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.

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

statstics: correctly handle error when merging global stats #47770

Merged
merged 5 commits into from
Oct 19, 2023
Merged
Show file tree
Hide file tree
Changes from 3 commits
Commits
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/statistics/handle/globalstats/BUILD.bazel
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,7 @@ go_test(
"topn_bench_test.go",
],
flaky = True,
shard_count = 14,
shard_count = 16,
deps = [
":globalstats",
"//pkg/config",
Expand Down
79 changes: 51 additions & 28 deletions pkg/statistics/handle/globalstats/global_stats_async.go
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,8 @@ import (
"github.com/pingcap/tidb/pkg/statistics/handle/storage"
"github.com/pingcap/tidb/pkg/statistics/handle/util"
"github.com/pingcap/tidb/pkg/types"
"github.com/pingcap/tidb/pkg/util/logutil"
"go.uber.org/zap"
"golang.org/x/sync/errgroup"
)

Expand Down Expand Up @@ -79,8 +81,11 @@ type AsyncMergePartitionStats2GlobalStats struct {
PartitionDefinition map[int64]model.PartitionDefinition
tableInfo map[int64]*model.TableInfo
// key is partition id and histID
skipPartition map[skipItem]struct{}
exitWhenErrChan chan struct{}
skipPartition map[skipItem]struct{}
// ioWorker meet error, it will close this channel to notify cpuWorker.
ioWorkerExitWhenErrChan chan struct{}
// cpuWorker exit, it will close this channel to notify ioWorker.
cpuWorkerExitChan chan struct{}
globalTableInfo *model.TableInfo
histIDs []int64
globalStatsNDV []int64
Expand All @@ -97,20 +102,21 @@ func NewAsyncMergePartitionStats2GlobalStats(
is infoschema.InfoSchema) (*AsyncMergePartitionStats2GlobalStats, error) {
partitionNum := len(globalTableInfo.Partition.Definitions)
return &AsyncMergePartitionStats2GlobalStats{
statsHandle: statsHandle,
cmsketch: make(chan mergeItem[*statistics.CMSketch], 5),
fmsketch: make(chan mergeItem[*statistics.FMSketch], 5),
histogramAndTopn: make(chan mergeItem[*StatsWrapper]),
PartitionDefinition: make(map[int64]model.PartitionDefinition),
tableInfo: make(map[int64]*model.TableInfo),
partitionIDs: make([]int64, 0, partitionNum),
exitWhenErrChan: make(chan struct{}),
skipPartition: make(map[skipItem]struct{}),
allPartitionStats: make(map[int64]*statistics.Table),
globalTableInfo: globalTableInfo,
histIDs: histIDs,
is: is,
partitionNum: partitionNum,
statsHandle: statsHandle,
cmsketch: make(chan mergeItem[*statistics.CMSketch], 5),
fmsketch: make(chan mergeItem[*statistics.FMSketch], 5),
histogramAndTopn: make(chan mergeItem[*StatsWrapper]),
PartitionDefinition: make(map[int64]model.PartitionDefinition),
tableInfo: make(map[int64]*model.TableInfo),
partitionIDs: make([]int64, 0, partitionNum),
ioWorkerExitWhenErrChan: make(chan struct{}),
cpuWorkerExitChan: make(chan struct{}),
skipPartition: make(map[skipItem]struct{}),
allPartitionStats: make(map[int64]*statistics.Table),
globalTableInfo: globalTableInfo,
histIDs: histIDs,
is: is,
partitionNum: partitionNum,
}, nil
}

