acid-ant/frostfs-node
1
0
Fork 0
forked from TrueCloudLab/frostfs-node
Code Issues Pull requests Projects Releases Packages Wiki Activity

[#536] blobovnicza: Drop cache

Browse source 
Each blobovnicza instance is opened
while is in use.

Signed-off-by: Dmitrii Stepanov <d.stepanov@yadro.com>
This commit is contained in:
Dmitrii Stepanov 2023-08-30 23:36:48 +03:00
parent b9b86d2ec8
commit c672f59ab8
16 changed files with 586 additions and 474 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
internal/logs/logs.go
View file

@ -509,4 +509,5 @@ const (
RuntimeSoftMemoryLimitUpdated = "soft runtime memory limit value updated"
RuntimeSoftMemoryDefinedWithGOMEMLIMIT = "soft runtime memory defined with GOMEMLIMIT environment variable, config value skipped"
FailedToCountWritecacheItems = "failed to count writecache items"
AttemtToCloseAlreadyClosedBlobovnicza = "attempt to close an already closed blobovnicza"
)

5
pkg/local_object_storage/blobovnicza/blobovnicza_test.go
View file

@ -3,7 +3,6 @@ package blobovnicza
import (
"context"
"crypto/rand"
"errors"
"os"
"testing"
@ -98,9 +97,9 @@ func TestBlobovnicza(t *testing.T) {
testPutGet(t, blz, oidtest.Address(), objSizeLim, nil, nil)
}
// from now objects should not be saved
// blobovnizca accepts object event if full
testPutGet(t, blz, oidtest.Address(), 1024, func(err error) bool {
return errors.Is(err, ErrFull)
return err == nil
}, nil)
require.NoError(t, blz.Close())

8
pkg/local_object_storage/blobovnicza/put.go
View file

@ -23,10 +23,6 @@ type PutPrm struct {
type PutRes struct {
}
// ErrFull is returned when trying to save an
// object to a filled blobovnicza.
var ErrFull = logicerr.New("blobovnicza is full")
// SetAddress sets the address of the saving object.
func (p *PutPrm) SetAddress(addr oid.Address) {
p.addr = addr
@ -65,10 +61,6 @@ func (b *Blobovnicza) Put(ctx context.Context, prm PutPrm) (PutRes, error) {
key := addressKey(prm.addr)
err := b.boltDB.Batch(func(tx *bbolt.Tx) error {
if b.full() {
return ErrFull
}
buck := tx.Bucket(bucketName)
if buck == nil {
// expected to happen:

2
pkg/local_object_storage/blobovnicza/sizes.go
View file

@ -55,6 +55,6 @@ func (b *Blobovnicza) itemDeleted(itemSize uint64) {
b.metrics.SubOpenBlobovniczaItems(1)
}
func (b *Blobovnicza) full() bool {
func (b *Blobovnicza) IsFull() bool {
return b.dataSize.Load() >= b.fullSizeLimit
}

213
pkg/local_object_storage/blobstor/blobovniczatree/active.go Normal file
View file

