minio/cmd/data-update-tracker_test.go
Klaus Post 073aac3d92
add data update tracking using bloom filter (#9208)
By monitoring PUT/DELETE and heal operations it is possible
to track changed paths and keep a bloom filter for this data. 

This can help prioritize paths to scan. The bloom filter can identify
paths that have not changed, and the few collisions will only result
in a marginal extra workload. This can be implemented on either a
bucket+(1 prefix level) with reasonable performance.

The bloom filter is set to have a false positive rate at 1% at 1M 
entries. A bloom table of this size is about ~2500 bytes when serialized.

To not force a full scan of all paths that have changed cycle bloom
filters would need to be kept, so we guarantee that dirty paths have
been scanned within cycle runs. Until cycle bloom filters have been
collected all paths are considered dirty.
2020-04-27 10:06:21 -07:00

263 lines
6.6 KiB
Go

/*
* MinIO Cloud Storage, (C) 2020 MinIO, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package cmd
import (
"bytes"
"context"
"fmt"
"io/ioutil"
"math/rand"
"os"
"path"
"path/filepath"
"sync"
"testing"
"github.com/minio/minio/cmd/logger"
"github.com/minio/minio/cmd/logger/message/log"
)
type testLoggerI interface {
Helper()
Log(args ...interface{})
}
type testingLogger struct {
mu sync.Mutex
t testLoggerI
}
func (t *testingLogger) Send(entry interface{}, errKind string) error {
t.mu.Lock()
defer t.mu.Unlock()
if t.t == nil {
return nil
}
e, ok := entry.(log.Entry)
if !ok {
return fmt.Errorf("unexpected log entry structure %#v", entry)
}
t.t.Helper()
t.t.Log(e.Level, ":", errKind, e.Message)
return nil
}
func addTestingLogging(t testLoggerI) func() {
tl := &testingLogger{t: t}
logger.AddTarget(tl)
return func() {
tl.mu.Lock()
defer tl.mu.Unlock()
tl.t = nil
}
}
func TestDataUpdateTracker(t *testing.T) {
dut := newDataUpdateTracker()
// Change some defaults.
dut.debug = testing.Verbose()
dut.input = make(chan string)
dut.save = make(chan struct{})
defer addTestingLogging(t)()
dut.Current.bf = dut.newBloomFilter()
tmpDir, err := ioutil.TempDir("", "TestDataUpdateTracker")
if err != nil {
t.Fatal(err)
}
err = os.MkdirAll(filepath.Dir(filepath.Join(tmpDir, dataUpdateTrackerFilename)), os.ModePerm)
if err != nil {
t.Fatal(err)
}
defer os.RemoveAll(tmpDir)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
dut.start(ctx, tmpDir)
var tests = []struct {
in string
check []string // if not empty, check against these instead.
exist bool
}{
{
in: "bucket/directory/file.txt",
check: []string{"bucket", "bucket/", "/bucket", "bucket/directory", "bucket/directory/", "bucket/directory/file.txt", "/bucket/directory/file.txt"},
exist: true,
},
{
// System bucket
in: ".minio.sys/ignoreme/pls",
exist: false,
},
{
// Not a valid bucket
in: "./bucket/okfile.txt",
check: []string{"./bucket/okfile.txt", "/bucket/okfile.txt", "bucket/okfile.txt"},
exist: false,
},
{
// Not a valid bucket
in: "æ/okfile.txt",
check: []string{"æ/okfile.txt", "æ/okfile.txt", "æ"},
exist: false,
},
{
in: "/bucket2/okfile2.txt",
check: []string{"./bucket2/okfile2.txt", "/bucket2/okfile2.txt", "bucket2/okfile2.txt", "bucket2"},
exist: true,
},
{
in: "/bucket3/prefix/okfile2.txt",
check: []string{"./bucket3/prefix/okfile2.txt", "/bucket3/prefix/okfile2.txt", "bucket3/prefix/okfile2.txt", "bucket3/prefix", "bucket3"},
