add support for tuning healing to make healing more aggressive (#11003)

supports `mc admin config set <alias> heal sleep=100ms` to
enable more aggressive healing under certain times.

also optimize some areas that were doing extra checks than
necessary when bitrotscan was enabled, avoid double sleeps
make healing more predictable.

fixes #10497
master
Harshavardhana 4 years ago committed by GitHub
parent fe11e9047d
commit 96c0ce1f0c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
  1. 9
      cmd/admin-heal-ops.go
  2. 26
      cmd/background-heal-ops.go
  3. 2
      cmd/background-newdisks-heal-ops.go
  4. 51
      cmd/config/heal/heal.go
  5. 9
      cmd/data-crawler.go
  6. 10
      cmd/fs-v1.go
  7. 5
      cmd/global-heal.go
  8. 4
      cmd/http/server.go
  9. 60
      cmd/xl-storage.go

@ -657,6 +657,8 @@ func (h *healSequence) healSequenceStart(objAPI ObjectLayer) {
}
func (h *healSequence) queueHealTask(source healSource, healType madmin.HealItemType) error {
opts := globalHealConfig
// Send heal request
task := healTask{
bucket: source.bucket,
@ -667,8 +669,15 @@ func (h *healSequence) queueHealTask(source healSource, healType madmin.HealItem
}
if source.opts != nil {
task.opts = *source.opts
} else {
if opts.Bitrot {
task.opts.ScanMode = madmin.HealDeepScan
}
}
// Wait and proceed if there are active requests
waitForLowHTTPReq(opts.IOCount, opts.Sleep)
h.mutex.Lock()
h.scannedItemsMap[healType]++
h.lastHealActivity = UTCNow()

@ -20,6 +20,7 @@ import (
"context"
"time"
"github.com/minio/minio/cmd/logger"
"github.com/minio/minio/pkg/madmin"
)
@ -53,20 +54,28 @@ func (h *healRoutine) queueHealTask(task healTask) {
h.tasks <- task
}
func waitForLowHTTPReq(tolerance int32, maxWait time.Duration) {
const wait = 10 * time.Millisecond
waitCount := maxWait / wait
func waitForLowHTTPReq(tolerance int, maxWait time.Duration) {
// At max 10 attempts to wait with 100 millisecond interval before proceeding
waitCount := 10
waitTick := 100 * time.Millisecond
// Bucket notification and http trace are not costly, it is okay to ignore them
// while counting the number of concurrent connections
tolerance += int32(globalHTTPListen.NumSubscribers() + globalHTTPTrace.NumSubscribers())
toleranceFn := func() int {
return tolerance + globalHTTPListen.NumSubscribers() + globalHTTPTrace.NumSubscribers()
}
if httpServer := newHTTPServerFn(); httpServer != nil {
// Any requests in progress, delay the heal.
for (httpServer.GetRequestCount() >= tolerance) &&
waitCount > 0 {
for httpServer.GetRequestCount() >= toleranceFn() {
time.Sleep(waitTick)
waitCount--
time.Sleep(wait)
if waitCount == 0 {
if intDataUpdateTracker.debug {
logger.Info("waitForLowHTTPReq: waited %d times, resuming", waitCount)
}
break
}
}
}
}
@ -80,9 +89,6 @@ func (h *healRoutine) run(ctx context.Context, objAPI ObjectLayer) {
break
}
// Wait and proceed if there are active requests
waitForLowHTTPReq(int32(globalEndpoints.NEndpoints()), time.Second)
var res madmin.HealResultItem
var err error
switch {

@ -124,8 +124,6 @@ wait:
case <-ctx.Done():
return
case <-time.After(defaultMonitorNewDiskInterval):
waitForLowHTTPReq(int32(globalEndpoints.NEndpoints()), time.Second)
var erasureSetInZoneDisksToHeal []map[int][]StorageAPI
healDisks := globalBackgroundHealState.getHealLocalDisks()

