gian/cubefs
1
0
Fork 0
mirror of https://github.com/cubefs/cubefs.git synced 2026-08-02 02:00:56 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

fix(data): bad disk judgment logic fix. #1000184384

Browse Source 
Signed-off-by: zhumingze <zhumingze@oppo.com>
This commit is contained in:
zhumingze 2025-06-18 19:31:53 +08:00 committed by zhumingze1108
parent 33c6e2d96d
commit 9e9eb0e753
2 changed files with 16 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
datanode/disk.go
View File

@ -616,6 +616,10 @@ func (d *Disk) triggerDiskError(rwFlag uint8, dpId uint64) {
// exporter.Warning(mesg)
log.LogWarnf(mesg)
if d.HasDiskErrPartition(dpId) {
return
}
if rwFlag == WriteFlag {
d.incWriteErrCnt()
} else if rwFlag == ReadFlag {
@ -628,7 +632,7 @@ func (d *Disk) triggerDiskError(rwFlag uint8, dpId uint64) {
d.AddDiskErrPartition(dpId)
diskErrCnt := d.getTotalErrCnt()
diskErrPartitionCnt := d.GetDiskErrPartitionCount()
if diskErrPartitionCnt >= d.dataNode.diskUnavailablePartitionErrorCount {
if diskErrCnt >= d.dataNode.diskUnavailableErrorCount || diskErrPartitionCnt >= d.dataNode.diskUnavailablePartitionErrorCount {
msg := fmt.Sprintf("set disk unavailable for too many disk error, "+
"disk path(%v), ip(%v), diskErrCnt(%v), diskErrPartitionCnt(%v) threshold(%v)",
d.Path, LocalIP, diskErrCnt, diskErrPartitionCnt, d.dataNode.diskUnavailablePartitionErrorCount)

11
datanode/server.go
View File

@ -145,6 +145,7 @@ const (
ConfigGcRecyclePercent = "gcRecyclePercent"
// disk status becomes unavailable if disk error partition count reaches this value
ConfigKeyDiskUnavailableErrorCount = "diskUnavailableErrorCount"
ConfigKeyDiskUnavailablePartitionErrorCount = "diskUnavailablePartitionErrorCount"
ConfigKeyCacheCap = "cacheCap"
ConfigExtentCacheTtlByMin = "extentCacheTtlByMin"
@ -234,6 +235,7 @@ type DataNode struct {
gcRecyclePercent float64
gcTimer *util.RecycleTimer
diskUnavailableErrorCount uint64 // disk status becomes unavailable when disk error count reaches this value
diskUnavailablePartitionErrorCount uint64 // disk status becomes unavailable when disk error partition count reaches this value
started int32
dpBackupTimeout time.Duration
@ -472,6 +474,15 @@ func (s *DataNode) parseConfig(cfg *config.Config) (err error) {
s.gcRecyclePercent = defaultGcRecyclePercent
}
diskUnavailableErrorCount := cfg.GetInt64(ConfigKeyDiskUnavailableErrorCount)
if diskUnavailableErrorCount <= 0 || diskUnavailableErrorCount > 100 {
diskUnavailableErrorCount = DefaultDiskUnavailableErrorCount
log.LogDebugf("action[parseConfig] ConfigKeyDiskUnavailableErrorCount(%v) out of range, set as default(%v)",
diskUnavailableErrorCount, DefaultDiskUnavailableErrorCount)
}
s.diskUnavailableErrorCount = uint64(diskUnavailableErrorCount)
log.LogDebugf("action[parseConfig] load diskUnavailableErrorCount(%v)", s.diskUnavailableErrorCount)
diskUnavailablePartitionErrorCount := cfg.GetInt64(ConfigKeyDiskUnavailablePartitionErrorCount)
if diskUnavailablePartitionErrorCount <= 0 || diskUnavailablePartitionErrorCount > 100 {
diskUnavailablePartitionErrorCount = DefaultDiskUnavailablePartitionErrorCount