@ -0,0 +1,213 @@
package blobovniczatree
import (
"path/filepath"
"sync"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobovnicza"
utilSync "git.frostfs.info/TrueCloudLab/frostfs-node/pkg/util/sync"
)
type activeDB struct {
blz *blobovnicza.Blobovnicza
shDB *sharedDB
}
func (db *activeDB) Blobovnicza() *blobovnicza.Blobovnicza {
return db.blz
}
func (db *activeDB) Close() {
db.shDB.Close()
}
func (db *activeDB) Path() string {
return db.shDB.Path()
}
// activeDBManager manages active blobovnicza instances (that is, those that are being used for Put).
//
// Uses dbManager for opening/closing sharedDB instances.
// Stores a reference to an open active sharedDB, so dbManager does not close it.
// When changing the active sharedDB, releases the reference to the previous active sharedDB.
type activeDBManager struct {
levelToActiveDBGuard *sync.RWMutex
levelToActiveDB map[string]*sharedDB
levelLock *utilSync.KeyLocker[string]
closed bool
dbManager *dbManager
leafWidth uint64
}
func newActiveDBManager(dbManager *dbManager, leafWidth uint64) *activeDBManager {
return &activeDBManager{
levelToActiveDBGuard: &sync.RWMutex{},
levelToActiveDB: make(map[string]*sharedDB),
levelLock: utilSync.NewKeyLocker[string](),
dbManager: dbManager,
leafWidth: leafWidth,
}
}
// GetOpenedActiveDBForLevel returns active DB for level.
// DB must be closed after use.
func (m *activeDBManager) GetOpenedActiveDBForLevel(lvlPath string) (*activeDB, error) {
activeDB, err := m.getCurrentActiveIfOk(lvlPath)
if err != nil {
return nil, err
}
if activeDB != nil {
return activeDB, nil
}
return m.updateAndGetActive(lvlPath)
}
func (m *activeDBManager) Open() {
m.levelToActiveDBGuard.Lock()
defer m.levelToActiveDBGuard.Unlock()
m.closed = false
}
func (m *activeDBManager) Close() {
m.levelToActiveDBGuard.Lock()
defer m.levelToActiveDBGuard.Unlock()
for _, db := range m.levelToActiveDB {
db.Close()
}
m.levelToActiveDB = make(map[string]*sharedDB)
m.closed = true
}
func (m *activeDBManager) getCurrentActiveIfOk(lvlPath string) (*activeDB, error) {
m.levelToActiveDBGuard.RLock()
defer m.levelToActiveDBGuard.RUnlock()
if m.closed {
return nil, errClosed
}
db, ok := m.levelToActiveDB[lvlPath]
if !ok {
return nil, nil
}
blz, err := db.Open() //open db for usage, will be closed on activeDB.Close()
if err != nil {
return nil, err
}
if blz.IsFull() {
db.Close()
return nil, nil
}
return &activeDB{
blz: blz,
shDB: db,
}, nil
}
func (m *activeDBManager) updateAndGetActive(lvlPath string) (*activeDB, error) {
m.levelLock.Lock(lvlPath)
defer m.levelLock.Unlock(lvlPath)
current, err := m.getCurrentActiveIfOk(lvlPath)
if err != nil {
return nil, err
}
if current != nil {
return current, nil
}
nextShDB, err := m.getNextSharedDB(lvlPath)
if err != nil {
return nil, err
}
if nextShDB == nil {
return nil, nil
}
blz, err := nextShDB.Open() // open db for client, client must call Close() after usage
if err != nil {
return nil, err
}
return &activeDB{
blz: blz,
shDB: nextShDB,
}, nil
}
func (m *activeDBManager) getNextSharedDB(lvlPath string) (*sharedDB, error) {
var idx uint64
var iterCount uint64
hasActive, currentIdx := m.hasActiveDB(lvlPath)
if hasActive {
idx = (currentIdx + 1) % m.leafWidth
}
var next *sharedDB
for iterCount < m.leafWidth {
path := filepath.Join(lvlPath, u64ToHexString(idx))
shDB := m.dbManager.GetByPath(path)
db, err := shDB.Open() //open db to hold active DB open, will be closed if db is full, after m.replace or by activeDBManager.Close()
if err != nil {
return nil, err
}
if db.IsFull() {
shDB.Close()
} else {
next = shDB
break
}
idx = (idx + 1) % m.leafWidth
iterCount++
}
previous, updated := m.replace(lvlPath, next)
if !updated && next != nil {
next.Close() // manager is closed, so don't hold active DB open
}
if updated && previous != nil {
previous.Close()
}
return next, nil
}
func (m *activeDBManager) hasActiveDB(lvlPath string) (bool, uint64) {
m.levelToActiveDBGuard.RLock()
defer m.levelToActiveDBGuard.RUnlock()
if m.closed {
return false, 0
}
db, ok := m.levelToActiveDB[lvlPath]
if !ok {
return false, 0
}
return true, u64FromHexString(filepath.Base(db.Path()))
}
func (m *activeDBManager) replace(lvlPath string, shDB *sharedDB) (*sharedDB, bool) {
m.levelToActiveDBGuard.Lock()
defer m.levelToActiveDBGuard.Unlock()
if m.closed {
return nil, false
}
previous := m.levelToActiveDB[lvlPath]
if shDB == nil {
delete(m.levelToActiveDB, lvlPath)
} else {
m.levelToActiveDB[lvlPath] = shDB
}
return previous, true
}

