Files
seaweedfs/weed/storage/store_load_balancing_test.go
T
e3e02d3364 [CheckDisk]: implement disk health detection (#9560)
* [CheckDisk][GRPC]: implement MVP for disk health detection, added timeout for new grpc connections

* fix(volume): build disk health check on every platform

setDiskStatus only existed behind the statfs build tag, so disk.go failed
to compile on windows, openbsd, solaris, netbsd and plan9. Move the timeout
wrapper and failure tracking into the shared disk.go and have each platform's
fillInDiskStatus return an error, so every platform gets the same protection
from a stuck filesystem.

Also restore the uint64(fs.Bavail) cast: Bavail is int64 on freebsd, so the
unguarded multiply broke the freebsd build.

* fix(volume): keep one outstanding statfs probe per disk

A stuck statfs used to leave isChecking cleared by the timeout path, so the
next check spawned another goroutine while the previous one was still blocked
in the syscall, leaking one goroutine per minute on a hung disk. Clear the
flag only when statfs returns and treat an overlapping check as a failure, so
a hung filesystem keeps a single outstanding probe and still gets reported.

* fix(volume): assume disk available until the first health check

isDiskAvailable defaulted to false, and CollectHeartbeat skips locations that
are not available. A freshly started volume server would therefore omit every
volume from its first heartbeats until the async CheckDiskSpace ran, so the
master could briefly treat all of them as missing.

* fix(volume): label the disk error metric by data directory

The new gauge tagged the series with IdxDirectory while every neighbouring
resource gauge uses Directory, so the error series would not line up with them
in dashboards. Also log the underlying error instead of a generic message.

* test(volume): cover disk health success and repeated-failure paths

* fix(volume): make a healthy disk the zero-value default

Track the disk as isDiskUnavailable instead of isDiskAvailable so the safe
state is the zero value, matching isDiskSpaceLow. CollectHeartbeat only skips a
location once a check has actively marked it unavailable, so any DiskLocation
built without running CheckDiskSpace (tests, future call sites) still reports
its volumes instead of silently dropping them.

* feat(disk): detect degraded disks using IO latency probes

* feat(stats): introduce configurable disk I/O health probe with EWMA-based latency detection

* feat(disk): replace EWMA with sliding window algorithm for disk health detection and added user-friendly options

* feat(disk): improve disk health probing and recovery

* feat(volume): configure disk health checks via volume.toml

* fix(volume): Remove disk IO probe CLI options

---------

Co-authored-by: ptukha <ptukha@tochka.com>
Co-authored-by: Chris Lu <chris.lu@gmail.com>
2026-06-02 09:02:05 -07:00