exist: true,
},
}
for _, tt := range tests {
t.Run(tt.in, func(t *testing.T) {
dut.input <- tt.in
dut.input <- "" // Sending empty string ensures the previous is added to filter.
dut.mu.Lock()
defer dut.mu.Unlock()
if len(tt.check) == 0 {
got := dut.Current.bf.containsDir(tt.in)
if got != tt.exist {
// For unlimited tests this could lead to false positives,
// but it should be deterministic.
t.Errorf("entry %q, got: %v, want %v", tt.in, got, tt.exist)
}
return
}
for _, check := range tt.check {
got := dut.Current.bf.containsDir(check)
if got != tt.exist {
// For unlimited tests this could lead to false positives,
// but it should be deterministic.
t.Errorf("entry %q, check: %q, got: %v, want %v", tt.in, check, got, tt.exist)
}
continue
}
})
}
// Cycle to history
_, err = dut.cycleFilter(ctx, 1, 2)
if err != nil {
t.Fatal(err)
}
dut.input <- "cycle2/file.txt"
dut.input <- "" // Sending empty string ensures the previous is added to filter.
tests = append(tests, struct {
in string
check []string
exist bool
}{in: "cycle2/file.txt", exist: true})
// Shut down
cancel()
<-dut.saveExited
if dut.current() != 2 {
t.Fatal("wrong current idx after save. want 2, got:", dut.current())
}
ctx, cancel = context.WithCancel(context.Background())
defer cancel()
// Reload...
dut = newDataUpdateTracker()
dut.start(ctx, tmpDir)
if dut.current() != 2 {
t.Fatal("current idx after load not preserved. want 2, got:", dut.current())
}
bfr2, err := dut.cycleFilter(ctx, 1, 3)
if err != nil {
t.Fatal(err)
}
if !bfr2.Complete {
t.Fatal("Wanted complete, didn't get it")
}
if bfr2.CurrentIdx != 3 {
t.Fatal("wanted index 3, got", bfr2.CurrentIdx)
}
if bfr2.OldestIdx != 1 {
t.Fatal("wanted oldest index 3, got", bfr2.OldestIdx)
}
// Rerun test with returned bfr2
bf := dut.newBloomFilter()
_, err = bf.ReadFrom(bytes.NewBuffer(bfr2.Filter))
if err != nil {
t.Fatal(err)
}
for _, tt := range tests {
t.Run(tt.in+"-reloaded", func(t *testing.T) {
if len(tt.check) == 0 {
got := bf.containsDir(tt.in)
if got != tt.exist {
// For unlimited tests this could lead to false positives,
// but it should be deterministic.
t.Errorf("entry %q, got: %v, want %v", tt.in, got, tt.exist)
}
return
}
for _, check := range tt.check {
got := bf.containsDir(check)
if got != tt.exist {
// For unlimited tests this could lead to false positives,
// but it should be deterministic.
t.Errorf("entry %q, check: %q, got: %v, want %v", tt.in, check, got, tt.exist)
}
continue
}
})
}
}
func BenchmarkDataUpdateTracker(b *testing.B) {
dut := newDataUpdateTracker()
// Change some defaults.
dut.debug = false
dut.input = make(chan string)
dut.save = make(chan struct{})
defer addTestingLogging(b)()
dut.Current.bf = dut.newBloomFilter()
// We do this unbuffered. This will very significantly reduce throughput, so this is a worst case.
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
go dut.startCollector(ctx)
input := make([]string, 1000)
rng := rand.New(rand.NewSource(0xabad1dea))
tmp := []string{"bucket", "aprefix", "nextprefixlevel", "maybeobjname", "evendeeper", "ok-one-morelevel", "final.object"}
for i := range input {
tmp := tmp[:1+rng.Intn(cap(tmp)-1)]
input[i] = path.Join(tmp...)
}
b.SetBytes(1)
b.ResetTimer()
b.ReportAllocs()
for i := 0; i < b.N; i++ {
dut.input <- input[rng.Intn(len(input))]
}
}