143
pkg/local_object_storage/blobstor/blobovniczatree/blobovnicza.go
View file

@ -3,19 +3,12 @@ package blobovniczatree
import (
"errors"
"fmt"
"path/filepath"
"strconv"
"sync"
"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobovnicza"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobstor/common"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobstor/compression"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/util/logicerr"
oid "git.frostfs.info/TrueCloudLab/frostfs-sdk-go/object/id"
"git.frostfs.info/TrueCloudLab/hrw"
"github.com/hashicorp/golang-lru/v2/simplelru"
"go.uber.org/zap"
)
// Blobovniczas represents the storage of the "small" objects.
@ -61,28 +54,8 @@ import (
type Blobovniczas struct {
cfg
// cache of opened filled Blobovniczas
opened *simplelru.LRU[string, *blobovnicza.Blobovnicza]
// lruMtx protects opened cache.
// It isn't RWMutex because `Get` calls must
// lock this mutex on write, as LRU info is updated.
// It must be taken after activeMtx in case when eviction is possible
// i.e. `Add`, `Purge` and `Remove` calls.
lruMtx sync.Mutex
// mutex to exclude parallel bbolt.Open() calls
// bbolt.Open() deadlocks if it tries to open already opened file
openMtx sync.Mutex
// list of active (opened, non-filled) Blobovniczas
activeMtx sync.RWMutex
active map[string]blobovniczaWithIndex
}
type blobovniczaWithIndex struct {
ind uint64
blz *blobovnicza.Blobovnicza
commondbManager *dbManager
activeDBManager *activeDBManager
}
var _ common.Storage = (*Blobovniczas)(nil)
@ -102,120 +75,12 @@ func NewBlobovniczaTree(opts ...Option) (blz *Blobovniczas) {
blz.blzLeafWidth = blz.blzShallowWidth
}
cache, err := simplelru.NewLRU[string, *blobovnicza.Blobovnicza](blz.openedCacheSize, func(p string, value *blobovnicza.Blobovnicza) {
lvlPath := filepath.Dir(p)
if b, ok := blz.active[lvlPath]; ok && b.ind == u64FromHexString(filepath.Base(p)) {
// This branch is taken if we have recently updated active blobovnicza and remove
// it from opened cache.
return
} else if err := value.Close(); err != nil {
blz.log.Error(logs.BlobovniczatreeCouldNotCloseBlobovnicza,
zap.String("id", p),
zap.String("error", err.Error()),
)
} else {
blz.log.Debug(logs.BlobovniczatreeBlobovniczaSuccessfullyClosedOnEvict,
zap.String("id", p),
)
}
})
if err != nil {
// occurs only if the size is not positive
panic(fmt.Errorf("could not create LRU cache of size %d: %w", blz.openedCacheSize, err))
}
activeMapCapacity := uint64(1)
for i := uint64(0); i < blz.blzShallowDepth; i++ {
if i+1 == blz.blzShallowDepth {
activeMapCapacity *= blz.blzLeafWidth
} else {
activeMapCapacity *= blz.blzShallowWidth
}
}
blz.opened = cache
blz.active = make(map[string]blobovniczaWithIndex, activeMapCapacity)
blz.commondbManager = newDBManager(blz.rootPath, blz.blzOpts, blz.blzLeafWidth, blz.readOnly, blz.metrics.Blobovnicza(), blz.log)
blz.activeDBManager = newActiveDBManager(blz.commondbManager, blz.blzLeafWidth)
return blz
}
// activates and returns activated blobovnicza of p-level (dir).
//
// returns error if blobvnicza could not be activated.
func (b *Blobovniczas) getActivated(lvlPath string) (blobovniczaWithIndex, error) {
return b.updateAndGet(lvlPath, nil)
}
// updates active blobovnicza of p-level (dir).
//
// if current active blobovnicza's index is not old, it remains unchanged.
func (b *Blobovniczas) updateActive(lvlPath string, old *uint64) error {
b.log.Debug(logs.BlobovniczatreeUpdatingActiveBlobovnicza, zap.String("path", lvlPath))
_, err := b.updateAndGet(lvlPath, old)
b.log.Debug(logs.BlobovniczatreeActiveBlobovniczaSuccessfullyUpdated, zap.String("path", lvlPath))
return err
}
// updates and returns active blobovnicza of p-level (dir).
//
// if current active blobovnicza's index is not old, it is returned unchanged.
func (b *Blobovniczas) updateAndGet(lvlPath string, old *uint64) (blobovniczaWithIndex, error) {
b.activeMtx.RLock()
active, ok := b.active[lvlPath]
b.activeMtx.RUnlock()
if ok {
if old != nil {
if active.ind == b.blzLeafWidth-1 {
return active, logicerr.New("no more Blobovniczas")
} else if active.ind != *old {
// sort of CAS in order to control concurrent
// updateActive calls
return active, nil
}
} else {
return active, nil
}
active.ind++
}
var err error
if active.blz, err = b.openBlobovnicza(filepath.Join(lvlPath, u64ToHexString(active.ind))); err != nil {
return active, err
}
b.activeMtx.Lock()
defer b.activeMtx.Unlock()
// check 2nd time to find out if it blobovnicza was activated while thread was locked
tryActive, ok := b.active[lvlPath]
if ok && tryActive.blz == active.blz {
return tryActive, nil
}
// Remove from opened cache (active blobovnicza should always be opened).
// Because `onEvict` callback is called in `Remove`, we need to update
// active map beforehand.
b.active[lvlPath] = active
activePath := filepath.Join(lvlPath, u64ToHexString(active.ind))
b.lruMtx.Lock()
b.opened.Remove(activePath)
if ok {
b.opened.Add(filepath.Join(lvlPath, u64ToHexString(tryActive.ind)), tryActive.blz)
}
b.lruMtx.Unlock()
b.log.Debug(logs.BlobovniczatreeBlobovniczaSuccessfullyActivated,
zap.String("path", activePath))
return active, nil
}
// returns hash of the object address.
func addressHash(addr *oid.Address, path string) uint64 {
var a string

59
pkg/local_object_storage/blobstor/blobovniczatree/concurrency_test.go Normal file
View file

@ -0,0 +1,59 @@
package blobovniczatree
import (
"context"
"sync"
"sync/atomic"
"testing"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobstor/common"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/internal/testutil"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/util/logger/test"
"github.com/stretchr/testify/require"
)
func TestBlobovniczaTree_Concurrency(t *testing.T) {
t.Parallel()
const n = 1000
st := NewBlobovniczaTree(
WithLogger(test.NewLogger(t, true)),
WithObjectSizeLimit(1024),
WithBlobovniczaShallowWidth(10),
WithBlobovniczaShallowDepth(1),
WithRootPath(t.TempDir()))
require.NoError(t, st.Open(false))
require.NoError(t, st.Init())
t.Cleanup(func() {
require.NoError(t, st.Close())
})
objGen := &testutil.SeqObjGenerator{ObjSize: 1}
var cnt atomic.Int64
var wg sync.WaitGroup
for i := 0; i < 1000; i++ {
wg.Add(1)
go func() {
defer wg.Done()
for cnt.Add(1) <= n {
obj := objGen.Next()
addr := testutil.AddressFromObject(t, obj)
raw, err := obj.Marshal()
require.NoError(t, err)
_, err = st.Put(context.Background(), common.PutPrm{
Address: addr,
RawData: raw,
})
require.NoError(t, err)
_, err = st.Get(context.Background(), common.GetPrm{Address: addr})
require.NoError(t, err) // fails very often, correlated to how many goroutines are started
}
}()
}
wg.Wait()
}

