Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

statistics: add gc in hot peer cache #8702

Merged
merged 7 commits into from
Oct 28, 2024
Merged

statistics: add gc in hot peer cache #8702

Show file tree
Hide file tree
Changes from 4 commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
d8b8091
statistics: add gc in hot peer cache
lhy1024 Oct 14, 2024
9266c24
fix test
lhy1024 Oct 14, 2024
fe4ebc4
fix test
lhy1024 Oct 14, 2024
f2d03a9
fix test
lhy1024 Oct 15, 2024
9d00e7a
Merge branch 'master' of github.com:tikv/pd into hot-gc
lhy1024 Oct 23, 2024
9331644
Merge branch 'master' of github.com:tikv/pd into hot-gc
lhy1024 Oct 25, 2024
9481fb7
address comments
lhy1024 Oct 25, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion pkg/mcs/scheduling/server/cluster.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -93,7 +93,7 @@ func NewCluster(parentCtx context.Context, persistConfig *config.PersistConfig,
ruleManager: ruleManager,
labelerManager: labelerManager,
persistConfig: persistConfig,
hotStat: statistics.NewHotStat(ctx),
hotStat: statistics.NewHotStat(ctx, basicCluster),
labelStats: statistics.NewLabelStatistics(),
regionStats: statistics.NewRegionStatistics(basicCluster, persistConfig, ruleManager),
storage: storage,
Expand Down
5 changes: 3 additions & 2 deletions pkg/mock/mockcluster/mockcluster.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -63,11 +63,12 @@ type Cluster struct {

// NewCluster creates a new Cluster
func NewCluster(ctx context.Context, opts *config.PersistOptions) *Cluster {
bc := core.NewBasicCluster()
c := &Cluster{
ctx: ctx,
BasicCluster: core.NewBasicCluster(),
BasicCluster: bc,
IDAllocator: mockid.NewIDAllocator(),
HotStat: statistics.NewHotStat(ctx),
HotStat: statistics.NewHotStat(ctx, bc),
HotBucketCache: buckets.NewBucketsCache(ctx),
PersistOptions: opts,
pendingProcessedRegions: map[uint64]struct{}{},
Expand Down
18 changes: 18 additions & 0 deletions pkg/schedule/schedulers/hot_region_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1614,6 +1614,9 @@ func TestHotCacheUpdateCache(t *testing.T) {
re := require.New(t)
cancel, _, tc, _ := prepareSchedulersTest()
defer cancel()
for i := 0; i < 3; i++ {
tc.PutStore(core.NewStoreInfo(&metapb.Store{Id: uint64(i + 1)}))
}

