (10), which is correct for the
// fixed 10+4 layout. Forks with per-volume ratio metadata (e.g. the
// data_shards field carried on an extended VolumeEcShardInformationMessage)
// should pass the per-volume value so logical sizes remain accurate under
// custom EC policies like 6+3 or 16+6.
func EcShardsDataSize(vi *master_pb.VolumeEcShardInformationMessage, dataShards int) int64 {
if vi == nil {
return 0
}
if dataShards <= 0 {
dataShards = DataShardsCount
}
var total int64
var id ShardId
var j int
for bitmap := vi.EcIndexBits; bitmap != 0; bitmap >>= 1 {
if bitmap&1 != 0 {
if int(id) < dataShards && j < len(vi.ShardSizes) {
total += vi.ShardSizes[j]
}
j++
}
id++
}
return total
}
// Returns a string representation for a ShardsInfo.
func (sp *ShardsInfo) String() string {
sp.mu.RLock()
defer sp.mu.RUnlock()
var sb strings.Builder
for i, s := range sp.shards {
if i > 0 {
sb.WriteString(" ")
}
fmt.Fprintf(&sb, "%d:%s", s.Id, humanize.Bytes(uint64(s.Size)))
}
return sb.String()
}
// AsSlice converts a ShardsInfo to a slice of ShardInfo structs, ordered by shard ID.
func (si *ShardsInfo) AsSlice() []ShardInfo {
si.mu.RLock()
defer si.mu.RUnlock()
res := make([]ShardInfo, len(si.shards))
copy(res, si.shards)
return res
}
// Count returns the number of EC shards using popcount on the bitmap.
func (si *ShardsInfo) Count() int {
si.mu.RLock()
defer si.mu.RUnlock()
return si.shardBits.Count()
}
// Has verifies if a shard ID is present using bitmap check.
func (si *ShardsInfo) Has(id ShardId) bool {
si.mu.RLock()
defer si.mu.RUnlock()
return si.shardBits.Has(id)
}
// Ids returns a list of shard IDs, in ascending order.
func (si *ShardsInfo) Ids() []ShardId {
si.mu.RLock()
defer si.mu.RUnlock()
ids := make([]ShardId, len(si.shards))
for i, s := range si.shards {
ids[i] = s.Id
}
return ids
}
// IdsInt returns a list of shards ID as int, in ascending order.
func (si *ShardsInfo) IdsInt() []int {
ids := si.Ids()
res := make([]int, len(ids))
for i, id := range ids {
res[i] = int(id)
}
return res
}
// IdsUint32 returns a list of shards ID as uint32, in ascending order.
func (si *ShardsInfo) IdsUint32() []uint32 {
return ShardIdsToUint32(si.Ids())
}
// Set sets or updates a shard's information.
func (si *ShardsInfo) Set(shard ShardInfo) {
if shard.Id >= MaxShardCount {
return
}
si.mu.Lock()
defer si.mu.Unlock()
// Check if already exists
if si.shardBits.Has(shard.Id) {
// Find and update
idx := si.findIndex(shard.Id)
if idx >= 0 {
si.shards[idx] = shard
}
return
}
// Add new shard
si.shardBits = si.shardBits.Set(shard.Id)
// Find insertion point to keep sorted
idx := sort.Search(len(si.shards), func(i int) bool {
return si.shards[i].Id > shard.Id
})
// Insert at idx
si.shards = append(si.shards, ShardInfo{})
copy(si.shards[idx+1:], si.shards[idx:])
si.shards[idx] = shard
}
// Delete deletes a shard by ID.
func (si *ShardsInfo) Delete(id ShardId) {
if id >= MaxShardCount {
return
}
si.mu.Lock()
defer si.mu.Unlock()
if !si.shardBits.Has(id) {
return // Not present
}
si.shardBits = si.shardBits.Clear(id)
// Find and remove from slice
idx := si.findIndex(id)
if idx >= 0 {
si.shards = append(si.shards[:idx], si.shards[idx+1:]...)