111
pkg/local_object_storage/blobstor/blobovniczatree/control.go
View file

@ -2,11 +2,8 @@ package blobovniczatree
import (
"context"
"fmt"
"path/filepath"
"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobovnicza"
"go.uber.org/zap"
)
@ -14,6 +11,7 @@ import (
func (b *Blobovniczas) Open(readOnly bool) error {
b.readOnly = readOnly
b.metrics.SetMode(readOnly)
b.openManagers()
return nil
}
@ -23,59 +21,35 @@ func (b *Blobovniczas) Open(readOnly bool) error {
func (b *Blobovniczas) Init() error {
b.log.Debug(logs.BlobovniczatreeInitializingBlobovniczas)
b.openManagers()
if b.readOnly {
b.log.Debug(logs.BlobovniczatreeReadonlyModeSkipBlobovniczasInitialization)
return nil
}
return b.iterateLeaves(context.TODO(), func(p string) (bool, error) {
blz, err := b.openBlobovniczaNoCache(p)
shBlz := b.openBlobovniczaNoCache(p)
_, err := shBlz.Open()
if err != nil {
return true, err
}
defer blz.Close()
if err := blz.Init(); err != nil {
return true, fmt.Errorf("could not initialize blobovnicza structure %s: %w", p, err)
}
defer shBlz.Close()
b.log.Debug(logs.BlobovniczatreeBlobovniczaSuccessfullyInitializedClosing, zap.String("id", p))
return false, nil
})
}
func (b *Blobovniczas) openManagers() {
b.commondbManager.Open() //order important
b.activeDBManager.Open()
}
// Close implements common.Storage.
func (b *Blobovniczas) Close() error {
b.activeMtx.Lock()
b.lruMtx.Lock()
for p, v := range b.active {
if err := v.blz.Close(); err != nil {
b.log.Debug(logs.BlobovniczatreeCouldNotCloseActiveBlobovnicza,
zap.String("path", p),
zap.String("error", err.Error()),
)
}
b.opened.Remove(p)
}
for _, k := range b.opened.Keys() {
blz, _ := b.opened.Get(k)
if err := blz.Close(); err != nil {
b.log.Debug(logs.BlobovniczatreeCouldNotCloseActiveBlobovnicza,
zap.String("path", k),
zap.String("error", err.Error()),
)
}
b.opened.Remove(k)
}
b.active = make(map[string]blobovniczaWithIndex)
b.metrics.Close()
b.lruMtx.Unlock()
b.activeMtx.Unlock()
b.activeDBManager.Close() //order important
b.commondbManager.Close()
return nil
}
@ -83,61 +57,10 @@ func (b *Blobovniczas) Close() error {
// opens and returns blobovnicza with path p.
//
// If blobovnicza is already opened and cached, instance from cache is returned w/o changes.
func (b *Blobovniczas) openBlobovnicza(p string) (*blobovnicza.Blobovnicza, error) {
b.lruMtx.Lock()
v, ok := b.opened.Get(p)
b.lruMtx.Unlock()
if ok {
// blobovnicza should be opened in cache
return v, nil
}
lvlPath := filepath.Dir(p)
curIndex := u64FromHexString(filepath.Base(p))
b.activeMtx.RLock()
defer b.activeMtx.RUnlock()
active, ok := b.active[lvlPath]
if ok && active.ind == curIndex {
return active.blz, nil
}
b.lruMtx.Lock()
defer b.lruMtx.Unlock()
v, ok = b.opened.Get(p)
if ok {
return v, nil
}
blz, err := b.openBlobovniczaNoCache(p)
if err != nil {
return nil, err
}
b.opened.Add(p, blz)
return blz, nil
func (b *Blobovniczas) openBlobovnicza(p string) *sharedDB {
return b.openBlobovniczaNoCache(p)
}
func (b *Blobovniczas) openBlobovniczaNoCache(p string) (*blobovnicza.Blobovnicza, error) {
b.openMtx.Lock()
defer b.openMtx.Unlock()
path := filepath.Join(b.rootPath, p)
blz := blobovnicza.New(append(b.blzOpts,
blobovnicza.WithReadOnly(b.readOnly),
blobovnicza.WithPath(path),
blobovnicza.WithMetrics(b.metrics.Blobovnicza()),
)...)
if err := blz.Open(); err != nil {
return nil, fmt.Errorf("could not open blobovnicza %s: %w", p, err)
}
if err := blz.Init(); err != nil {
return nil, fmt.Errorf("could not init blobovnicza %s: %w", p, err)
}
return blz, nil
func (b *Blobovniczas) openBlobovniczaNoCache(p string) *sharedDB {
return b.commondbManager.GetByPath(p)
}

67
pkg/local_object_storage/blobstor/blobovniczatree/delete.go
View file

@ -3,7 +3,6 @@ package blobovniczatree
import (
"context"
"encoding/hex"
"path/filepath"
"time"
"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
@ -48,10 +47,12 @@ func (b *Blobovniczas) Delete(ctx context.Context, prm common.DeletePrm) (res co
if prm.StorageID != nil {
id := blobovnicza.NewIDFromBytes(prm.StorageID)
blz, err := b.openBlobovnicza(id.String())
shBlz := b.openBlobovnicza(id.String())
blz, err := shBlz.Open()
if err != nil {
return res, err
}
defer shBlz.Close()
if res, err = b.deleteObject(ctx, blz, bPrm); err == nil {
success = true
@ -59,16 +60,10 @@ func (b *Blobovniczas) Delete(ctx context.Context, prm common.DeletePrm) (res co
return res, err
}
activeCache := make(map[string]struct{})
objectFound := false
err = b.iterateSortedLeaves(ctx, &prm.Address, func(p string) (bool, error) {
dirPath := filepath.Dir(p)
// don't process active blobovnicza of the level twice
_, ok := activeCache[dirPath]
res, err = b.deleteObjectFromLevel(ctx, bPrm, p, !ok)
res, err = b.deleteObjectFromLevel(ctx, bPrm, p)
if err != nil {
if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotRemoveObjectFromLevel,
@ -78,8 +73,6 @@ func (b *Blobovniczas) Delete(ctx context.Context, prm common.DeletePrm) (res co
}
}
activeCache[dirPath] = struct{}{}
if err == nil {
objectFound = true
}
@ -100,57 +93,13 @@ func (b *Blobovniczas) Delete(ctx context.Context, prm common.DeletePrm) (res co
// tries to delete object from particular blobovnicza.
//
// returns no error if object was removed from some blobovnicza of the same level.
func (b *Blobovniczas) deleteObjectFromLevel(ctx context.Context, prm blobovnicza.DeletePrm, blzPath string, tryActive bool) (common.DeleteRes, error) {
lvlPath := filepath.Dir(blzPath)
// try to remove from blobovnicza if it is opened
b.lruMtx.Lock()
v, ok := b.opened.Get(blzPath)
b.lruMtx.Unlock()
if ok {
if res, err := b.deleteObject(ctx, v, prm); err == nil {
return res, err
} else if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotRemoveObjectFromOpenedBlobovnicza,
zap.String("path", blzPath),
zap.String("error", err.Error()),
)
}
}
// therefore the object is possibly placed in a lighter blobovnicza
// next we check in the active level blobobnicza:
// * the active blobovnicza is always opened.
b.activeMtx.RLock()
active, ok := b.active[lvlPath]
b.activeMtx.RUnlock()
if ok && tryActive {
if res, err := b.deleteObject(ctx, active.blz, prm); err == nil {
return res, err
} else if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotRemoveObjectFromActiveBlobovnicza,
zap.String("path", blzPath),
zap.String("error", err.Error()),
)
}
}
// then object is possibly placed in closed blobovnicza
// check if it makes sense to try to open the blob
// (Blobovniczas "after" the active one are empty anyway,
// and it's pointless to open them).
if u64FromHexString(filepath.Base(blzPath)) > active.ind {
return common.DeleteRes{}, logicerr.Wrap(new(apistatus.ObjectNotFound))
}
// open blobovnicza (cached inside)
blz, err := b.openBlobovnicza(blzPath)
func (b *Blobovniczas) deleteObjectFromLevel(ctx context.Context, prm blobovnicza.DeletePrm, blzPath string) (common.DeleteRes, error) {
shBlz := b.openBlobovnicza(blzPath)
blz, err := shBlz.Open()
if err != nil {
return common.DeleteRes{}, err
}
defer shBlz.Close()
return b.deleteObject(ctx, blz, prm)
}