// For read flow
addRegionInfo(tc, utils.Read, []testRegionInfo{
Expand Down Expand Up @@ -1680,6 +1683,9 @@ func TestHotCacheKeyThresholds(t *testing.T) {
{ // only a few regions
cancel, _, tc, _ := prepareSchedulersTest()
defer cancel()
for i := 0; i < 6; i++ {
tc.PutStore(core.NewStoreInfo(&metapb.Store{Id: uint64(i + 1)}))
}
addRegionInfo(tc, utils.Read, []testRegionInfo{
{1, []uint64{1, 2, 3}, 0, 1, 0},
{2, []uint64{1, 2, 3}, 0, 1 * units.KiB, 0},
Expand All @@ -1698,6 +1704,9 @@ func TestHotCacheKeyThresholds(t *testing.T) {
{ // many regions
cancel, _, tc, _ := prepareSchedulersTest()
defer cancel()
for i := 0; i < 3; i++ {
tc.PutStore(core.NewStoreInfo(&metapb.Store{Id: uint64(i + 1)}))
}
regions := []testRegionInfo{}
for i := 1; i <= 1000; i += 2 {
regions = append(regions,
Expand Down Expand Up @@ -1751,6 +1760,9 @@ func TestHotCacheByteAndKey(t *testing.T) {
re := require.New(t)
cancel, _, tc, _ := prepareSchedulersTest()
defer cancel()
for i := 0; i < 3; i++ {
tc.PutStore(core.NewStoreInfo(&metapb.Store{Id: uint64(i + 1)}))
}
statistics.ThresholdsUpdateInterval = 0
defer func() {
statistics.ThresholdsUpdateInterval = 8 * time.Second
Expand Down Expand Up @@ -1877,6 +1889,9 @@ func TestHotCacheCheckRegionFlow(t *testing.T) {
func checkHotCacheCheckRegionFlow(re *require.Assertions, testCase testHotCacheCheckRegionFlowCase, enablePlacementRules bool) {
cancel, _, tc, oc := prepareSchedulersTest()
defer cancel()
for i := 0; i < 3; i++ {
tc.PutStore(core.NewStoreInfo(&metapb.Store{Id: uint64(i + 1)}))
}
tc.SetClusterVersion(versioninfo.MinSupportedVersion(versioninfo.Version4_0))
tc.SetEnablePlacementRules(enablePlacementRules)
labels := []string{"zone", "host"}
Expand Down Expand Up @@ -1953,6 +1968,9 @@ func TestHotCacheCheckRegionFlowWithDifferentThreshold(t *testing.T) {
func checkHotCacheCheckRegionFlowWithDifferentThreshold(re *require.Assertions, enablePlacementRules bool) {
cancel, _, tc, _ := prepareSchedulersTest()
defer cancel()
for i := 0; i < 3; i++ {
tc.PutStore(core.NewStoreInfo(&metapb.Store{Id: uint64(i + 1)}))
}
tc.SetClusterVersion(versioninfo.MinSupportedVersion(versioninfo.Version4_0))
tc.SetEnablePlacementRules(enablePlacementRules)
labels := []string{"zone", "host"}
Expand Down
8 changes: 4 additions & 4 deletions pkg/statistics/hot_cache.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -39,11 +39,11 @@ type HotCache struct {
}

// NewHotCache creates a new hot spot cache.
func NewHotCache(ctx context.Context) *HotCache {
func NewHotCache(ctx context.Context, cluster *core.BasicCluster) *HotCache {
w := &HotCache{
ctx: ctx,
writeCache: NewHotPeerCache(ctx, utils.Write),
readCache: NewHotPeerCache(ctx, utils.Read),
writeCache: NewHotPeerCache(ctx, cluster, utils.Write),
readCache: NewHotPeerCache(ctx, cluster, utils.Read),
}
go w.updateItems(w.readCache.taskQueue, w.runReadTask)
go w.updateItems(w.writeCache.taskQueue, w.runWriteTask)
Expand Down Expand Up @@ -80,7 +80,7 @@ func (w *HotCache) CheckReadAsync(task func(cache *HotPeerCache)) bool {
func (w *HotCache) RegionStats(kind utils.RWType, minHotDegree int) map[uint64][]*HotPeerStat {
ret := make(chan map[uint64][]*HotPeerStat, 1)
collectRegionStatsTask := func(cache *HotPeerCache) {
ret <- cache.RegionStats(minHotDegree)
ret <- cache.PeerStats(minHotDegree)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Why change the name, it's not consistent with other places.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Okay, I'll keep the TODO and change them in another PR uniform.

}
var succ bool
switch kind {
Expand Down
6 changes: 4 additions & 2 deletions pkg/statistics/hot_cache_test.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,7 @@ import (
"testing"

"github.com/stretchr/testify/require"
"github.com/tikv/pd/pkg/core"
"github.com/tikv/pd/pkg/statistics/utils"
)