@ -17,20 +17,32 @@
package heal
import (
"errors"
"fmt"
"strconv"
"time"
"github.com/minio/minio/cmd/config"
"github.com/minio/minio/pkg/env"
)
// Compression environment variables
const (
Bitrot = "bitrotscan"
Bitrot = "bitrotscan"
Sleep = "max_sleep"
IOCount = "max_io"
EnvBitrot = "MINIO_HEAL_BITROTSCAN"
EnvSleep = "MINIO_HEAL_MAX_SLEEP"
EnvIOCount = "MINIO_HEAL_MAX_IO"
)
// Config represents the heal settings.
type Config struct {
// Bitrot will perform bitrot scan on local disk when checking objects.
Bitrot bool `json:"bitrotscan"`
// maximum sleep duration between objects to slow down heal operation.
Sleep time.Duration `json:"sleep"`
IOCount int `json:"iocount"`
}
var (
@ -40,6 +52,14 @@ var (
Key: Bitrot,
Value: config.EnableOff,
},
config.KV{
Key: Sleep,
Value: "1s",
},
config.KV{
Key: IOCount,
Value: "10",
},
}
// Help provides help for config values
@ -50,6 +70,18 @@ var (
Optional: true,
Type: "on|off",
},
config.HelpKV{
Key: Sleep,
Description: `maximum sleep duration between objects to slow down heal operation. eg. 2s`,
Optional: true,
Type: "duration",
},
config.HelpKV{
Key: IOCount,
Description: `maximum IO requests allowed between objects to slow down heal operation. eg. 3`,
Optional: true,
Type: "int",
},
}
)
@ -58,10 +90,17 @@ func LookupConfig(kvs config.KVS) (cfg Config, err error) {
if err = config.CheckValidKeys(config.HealSubSys, kvs, DefaultKVS); err != nil {
return cfg, err
}
bitrot := kvs.Get(Bitrot)
if bitrot != config.EnableOn && bitrot != config.EnableOff {
return cfg, errors.New(Bitrot + ": must be 'on' or 'off'")
cfg.Bitrot, err = config.ParseBool(env.Get(EnvBitrot, kvs.Get(Bitrot)))
if err != nil {
return cfg, fmt.Errorf("'heal:bitrotscan' value invalid: %w", err)
}
cfg.Sleep, err = time.ParseDuration(env.Get(EnvSleep, kvs.Get(Sleep)))
if err != nil {
return cfg, fmt.Errorf("'heal:max_sleep' value invalid: %w", err)
}
cfg.IOCount, err = strconv.Atoi(env.Get(EnvIOCount, kvs.Get(IOCount)))
if err != nil {
return cfg, fmt.Errorf("'heal:max_io' value invalid: %w", err)
}
cfg.Bitrot = bitrot == config.EnableOn
return cfg, nil
}

@ -153,7 +153,7 @@ type folderScanner struct {
// crawlDataFolder will crawl the basepath+cache.Info.Name and return an updated cache.
// The returned cache will always be valid, but may not be updated from the existing.
// Before each operation waitForLowActiveIO is called which can be used to temporarily halt the crawler.
// Before each operation sleepDuration is called which can be used to temporarily halt the crawler.
// If the supplied context is canceled the function will return at the first chance.
func crawlDataFolder(ctx context.Context, basePath string, cache dataUsageCache, getSize getSizeFn) (dataUsageCache, error) {
t := UTCNow()
@ -507,6 +507,7 @@ func (f *folderScanner) scanQueuedLevels(ctx context.Context, folders []cachedFo
// Dynamic time delay.
t := UTCNow()
resolver.bucket = bucket
foundObjs := false
@ -604,12 +605,6 @@ func (f *folderScanner) scanQueuedLevels(ctx context.Context, folders []cachedFo
// If we have quorum, found directories, but no objects, issue heal to delete the dangling.
objAPI.HealObjects(ctx, bucket, prefix, madmin.HealOpts{Recursive: true, Remove: true},
func(bucket, object, versionID string) error {
// Wait for each heal as per crawler frequency.
sleepDuration(time.Since(t), f.dataUsageCrawlMult)
defer func() {
t = UTCNow()
}()
return bgSeq.queueHealTask(healSource{
bucket: bucket,
object: object,

@ -55,8 +55,6 @@ type FSObjects struct {
// The count of concurrent calls on FSObjects API
activeIOCount int64
// The active IO count ceiling for crawling to work
maxActiveIOCount int64
// Path to be exported over S3 API.
fsPath string
@ -168,8 +166,6 @@ func NewFSObjectLayer(fsPath string) (ObjectLayer, error) {
listPool: NewTreeWalkPool(globalLookupTimeout),
appendFileMap: make(map[string]*fsAppendFile),
diskMount: mountinfo.IsLikelyMountPoint(fsPath),
maxActiveIOCount: 10,
}
// Once the filesystem has initialized hold the read lock for
@ -230,12 +226,6 @@ func (fs *FSObjects) StorageInfo(ctx context.Context, _ bool) (StorageInfo, []er
return storageInfo, nil
}
func (fs *FSObjects) waitForLowActiveIO() {
for atomic.LoadInt64(&fs.activeIOCount) >= fs.maxActiveIOCount {
time.Sleep(lowActiveIOWaitTick)
}
}
// CrawlAndGetDataUsage returns data usage stats of the current FS deployment
func (fs *FSObjects) CrawlAndGetDataUsage(ctx context.Context, bf *bloomFilter, updates chan<- DataUsageInfo) error {
// Load bucket totals