2
pkg/local_object_storage/blobstor/blobovniczatree/errors.go
View file

@ -7,6 +7,8 @@ import (
apistatus "git.frostfs.info/TrueCloudLab/frostfs-sdk-go/client/status"
)
var errClosed = logicerr.New("blobvnicza is closed")
func isErrOutOfRange(err error) bool {
var target *apistatus.ObjectOutOfRange
return errors.As(err, &target)

14
pkg/local_object_storage/blobstor/blobovniczatree/exists.go
View file

@ -3,7 +3,6 @@ package blobovniczatree
import (
"context"
"encoding/hex"
"path/filepath"
"time"
"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
@ -37,26 +36,22 @@ func (b *Blobovniczas) Exists(ctx context.Context, prm common.ExistsPrm) (common
if prm.StorageID != nil {
id := blobovnicza.NewIDFromBytes(prm.StorageID)
blz, err := b.openBlobovnicza(id.String())
shBlz := b.openBlobovnicza(id.String())
blz, err := shBlz.Open()
if err != nil {
return common.ExistsRes{}, err
}
defer shBlz.Close()
exists, err := blz.Exists(ctx, prm.Address)
return common.ExistsRes{Exists: exists}, err
}
activeCache := make(map[string]struct{})
var gPrm blobovnicza.GetPrm
gPrm.SetAddress(prm.Address)
err := b.iterateSortedLeaves(ctx, &prm.Address, func(p string) (bool, error) {
dirPath := filepath.Dir(p)
_, ok := activeCache[dirPath]
_, err := b.getObjectFromLevel(ctx, gPrm, p, !ok)
_, err := b.getObjectFromLevel(ctx, gPrm, p)
if err != nil {
if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotGetObjectFromLevel,
@ -65,7 +60,6 @@ func (b *Blobovniczas) Exists(ctx context.Context, prm common.ExistsPrm) (common
}
}
activeCache[dirPath] = struct{}{}
found = err == nil
return found, nil
})

67
pkg/local_object_storage/blobstor/blobovniczatree/get.go
View file

@ -4,7 +4,6 @@ import (
"context"
"encoding/hex"
"fmt"
"path/filepath"
"time"
"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
@ -48,10 +47,12 @@ func (b *Blobovniczas) Get(ctx context.Context, prm common.GetPrm) (res common.G
if prm.StorageID != nil {
id := blobovnicza.NewIDFromBytes(prm.StorageID)
blz, err := b.openBlobovnicza(id.String())
shBlz := b.openBlobovnicza(id.String())
blz, err := shBlz.Open()
if err != nil {
return res, err
}
defer shBlz.Close()
res, err = b.getObject(ctx, blz, bPrm)
if err == nil {
@ -61,14 +62,8 @@ func (b *Blobovniczas) Get(ctx context.Context, prm common.GetPrm) (res common.G
return res, err
}
activeCache := make(map[string]struct{})
err = b.iterateSortedLeaves(ctx, &prm.Address, func(p string) (bool, error) {
dirPath := filepath.Dir(p)
_, ok := activeCache[dirPath]
res, err = b.getObjectFromLevel(ctx, bPrm, p, !ok)
res, err = b.getObjectFromLevel(ctx, bPrm, p)
if err != nil {
if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotGetObjectFromLevel,
@ -78,8 +73,6 @@ func (b *Blobovniczas) Get(ctx context.Context, prm common.GetPrm) (res common.G
}
}
activeCache[dirPath] = struct{}{}
// abort iterator if found, otherwise process all Blobovniczas
return err == nil, nil
})
@ -98,58 +91,14 @@ func (b *Blobovniczas) Get(ctx context.Context, prm common.GetPrm) (res common.G
// tries to read object from particular blobovnicza.
//
// returns error if object could not be read from any blobovnicza of the same level.
func (b *Blobovniczas) getObjectFromLevel(ctx context.Context, prm blobovnicza.GetPrm, blzPath string, tryActive bool) (common.GetRes, error) {
lvlPath := filepath.Dir(blzPath)
// try to read from blobovnicza if it is opened
b.lruMtx.Lock()
v, ok := b.opened.Get(blzPath)
b.lruMtx.Unlock()
if ok {
if res, err := b.getObject(ctx, v, prm); err == nil {
return res, err
} else if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotReadObjectFromOpenedBlobovnicza,
zap.String("path", blzPath),
zap.String("error", err.Error()),
)
}
}
// therefore the object is possibly placed in a lighter blobovnicza
// next we check in the active level blobobnicza:
// * the freshest objects are probably the most demanded;
// * the active blobovnicza is always opened.
b.activeMtx.RLock()
active, ok := b.active[lvlPath]
b.activeMtx.RUnlock()
if ok && tryActive {
if res, err := b.getObject(ctx, active.blz, prm); err == nil {
return res, err
} else if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotGetObjectFromActiveBlobovnicza,
zap.String("path", blzPath),
zap.String("error", err.Error()),
)
}
}
// then object is possibly placed in closed blobovnicza
// check if it makes sense to try to open the blob
// (Blobovniczas "after" the active one are empty anyway,
// and it's pointless to open them).
if u64FromHexString(filepath.Base(blzPath)) > active.ind {
return common.GetRes{}, logicerr.Wrap(new(apistatus.ObjectNotFound))
}
func (b *Blobovniczas) getObjectFromLevel(ctx context.Context, prm blobovnicza.GetPrm, blzPath string) (common.GetRes, error) {
// open blobovnicza (cached inside)
blz, err := b.openBlobovnicza(blzPath)
shBlz := b.openBlobovnicza(blzPath)
blz, err := shBlz.Open()
if err != nil {
return common.GetRes{}, err
}
defer shBlz.Close()
return b.getObject(ctx, blz, prm)
}