}
}
// Bitmap returns a bitmap for all existing shard IDs.
func (si *ShardsInfo) Bitmap() uint32 {
si.mu.RLock()
defer si.mu.RUnlock()
return uint32(si.shardBits)
}
// Size returns the size of a given shard ID, if present.
func (si *ShardsInfo) Size(id ShardId) ShardSize {
if id >= MaxShardCount {
return 0
}
si.mu.RLock()
defer si.mu.RUnlock()
if !si.shardBits.Has(id) {
return 0
}
idx := si.findIndex(id)
if idx >= 0 {
return si.shards[idx].Size
}
return 0
}
// TotalSize returns the size for all shards.
func (si *ShardsInfo) TotalSize() ShardSize {
si.mu.RLock()
defer si.mu.RUnlock()
var total ShardSize
for _, s := range si.shards {
total += s.Size
}
return total
}
// Sizes returns a compact slice of present shard sizes, from first to last.
func (si *ShardsInfo) Sizes() []ShardSize {
si.mu.RLock()
defer si.mu.RUnlock()
res := make([]ShardSize, len(si.shards))
for i, s := range si.shards {
res[i] = s.Size
}
return res
}
// SizesInt64 returns a compact slice of present shard sizes, from first to last, as int64.
func (si *ShardsInfo) SizesInt64() []int64 {
sizes := si.Sizes()
res := make([]int64, len(sizes))
for i, s := range sizes {
res[i] = int64(s)
}
return res
}
// Copy creates a copy of a ShardInfo.
func (si *ShardsInfo) Copy() *ShardsInfo {
si.mu.RLock()
defer si.mu.RUnlock()
newShards := make([]ShardInfo, len(si.shards))
copy(newShards, si.shards)
return &ShardsInfo{
shards: newShards,
shardBits: si.shardBits,
}
}
// DeleteParityShards removes parity shards from a ShardInfo.
func (si *ShardsInfo) DeleteParityShards() {
for id := DataShardsCount; id < TotalShardsCount; id++ {
si.Delete(ShardId(id))
}
}
// MinusParityShards creates a ShardInfo copy, but with parity shards removed.
func (si *ShardsInfo) MinusParityShards() *ShardsInfo {
result := si.Copy()
result.DeleteParityShards()
return result
}
// Add merges all shards from another ShardInfo into this one.
func (si *ShardsInfo) Add(other *ShardsInfo) {
other.mu.RLock()
// Copy shards to avoid holding lock on 'other' while calling si.Set, which could deadlock.
shardsToAdd := make([]ShardInfo, len(other.shards))
copy(shardsToAdd, other.shards)
other.mu.RUnlock()
for _, s := range shardsToAdd {
si.Set(s)
}
}
// Subtract removes all shards present on another ShardInfo.
func (si *ShardsInfo) Subtract(other *ShardsInfo) {
other.mu.RLock()
// Copy shards to avoid holding lock on 'other' while calling si.Delete, which could deadlock.
shardsToRemove := make([]ShardInfo, len(other.shards))
copy(shardsToRemove, other.shards)
other.mu.RUnlock()
for _, s := range shardsToRemove {
si.Delete(s.Id)
}
}
// Plus returns a new ShardInfo consisting of (this + other).
func (si *ShardsInfo) Plus(other *ShardsInfo) *ShardsInfo {
result := si.Copy()
result.Add(other)
return result
}
// Minus returns a new ShardInfo consisting of (this - other).
func (si *ShardsInfo) Minus(other *ShardsInfo) *ShardsInfo {
result := si.Copy()
result.Subtract(other)
return result
}
// findIndex finds the index of a shard by ID using binary search.
// Must be called with lock held. Returns -1 if not found.
func (si *ShardsInfo) findIndex(id ShardId) int {
idx := sort.Search(len(si.shards), func(i int) bool {
return si.shards[i].Id >= id
})
if idx < len(si.shards) && si.shards[idx].Id == id {
return idx
}
return -1
}