Expand Down Expand Up @@ -218,25 +224,26 @@ func (a *AsyncMergePartitionStats2GlobalStats) dealErrPartitionColumnStatsMissin
func (a *AsyncMergePartitionStats2GlobalStats) ioWorker(sctx sessionctx.Context, isIndex bool) (err error) {
defer func() {
if r := recover(); r != nil {
close(a.exitWhenErrChan)
logutil.BgLogger().Warn("ioWorker panic", zap.Stack("stack"), zap.Any("error", r))
close(a.ioWorkerExitWhenErrChan)
err = errors.New(fmt.Sprint(r))
}
}()
err = a.loadFmsketch(sctx, isIndex)
if err != nil {
close(a.exitWhenErrChan)
close(a.ioWorkerExitWhenErrChan)
return err
}
close(a.fmsketch)
err = a.loadCMsketch(sctx, isIndex)
if err != nil {
close(a.exitWhenErrChan)
close(a.ioWorkerExitWhenErrChan)
return err
}
close(a.cmsketch)
err = a.loadHistogramAndTopN(sctx, a.globalTableInfo, isIndex)
if err != nil {
close(a.exitWhenErrChan)
close(a.ioWorkerExitWhenErrChan)
return err
}
close(a.histogramAndTopn)
Expand All @@ -246,13 +253,14 @@ func (a *AsyncMergePartitionStats2GlobalStats) ioWorker(sctx sessionctx.Context,
func (a *AsyncMergePartitionStats2GlobalStats) cpuWorker(stmtCtx *stmtctx.StatementContext, sctx sessionctx.Context, opts map[ast.AnalyzeOptionType]uint64, isIndex bool, tz *time.Location, analyzeVersion int) (err error) {
defer func() {
if r := recover(); r != nil {
close(a.exitWhenErrChan)
logutil.BgLogger().Warn("cpuWorker panic", zap.Stack("stack"), zap.Any("error", r))
err = errors.New(fmt.Sprint(r))
}
close(a.cpuWorkerExitChan)
}()
a.dealFMSketch()
select {
case <-a.exitWhenErrChan:
case <-a.ioWorkerExitWhenErrChan:
return nil
default:
for i := 0; i < a.globalStats.Num; i++ {
Expand All @@ -267,10 +275,12 @@ func (a *AsyncMergePartitionStats2GlobalStats) cpuWorker(stmtCtx *stmtctx.Statem
}
err = a.dealCMSketch()
if err != nil {
logutil.BgLogger().Warn("dealCMSketch failed", zap.Error(err))
return err
}
err = a.dealHistogramAndTopN(stmtCtx, sctx, opts, isIndex, tz, analyzeVersion)
if err != nil {
logutil.BgLogger().Warn("dealHistogramAndTopN failed", zap.Error(err))
return err
}
return nil
Expand Down Expand Up @@ -337,7 +347,8 @@ func (a *AsyncMergePartitionStats2GlobalStats) loadFmsketch(sctx sessionctx.Cont
case a.fmsketch <- mergeItem[*statistics.FMSketch]{
fmsketch, i,
}:
case <-a.exitWhenErrChan:
case <-a.cpuWorkerExitChan:
logutil.BgLogger().Warn("ioWorker detects CPUWorker has exited")
return nil
}
}
Expand Down Expand Up @@ -367,7 +378,8 @@ func (a *AsyncMergePartitionStats2GlobalStats) loadCMsketch(sctx sessionctx.Cont
case a.cmsketch <- mergeItem[*statistics.CMSketch]{
cmsketch, i,
}:
case <-a.exitWhenErrChan:
case <-a.cpuWorkerExitChan:
logutil.BgLogger().Warn("ioWorker detects CPUWorker has exited")
return nil
}
}
Expand Down Expand Up @@ -402,7 +414,8 @@ func (a *AsyncMergePartitionStats2GlobalStats) loadHistogramAndTopN(sctx session
case a.histogramAndTopn <- mergeItem[*StatsWrapper]{
NewStatsWrapper(hists, topn), i,
}:
case <-a.exitWhenErrChan:
case <-a.cpuWorkerExitChan:
logutil.BgLogger().Warn("ioWorker detects CPUWorker has exited")
return nil
}
}
Expand All @@ -422,13 +435,18 @@ func (a *AsyncMergePartitionStats2GlobalStats) dealFMSketch() {
} else {
a.globalStats.Fms[fms.idx].MergeFMSketch(fms.item)
}
case <-a.exitWhenErrChan:
case <-a.ioWorkerExitWhenErrChan:
return
}
}
}

func (a *AsyncMergePartitionStats2GlobalStats) dealCMSketch() error {
failpoint.Inject("dealCMSketchErr", func(val failpoint.Value) {
if val, _ := val.(bool); val {
failpoint.Return(errors.New("dealCMSketch returned error"))
}
})
for {
select {
case cms, ok := <-a.cmsketch:
Expand All @@ -443,13 +461,18 @@ func (a *AsyncMergePartitionStats2GlobalStats) dealCMSketch() error {
return err
}
}
case <-a.exitWhenErrChan:
case <-a.ioWorkerExitWhenErrChan:
return nil
}
}
}

func (a *AsyncMergePartitionStats2GlobalStats) dealHistogramAndTopN(stmtCtx *stmtctx.StatementContext, sctx sessionctx.Context, opts map[ast.AnalyzeOptionType]uint64, isIndex bool, tz *time.Location, analyzeVersion int) (err error) {
failpoint.Inject("dealHistogramAndTopNErr", func(val failpoint.Value) {
if val, _ := val.(bool); val {
failpoint.Return(errors.New("dealHistogramAndTopNErr returned error"))
}
})
for {
select {
case item, ok := <-a.histogramAndTopn:
Expand Down Expand Up @@ -478,7 +501,7 @@ func (a *AsyncMergePartitionStats2GlobalStats) dealHistogramAndTopN(stmtCtx *stm
a.globalStats.Hg[item.idx].Buckets[j].NDV = 0
}
a.globalStats.Hg[item.idx].NDV = a.globalStatsNDV[item.idx]
case <-a.exitWhenErrChan:
case <-a.ioWorkerExitWhenErrChan:
return nil
}
}
Expand Down
18 changes: 18 additions & 0 deletions pkg/statistics/handle/globalstats/globalstats_test.go
Original file line number Diff line number Diff line change
Expand Up @@ -76,6 +76,24 @@ func TestGlobalStatsPanicInIOWorker(t *testing.T) {
simpleTest(t)
}

func TestGlobalStatsWithCMSketchErr(t *testing.T) {
qw4990 marked this conversation as resolved.
Show resolved Hide resolved
fpName := "github.com/pingcap/tidb/pkg/statistics/handle/globalstats/dealCMSketchErr"
require.NoError(t, failpoint.Enable(fpName, `return(true)`))
defer func() {
require.NoError(t, failpoint.Disable(fpName))
}()
simpleTest(t)
}

func TestGlobalStatsWithHistogramAndTopNErr(t *testing.T) {
fpName := "github.com/pingcap/tidb/pkg/statistics/handle/globalstats/dealHistogramAndTopNErr"
require.NoError(t, failpoint.Enable(fpName, `return(true)`))
defer func() {
require.NoError(t, failpoint.Disable(fpName))
}()
simpleTest(t)
}

func TestGlobalStatsPanicInCPUWorker(t *testing.T) {
fpName := "github.com/pingcap/tidb/pkg/statistics/handle/globalstats/PanicInCPUWorker"
require.NoError(t, failpoint.Enable(fpName, "panic(\"inject panic\")"))
Expand Down