76
pkg/local_object_storage/blobstor/blobovniczatree/get_range.go
View file

@ -4,7 +4,6 @@ import (
"context"
"encoding/hex"
"fmt"
"path/filepath"
"strconv"
"time"
@ -47,10 +46,12 @@ func (b *Blobovniczas) GetRange(ctx context.Context, prm common.GetRangePrm) (re
if prm.StorageID != nil {
id := blobovnicza.NewIDFromBytes(prm.StorageID)
blz, err := b.openBlobovnicza(id.String())
shBlz := b.openBlobovnicza(id.String())
blz, err := shBlz.Open()
if err != nil {
return common.GetRangeRes{}, err
}
defer shBlz.Close()
res, err := b.getObjectRange(ctx, blz, prm)
if err == nil {
@ -60,15 +61,10 @@ func (b *Blobovniczas) GetRange(ctx context.Context, prm common.GetRangePrm) (re
return res, err
}
activeCache := make(map[string]struct{})
objectFound := false
err = b.iterateSortedLeaves(ctx, &prm.Address, func(p string) (bool, error) {
dirPath := filepath.Dir(p)
_, ok := activeCache[dirPath]
res, err = b.getRangeFromLevel(ctx, prm, p, !ok)
res, err = b.getRangeFromLevel(ctx, prm, p)
if err != nil {
outOfBounds := isErrOutOfRange(err)
if !outOfBounds && !client.IsErrObjectNotFound(err) {
@ -82,8 +78,6 @@ func (b *Blobovniczas) GetRange(ctx context.Context, prm common.GetRangePrm) (re
}
}
activeCache[dirPath] = struct{}{}
objectFound = err == nil
// abort iterator if found, otherwise process all Blobovniczas
@ -106,68 +100,14 @@ func (b *Blobovniczas) GetRange(ctx context.Context, prm common.GetRangePrm) (re
// tries to read range of object payload data from particular blobovnicza.
//
// returns error if object could not be read from any blobovnicza of the same level.
func (b *Blobovniczas) getRangeFromLevel(ctx context.Context, prm common.GetRangePrm, blzPath string, tryActive bool) (common.GetRangeRes, error) {
lvlPath := filepath.Dir(blzPath)
// try to read from blobovnicza if it is opened
b.lruMtx.Lock()
v, ok := b.opened.Get(blzPath)
b.lruMtx.Unlock()
if ok {
res, err := b.getObjectRange(ctx, v, prm)
switch {
case err == nil,
isErrOutOfRange(err):
return res, err
default:
if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotReadPayloadRangeFromOpenedBlobovnicza,
zap.String("path", blzPath),
zap.String("error", err.Error()),
)
}
}
}
// therefore the object is possibly placed in a lighter blobovnicza
// next we check in the active level blobobnicza:
// * the freshest objects are probably the most demanded;
// * the active blobovnicza is always opened.
b.activeMtx.RLock()
active, ok := b.active[lvlPath]
b.activeMtx.RUnlock()
if ok && tryActive {
res, err := b.getObjectRange(ctx, active.blz, prm)
switch {
case err == nil,
isErrOutOfRange(err):
return res, err
default:
if !client.IsErrObjectNotFound(err) {
b.log.Debug(logs.BlobovniczatreeCouldNotReadPayloadRangeFromActiveBlobovnicza,
zap.String("path", blzPath),
zap.String("error", err.Error()),
)
}
}
}
// then object is possibly placed in closed blobovnicza
// check if it makes sense to try to open the blob
// (Blobovniczas "after" the active one are empty anyway,
// and it's pointless to open them).
if u64FromHexString(filepath.Base(blzPath)) > active.ind {
return common.GetRangeRes{}, logicerr.Wrap(new(apistatus.ObjectNotFound))
}
func (b *Blobovniczas) getRangeFromLevel(ctx context.Context, prm common.GetRangePrm, blzPath string) (common.GetRangeRes, error) {
// open blobovnicza (cached inside)
blz, err := b.openBlobovnicza(blzPath)
shBlz := b.openBlobovnicza(blzPath)
blz, err := shBlz.Open()
if err != nil {
return common.GetRangeRes{}, err
}
defer shBlz.Close()
return b.getObjectRange(ctx, blz, prm)
}