Expand All @@ -27,8 +28,9 @@ func TestIsHot(t *testing.T) {
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
for i := utils.RWType(0); i < utils.RWTypeLen; i++ {
cache := NewHotCache(ctx)
region := buildRegion(i, 3, 60)
cluster := core.NewBasicCluster()
cache := NewHotCache(ctx, cluster)
region := buildRegion(cluster, i, 3, 60)
stats := cache.CheckReadPeerSync(region, region.GetPeers(), []float64{100000000, 1000, 1000}, 60)
cache.Update(stats[0], i)
for i := 0; i < 100; i++ {
Expand Down
42 changes: 37 additions & 5 deletions pkg/statistics/hot_peer_cache.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,20 +60,22 @@ type thresholds struct {
// HotPeerCache saves the hot peer's statistics.
type HotPeerCache struct {
kind utils.RWType
cluster *core.BasicCluster
peersOfStore map[uint64]*utils.TopN // storeID -> hot peers
storesOfRegion map[uint64]map[uint64]struct{} // regionID -> storeIDs
regionsOfStore map[uint64]map[uint64]struct{} // storeID -> regionIDs
topNTTL time.Duration
taskQueue *chanx.UnboundedChan[func(*HotPeerCache)]
thresholdsOfStore map[uint64]*thresholds // storeID -> thresholds
metrics map[uint64][utils.ActionTypeLen]prometheus.Gauge // storeID -> metrics
// TODO: consider to remove store info when store is offline.
lastGCTime time.Time
}

// NewHotPeerCache creates a HotPeerCache
func NewHotPeerCache(ctx context.Context, kind utils.RWType) *HotPeerCache {
func NewHotPeerCache(ctx context.Context, cluster *core.BasicCluster, kind utils.RWType) *HotPeerCache {
return &HotPeerCache{
kind: kind,
cluster: cluster,
peersOfStore: make(map[uint64]*utils.TopN),
storesOfRegion: make(map[uint64]map[uint64]struct{}),
regionsOfStore: make(map[uint64]map[uint64]struct{}),
Expand All @@ -84,9 +86,8 @@ func NewHotPeerCache(ctx context.Context, kind utils.RWType) *HotPeerCache {
}
}

// TODO: rename RegionStats as PeerStats
// RegionStats returns hot items
func (f *HotPeerCache) RegionStats(minHotDegree int) map[uint64][]*HotPeerStat {
// PeerStats returns hot items
func (f *HotPeerCache) PeerStats(minHotDegree int) map[uint64][]*HotPeerStat {
res := make(map[uint64][]*HotPeerStat)
defaultAntiCount := f.kind.DefaultAntiCount()
for storeID, peers := range f.peersOfStore {
Expand Down Expand Up @@ -115,6 +116,7 @@ func (f *HotPeerCache) UpdateStat(item *HotPeerStat) {
return
}
f.incMetrics(item.actionType, item.StoreID)
f.removeExpiredItems()
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will it be called every time we receive a region heartbeat?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In previous, removeItem or putItem will call tn.maintain() every time. In this PR, it will call only after topNTTL

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

L554-L556

}

func (f *HotPeerCache) incMetrics(action utils.ActionType, storeID uint64) {
Expand Down Expand Up @@ -548,6 +550,36 @@ func (f *HotPeerCache) removeItem(item *HotPeerStat) {
}
}

func (f *HotPeerCache) removeExpiredItems() {
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

How about renaming to gc()?

if time.Since(f.lastGCTime) < f.topNTTL {
return
}
f.lastGCTime = time.Now()
// remove tombstone stores
stores := make(map[uint64]struct{})
for _, storeID := range f.cluster.GetStores() {
stores[storeID.GetID()] = struct{}{}
}
for storeID := range f.peersOfStore {
if _, ok := stores[storeID]; !ok {
delete(f.peersOfStore, storeID)
delete(f.regionsOfStore, storeID)
delete(f.thresholdsOfStore, storeID)
delete(f.metrics, storeID)
}
}
// remove expired items
for _, peers := range f.peersOfStore {
regions := peers.RemoveExpired()
for _, regionID := range regions {
delete(f.storesOfRegion, regionID)
for storeID := range f.regionsOfStore {
delete(f.regionsOfStore[storeID], regionID)
}
}
}
}

// removeAllItem removes all items of the cache.
// It is used for test.
func (f *HotPeerCache) removeAllItem() {
Expand Down
Loading