@ -186,7 +186,10 @@ func deepHealObject(bucket, object, versionID string) {
bucket: bucket,
object: object,
versionID: versionID,
opts: &madmin.HealOpts{ScanMode: madmin.HealDeepScan},
opts: &madmin.HealOpts{
Remove: true, // if found dangling purge it.
ScanMode: madmin.HealDeepScan,
},
}
}
}

@ -57,8 +57,8 @@ type Server struct {
}
// GetRequestCount - returns number of request in progress.
func (srv *Server) GetRequestCount() int32 {
return atomic.LoadInt32(&srv.requestCount)
func (srv *Server) GetRequestCount() int {
return int(atomic.LoadInt32(&srv.requestCount))
}
// Start - start HTTP server

@ -63,12 +63,6 @@ const (
// Size of each buffer.
readAheadBufSize = 1 << 20
// Wait interval to check if active IO count is low
// to proceed crawling to compute data usage.
// Wait up to lowActiveIOWaitMaxN times.
lowActiveIOWaitTick = 100 * time.Millisecond
lowActiveIOWaitMaxN = 10
// XL metadata file carries per object metadata.
xlStorageFormatFile = "xl.meta"
)
@ -90,8 +84,7 @@ func isValidVolname(volname string) bool {
// xlStorage - implements StorageAPI interface.
type xlStorage struct {
maxActiveIOCount int32
activeIOCount int32
activeIOCount int32
diskPath string
endpoint Endpoint
@ -262,13 +255,8 @@ func newXLStorage(ep Endpoint) (*xlStorage, error) {
},
},
globalSync: env.Get(config.EnvFSOSync, config.EnableOff) == config.EnableOn,
// Allow disk usage crawler to run with up to 2 concurrent
// I/O ops, if and when activeIOCount reaches this
// value disk usage routine suspends the crawler
// and waits until activeIOCount reaches below this threshold.
maxActiveIOCount: 3,
ctx: GlobalContext,
rootDisk: rootDisk,
ctx: GlobalContext,
rootDisk: rootDisk,
}
// Success.
@ -336,20 +324,6 @@ func (s *xlStorage) Healing() bool {
return err == nil
}
func (s *xlStorage) waitForLowActiveIO() {
max := lowActiveIOWaitMaxN
for atomic.LoadInt32(&s.activeIOCount) >= s.maxActiveIOCount {
time.Sleep(lowActiveIOWaitTick)
max--
if max == 0 {
if intDataUpdateTracker.debug {
logger.Info("waitForLowActiveIO: waited %d times, resuming", lowActiveIOWaitMaxN)
}
break
}
}
}
func (s *xlStorage) CrawlAndGetDataUsage(ctx context.Context, cache dataUsageCache) (dataUsageCache, error) {
// Check if the current bucket has a configured lifecycle policy
lc, err := globalLifecycleSys.Get(cache.Info.Name)
@ -402,24 +376,18 @@ func (s *xlStorage) CrawlAndGetDataUsage(ctx context.Context, cache dataUsageCac
if !version.Deleted {
// Bitrot check local data
if size > 0 && item.heal && opts.Bitrot {
s.waitForLowActiveIO()
err := s.VerifyFile(ctx, item.bucket, item.objectPath(), version)
switch err {
case errFileCorrupt:
res, err := objAPI.HealObject(ctx, item.bucket, item.objectPath(), oi.VersionID, madmin.HealOpts{
Remove: healDeleteDangling,
ScanMode: madmin.HealDeepScan,
})
if err != nil {
if !errors.Is(err, NotImplemented{}) {
logger.LogIf(ctx, err)
}
size = 0
} else {
size = res.ObjectSize
// HealObject verifies bitrot requirement internally
res, err := objAPI.HealObject(ctx, item.bucket, item.objectPath(), oi.VersionID, madmin.HealOpts{
Remove: healDeleteDangling,
ScanMode: madmin.HealDeepScan,
})
if err != nil {
if !errors.Is(err, NotImplemented{}) {
logger.LogIf(ctx, err)
}
default:
// VerifyFile already logs errors
size = 0
} else {
size = res.ObjectSize
}
}
totalSize += size

Loading…
Cancel
Save