4
pkg/local_object_storage/blobstor/blobovniczatree/iterate.go
View file

@ -68,13 +68,15 @@ func (b *Blobovniczas) Iterate(ctx context.Context, prm common.IteratePrm) (comm
// iterator over all Blobovniczas in unsorted order. Break on f's error return.
func (b *Blobovniczas) iterateBlobovniczas(ctx context.Context, ignoreErrors bool, f func(string, *blobovnicza.Blobovnicza) error) error {
return b.iterateLeaves(ctx, func(p string) (bool, error) {
blz, err := b.openBlobovnicza(p)
shBlz := b.openBlobovnicza(p)
blz, err := shBlz.Open()
if err != nil {
if ignoreErrors {
return false, nil
}
return false, fmt.Errorf("could not open blobovnicza %s: %w", p, err)
}
defer shBlz.Close()
err = f(p, blz)

243
pkg/local_object_storage/blobstor/blobovniczatree/manager.go Normal file
View file

@ -0,0 +1,243 @@
package blobovniczatree
import (
"fmt"
"path/filepath"
"sync"
"sync/atomic"
"git.frostfs.info/TrueCloudLab/frostfs-node/internal/logs"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobovnicza"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/util/logger"
"go.uber.org/zap"
)
// sharedDB is responsible for opening and closing a file of single blobovnicza.
type sharedDB struct {
guard *sync.RWMutex
blcza *blobovnicza.Blobovnicza
refCount uint32
openDBCounter *openDBCounter
closedFlag *atomic.Bool
options []blobovnicza.Option
path string
readOnly bool
metrics blobovnicza.Metrics
log *logger.Logger
}
func newSharedDB(options []blobovnicza.Option, path string, readOnly bool,
metrics blobovnicza.Metrics, openDBCounter *openDBCounter, closedFlag *atomic.Bool, log *logger.Logger) *sharedDB {
return &sharedDB{
guard: &sync.RWMutex{},
options: options,
path: path,
readOnly: readOnly,
metrics: metrics,
closedFlag: closedFlag,
log: log,
openDBCounter: openDBCounter,
}
}
func (b *sharedDB) Open() (*blobovnicza.Blobovnicza, error) {
if b.closedFlag.Load() {
return nil, errClosed
}
b.guard.Lock()
defer b.guard.Unlock()
if b.refCount > 0 {
b.refCount++
return b.blcza, nil
}
blz := blobovnicza.New(append(b.options,
blobovnicza.WithReadOnly(b.readOnly),
blobovnicza.WithPath(b.path),
blobovnicza.WithMetrics(b.metrics),
)...)
if err := blz.Open(); err != nil {
return nil, fmt.Errorf("could not open blobovnicza %s: %w", b.path, err)
}
if err := blz.Init(); err != nil {
return nil, fmt.Errorf("could not init blobovnicza %s: %w", b.path, err)
}
b.refCount++
b.blcza = blz
b.openDBCounter.Inc()
return blz, nil
}
func (b *sharedDB) Close() {
b.guard.Lock()
defer b.guard.Unlock()
if b.refCount == 0 {
b.log.Error(logs.AttemtToCloseAlreadyClosedBlobovnicza, zap.String("id", b.path))
return
}
if b.refCount == 1 {
b.refCount = 0
if err := b.blcza.Close(); err != nil {
b.log.Error(logs.BlobovniczatreeCouldNotCloseBlobovnicza,
zap.String("id", b.path),
zap.String("error", err.Error()),
)
}
b.blcza = nil
b.openDBCounter.Dec()
return
}
b.refCount--
}
func (b *sharedDB) Path() string {
return b.path
}
// levelDbManager stores pointers of the sharedDB's for the leaf directory of the blobovnicza tree.
type levelDbManager struct {
databases []*sharedDB
}
func newLevelDBManager(width uint64, options []blobovnicza.Option, rootPath string, lvlPath string,
readOnly bool, metrics blobovnicza.Metrics, openDBCounter *openDBCounter, closedFlog *atomic.Bool, log *logger.Logger) *levelDbManager {
result := &levelDbManager{
databases: make([]*sharedDB, width),
}
var idx uint64
for idx = 0; idx < width; idx++ {
result.databases[idx] = newSharedDB(options, filepath.Join(rootPath, lvlPath, u64ToHexString(idx)), readOnly, metrics, openDBCounter, closedFlog, log)
}
return result
}
func (m *levelDbManager) GetByIndex(idx uint64) *sharedDB {
return m.databases[idx]
}
// dbManager manages the opening and closing of blobovnicza instances.
//
// The blobovnicza opens at the first request, closes after the last request.
type dbManager struct {
levelToManager map[string]*levelDbManager
levelToManagerGuard *sync.RWMutex
closedFlag *atomic.Bool
dbCounter *openDBCounter
rootPath string
options []blobovnicza.Option
readOnly bool
metrics blobovnicza.Metrics
leafWidth uint64
log *logger.Logger
}
func newDBManager(rootPath string, options []blobovnicza.Option, leafWidth uint64, readOnly bool, metrics blobovnicza.Metrics, log *logger.Logger) *dbManager {
return &dbManager{
rootPath: rootPath,
options: options,
readOnly: readOnly,
metrics: metrics,
leafWidth: leafWidth,
levelToManager: make(map[string]*levelDbManager),
levelToManagerGuard: &sync.RWMutex{},
log: log,
closedFlag: &atomic.Bool{},
dbCounter: newOpenDBCounter(),
}
}
func (m *dbManager) GetByPath(path string) *sharedDB {
lvlPath := filepath.Dir(path)
curIndex := u64FromHexString(filepath.Base(path))
levelManager := m.getLevelManager(lvlPath)
return levelManager.GetByIndex(curIndex)
}
func (m *dbManager) Open() {
m.closedFlag.Store(false)
}
func (m *dbManager) Close() {
m.closedFlag.Store(true)
m.dbCounter.WaitUntilAllClosed()
}
func (m *dbManager) getLevelManager(lvlPath string) *levelDbManager {
result := m.getLevelManagerIfExists(lvlPath)
if result != nil {
return result
}
return m.getOrCreateLevelManager(lvlPath)
}
func (m *dbManager) getLevelManagerIfExists(lvlPath string) *levelDbManager {
m.levelToManagerGuard.RLock()
defer m.levelToManagerGuard.RUnlock()
return m.levelToManager[lvlPath]
}
func (m *dbManager) getOrCreateLevelManager(lvlPath string) *levelDbManager {
m.levelToManagerGuard.Lock()
defer m.levelToManagerGuard.Unlock()
if result, ok := m.levelToManager[lvlPath]; ok {
return result
}
result := newLevelDBManager(m.leafWidth, m.options, m.rootPath, lvlPath, m.readOnly, m.metrics, m.dbCounter, m.closedFlag, m.log)
m.levelToManager[lvlPath] = result
return result
}
type openDBCounter struct {
cond *sync.Cond
count uint64
}
func newOpenDBCounter() *openDBCounter {
return &openDBCounter{
cond: &sync.Cond{
L: &sync.Mutex{},
},
}
}
func (c *openDBCounter) Inc() {
c.cond.L.Lock()
defer c.cond.L.Unlock()
c.count++
}
func (c *openDBCounter) Dec() {
c.cond.L.Lock()
defer c.cond.L.Unlock()
if c.count > 0 {
c.count--
}
if c.count == 0 {
c.cond.Broadcast()
}
}
func (c *openDBCounter) WaitUntilAllClosed() {
c.cond.L.Lock()
for c.count > 0 {
c.cond.Wait()
}
c.cond.L.Unlock()
}