261 lines
7.2 KiB
Go

package storage
import (
"os"
"path/filepath"
"strconv"
"testing"
"github.com/seaweedfs/seaweedfs/weed/pb/volume_server_pb"
"github.com/seaweedfs/seaweedfs/weed/stats"
"github.com/seaweedfs/seaweedfs/weed/storage/needle"
"github.com/seaweedfs/seaweedfs/weed/storage/super_block"
"github.com/seaweedfs/seaweedfs/weed/storage/types"
"github.com/seaweedfs/seaweedfs/weed/util"
)
// newTestStore creates a test store with the specified number of directories
func newTestStore(t *testing.T, numDirs int) *Store {
tempDir := t.TempDir()
var dirs []string
var maxCounts []int32
var minFreeSpaces []util.MinFreeSpace
var diskTypes []types.DiskType
for i := 0; i < numDirs; i++ {
dir := filepath.Join(tempDir, "dir"+strconv.Itoa(i))
os.MkdirAll(dir, 0755)
dirs = append(dirs, dir)
maxCounts = append(maxCounts, 100) // high limit
minFreeSpaces = append(minFreeSpaces, util.MinFreeSpace{})
diskTypes = append(diskTypes, types.HardDriveType)
}
diskIOProbeConfig := stats.DefaultDiskIOProbeConfig()
store := NewStore(nil, "localhost", 8080, 18080, "http://localhost:8080", "",
dirs, maxCounts, minFreeSpaces, "", NeedleMapInMemory, diskTypes, nil, 3, diskIOProbeConfig)
// Consume channel messages to prevent blocking
done := make(chan bool)
go func() {
for {
select {
case <-store.NewVolumesChan:
case <-done:
return
}
}
}()
t.Cleanup(func() {
store.Close()
close(done)
})
return store
}
func TestLocalVolumesLen(t *testing.T) {
testCases := []struct {
name string
totalVolumes int
remoteVolumes int
expectedLocalCount int
}{
{
name: "all local volumes",
totalVolumes: 5,
remoteVolumes: 0,
expectedLocalCount: 5,
},
{
name: "all remote volumes",
totalVolumes: 5,
remoteVolumes: 5,
expectedLocalCount: 0,
},
{
name: "mixed local and remote",
totalVolumes: 10,
remoteVolumes: 3,
expectedLocalCount: 7,
},
{
name: "no volumes",
totalVolumes: 0,
remoteVolumes: 0,
expectedLocalCount: 0,
},
}
for _, tc := range testCases {
t.Run(tc.name, func(t *testing.T) {
diskLocation := &DiskLocation{
volumes: make(map[needle.VolumeId]*Volume),
}
// Add volumes
for i := 0; i < tc.totalVolumes; i++ {
vol := &Volume{
Id: needle.VolumeId(i + 1),
volumeInfo: &volume_server_pb.VolumeInfo{},
}
// Mark some as remote
if i < tc.remoteVolumes {
vol.hasRemoteFile = true
vol.volumeInfo.Files = []*volume_server_pb.RemoteFile{
{BackendType: "s3", BackendId: "test", Key: "test-key"},
}
}
diskLocation.volumes[vol.Id] = vol
}
result := diskLocation.LocalVolumesLen()
if result != tc.expectedLocalCount {
t.Errorf("Expected LocalVolumesLen() = %d; got %d (total: %d, remote: %d)",
tc.expectedLocalCount, result, tc.totalVolumes, tc.remoteVolumes)
}
})
}
}
func TestVolumeLoadBalancing(t *testing.T) {
testCases := []struct {
name string
locations []locationSetup
expectedLocations []int // which location index should get each volume
}{
{
name: "even distribution across empty locations",
locations: []locationSetup{
{localVolumes: 0, remoteVolumes: 0},
{localVolumes: 0, remoteVolumes: 0},
{localVolumes: 0, remoteVolumes: 0},
},
expectedLocations: []int{0, 1, 2, 0, 1, 2}, // round-robin
},
{
name: "prefers location with fewer local volumes",
locations: []locationSetup{
{localVolumes: 5, remoteVolumes: 0},
{localVolumes: 2, remoteVolumes: 0},
{localVolumes: 8, remoteVolumes: 0},
},
expectedLocations: []int{1, 1, 1}, // all go to location 1 (has fewest)
},
{
name: "ignores remote volumes in count",
locations: []locationSetup{
{localVolumes: 2, remoteVolumes: 10}, // 2 local, 10 remote
{localVolumes: 5, remoteVolumes: 0}, // 5 local
{localVolumes: 3, remoteVolumes: 0}, // 3 local
},
// expectedLocations: []int{0, 0, 2}
// Explanation:
// 1. Initial local counts: [2, 5, 3]. First volume goes to location 0 (2 local, ignoring 10 remote).
// 2. New local counts: [3, 5, 3]. Second volume goes to location 0 (first with min count 3).
// 3. New local counts: [4, 5, 3]. Third volume goes to location 2 (3 local < 4 local).
expectedLocations: []int{0, 0, 2},
},
{
name: "balances when some locations have remote volumes",
locations: []locationSetup{
{localVolumes: 1, remoteVolumes: 5},
{localVolumes: 1, remoteVolumes: 0},
{localVolumes: 0, remoteVolumes: 3},
},
// expectedLocations: []int{2, 0, 1}
// Explanation:
// 1. Initial local counts: [1, 1, 0]. First volume goes to location 2 (0 local).
// 2. New local counts: [1, 1, 1]. Second volume goes to location 0 (first with min count 1).
// 3. New local counts: [2, 1, 1]. Third volume goes to location 1 (next with min count 1).
expectedLocations: []int{2, 0, 1},
},
}
for _, tc := range testCases {
t.Run(tc.name, func(t *testing.T) {
// Create test store with multiple directories
store := newTestStore(t, len(tc.locations))
// Pre-populate locations with volumes
for locIdx, setup := range tc.locations {
location := store.Locations[locIdx]
vidCounter := 1000 + locIdx*100 // unique volume IDs per location
// Add local volumes
for i := 0; i < setup.localVolumes; i++ {
vol := createTestVolume(needle.VolumeId(vidCounter), false)
location.SetVolume(vol.Id, vol)
vidCounter++
}
// Add remote volumes
for i := 0; i < setup.remoteVolumes; i++ {
vol := createTestVolume(needle.VolumeId(vidCounter), true)
location.SetVolume(vol.Id, vol)
vidCounter++
}
}
// Create volumes and verify they go to expected locations
for i, expectedLoc := range tc.expectedLocations {
volumeId := needle.VolumeId(i + 1)
err := store.AddVolume(volumeId, "", NeedleMapInMemory, "000", "",
0, needle.GetCurrentVersion(), 0, types.HardDriveType, 3)
if err != nil {
t.Fatalf("Failed to add volume %d: %v", volumeId, err)
}
// Find which location got the volume
actualLoc := -1
for locIdx, location := range store.Locations {
if _, found := location.FindVolume(volumeId); found {
actualLoc = locIdx
break
}
}
if actualLoc != expectedLoc {
t.Errorf("Volume %d: expected location %d, got location %d",
volumeId, expectedLoc, actualLoc)
// Debug info
for locIdx, loc := range store.Locations {
localCount := loc.LocalVolumesLen()
totalCount := loc.VolumesLen()
t.Logf(" Location %d: %d local, %d total", locIdx, localCount, totalCount)
}
}
}
})
}
}
// Helper types and functions
type locationSetup struct {
localVolumes int
remoteVolumes int
}
func createTestVolume(vid needle.VolumeId, isRemote bool) *Volume {
vol := &Volume{
Id: vid,
SuperBlock: super_block.SuperBlock{},
volumeInfo: &volume_server_pb.VolumeInfo{},
}
if isRemote {
vol.hasRemoteFile = true
vol.volumeInfo.Files = []*volume_server_pb.RemoteFile{
{BackendType: "s3", BackendId: "test", Key: "remote-key-" + strconv.Itoa(int(vid))},
}
}
return vol
}