45
pkg/local_object_storage/blobstor/blobovniczatree/put.go
View file

@ -2,7 +2,6 @@ package blobovniczatree
import (
"context"
"errors"
"path/filepath"
"time"
@ -10,7 +9,6 @@ import (
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobovnicza"
"git.frostfs.info/TrueCloudLab/frostfs-node/pkg/local_object_storage/blobstor/common"
"git.frostfs.info/TrueCloudLab/frostfs-observability/tracing"
"go.etcd.io/bbolt"
"go.opentelemetry.io/otel/attribute"
"go.opentelemetry.io/otel/trace"
"go.uber.org/zap"
@ -76,8 +74,8 @@ type putIterator struct {
PutPrm blobovnicza.PutPrm
}
func (i *putIterator) iterate(ctx context.Context, path string) (bool, error) {
active, err := i.B.getActivated(path)
func (i *putIterator) iterate(ctx context.Context, lvlPath string) (bool, error) {
active, err := i.B.activeDBManager.GetOpenedActiveDBForLevel(lvlPath)
if err != nil {
if !isLogical(err) {
i.B.reportError(logs.BlobovniczatreeCouldNotGetActiveBlobovnicza, err)
@ -89,46 +87,29 @@ func (i *putIterator) iterate(ctx context.Context, path string) (bool, error) {
return false, nil
}
if _, err := active.blz.Put(ctx, i.PutPrm); err != nil {
// Check if blobovnicza is full. We could either receive `blobovnicza.ErrFull` error
// or update active blobovnicza in other thread. In the latter case the database will be closed
// and `updateActive` takes care of not updating the active blobovnicza twice.
if isFull := errors.Is(err, blobovnicza.ErrFull); isFull || errors.Is(err, bbolt.ErrDatabaseNotOpen) {
if isFull {
i.B.log.Debug(logs.BlobovniczatreeBlobovniczaOverflowed,
zap.String("path", filepath.Join(path, u64ToHexString(active.ind))))
}
if active == nil {
i.B.log.Debug(logs.BlobovniczatreeBlobovniczaOverflowed, zap.String("level", lvlPath))
return false, nil
}
defer active.Close()
if err := i.B.updateActive(path, &active.ind); err != nil {
if !isLogical(err) {
i.B.reportError(logs.BlobovniczatreeCouldNotUpdateActiveBlobovnicza, err)
} else {
i.B.log.Debug(logs.BlobovniczatreeCouldNotUpdateActiveBlobovnicza,
zap.String("level", path),
zap.String("error", err.Error()))
}
i.AllFull = false
return false, nil
}
return i.iterate(ctx, path)
}
i.AllFull = false
_, err = active.Blobovnicza().Put(ctx, i.PutPrm)
if err != nil {
if !isLogical(err) {
i.B.reportError(logs.BlobovniczatreeCouldNotPutObjectToActiveBlobovnicza, err)
} else {
i.B.log.Debug(logs.BlobovniczatreeCouldNotPutObjectToActiveBlobovnicza,
zap.String("path", filepath.Join(path, u64ToHexString(active.ind))),
zap.String("path", active.Path()),
zap.String("error", err.Error()))
}
return false, nil
}
path = filepath.Join(path, u64ToHexString(active.ind))
i.ID = blobovnicza.NewIDFromBytes([]byte(path))
idx := u64FromHexString(filepath.Base(active.Path()))
i.ID = blobovnicza.NewIDFromBytes([]byte(filepath.Join(lvlPath, u64ToHexString(idx))))
return true, nil
}