CurveBs|tools-v2|rebuild check-consistency,list may-broken-vol, check-chunkserver #2132

jolly-sy · 2022-12-03T13:50:10Z

the content of this pr is about rebuilding three command :check-consistency,list may-broken-vol, check-chunkserver

check-consistency:

list may broken vol:

check- chunkserver

jolly-sy · 2022-12-13T08:41:40Z

cicheck

Cyber-SiKu · 2022-12-13T11:13:38Z

All output should be printed using tableNew, please do not print by yourself, this output is not friendly.
If you need to print multiple tables, please consider splitting them into multiple commands and pay attention to the uniform style of the tables.

所有的输出都应该使用tableNew 打印，请不要自行打印，这样输出不友好。
如果需要打印多个表格，请考虑拆分成多个命令，注意表格风格的统一。

Cyber-SiKu · 2022-12-13T11:22:29Z

tools-v2/internal/error/error.go

@@ -357,7 +357,7 @@ var (
 		return NewInternalCmdError(36, "get time of day fail, the error is: %s")
 	}
 	ErrBsGetFileInfo = func() *CmdError {
-		return NewInternalCmdError(37, "get file info fail, the error is: %s")
+		return NewInternalCmdError(37, "%s")


why delete this？

why delete this？

coding error, fix

Cyber-SiKu · 2022-12-13T11:24:30Z

tools-v2/internal/proto/curvebs/topology/statuscode.proto

+    SegmentNotAllocated = 107; // segment不存在
+    FileNotExists = 102;  // 文件不存在
+    OtherError = -3;  // 其他错误
+}
+
+enum CopysetStatusCode{
+    CopysetSuccess = 0;
+    Exist = 1;     // copyset node 已经存在
+    NotExist = 2;
+    UnknowErr = 3;
+    Healthy = 4;
+}
+
+enum ChunkStatusCode{
+    ChunkSuccess = 0;
+    Redirect = 1;                     //不是leader,重定向
+    DiskErr = 2;                     //磁盘返回错误
+    CrcFail = 3;                    //Crc校验错误
+    InvalidReqParam = 4;           //请求参数错误
+    NoEnoughSpace = 5;            //空间不够
+    CopysetNotExist = 6;         //copyset不存在
+    ChunkNotExist = 7;          //chunk或其快照文件不存在
+    UnknowFailure = 8;          //未知错误
+    OverLoad = 9;               //服务端过载
+    Backward = 10;             // 请求的版本落后当前chunk的版本
+    ChunkExist = 11;           //chunk已存在
+    EpochOld = 12;             // 请求中的epoch太老了


English and alignment

English and alignment

fix

Cyber-SiKu · 2022-12-13T11:24:53Z

tools-v2/pkg/cli/command/curvebs/check/chunkserver/chunkserver.go

+	kCpsHealthy        = 0  // chunkserver上所有copyset健康
+	kNotHealthy        = -1 // chunkserver上有copyset不健康
+	kNotOnline         = -2 // chunkserver不在线


english

fix

Cyber-SiKu · 2022-12-13T11:26:38Z

tools-v2/pkg/cli/command/curvebs/check/chunkserver/chunkserver.go

+	row["total copysets"] = total
+	row["unhealthy copysets"] = unCnt
+	row["unhealthy ratio"] = ratio


add in internal/utils/row.go

add in internal/utils/row.go

fix

Cyber-SiKu · 2022-12-13T11:27:44Z

tools-v2/pkg/cli/command/curvebs/check/chunkserver/copyset.go

+	// 解析结果失败
+	kParseError = -1
+	// peer数量小于预期
+	kPeersNoSufficient = -2
+	// 副本间的index差距太大
+	kLogIndexGapTooBig = -3
+	// 有副本在安装快照
+	kInstallingSnapshot = -4
+	// 少数副本不在线
+	kMinorityPeerNotOnline = -5
+	// 大多数副本不在线


english

fix

Cyber-SiKu · 2022-12-13T11:28:05Z

tools-v2/pkg/cli/command/curvebs/check/chunkserver/copyset.go

+	for _, cpsInfo := range copysetInfos {
+		cpsIds = append(cpsIds, cpsInfo.GetCopysetId())
+	}
+	// fmt.Println("cpsIds is", cpsIds)


jolly-sy · 2022-12-19T06:28:11Z

这条命令会根据情况产生多种不一样的结果，表头无法统一，所以无法使用表格统一打印

Cyber-SiKu · 2022-12-19T06:34:52Z

这条命令会根据情况产生多种不一样的结果，表头无法统一，所以无法使用表格统一打印

The table header is not necessarily formulated in init, you can add the header when you are sure.

jolly-sy · 2022-12-19T09:57:44Z

cicheck

Cyber-SiKu · 2023-01-04T02:58:37Z

proto/topology.proto

+
+
+


remove extra blank lines

remove extra blank lines

fix

Cyber-SiKu · 2023-01-04T02:58:46Z

proto/topology.proto

+    required sint32 statusCode = 1;
+    repeated PoolsetInfo poolsetInfos = 2;
+}
+


remove extra blank lines

remove extra blank lines

fix

Cyber-SiKu · 2023-01-04T03:03:04Z

tools-v2/testTopo.json

@@ -0,0 +1,51 @@
+{


move to tools-v2/pkg/config/

move to tools-v2/pkg/config/

fix

Cyber-SiKu · 2023-01-04T03:03:28Z

tools-v2/pkg/config/template.yaml

@@ -23,3 +23,4 @@ curvebs:
  root:
    user: root
    password: root_password
+


remove extra blank lines

Cyber-SiKu · 2023-01-04T03:07:51Z

tools-v2/pkg/config/bs.go

+	CURVEBS_PASSWORD               = "password"
+	VIPER_CURVEBS_PASSWORD         = "curvebs.root.password"
+	CURVEBS_DEFAULT_PASSWORD       = "root_password"
+	CURVEBS_CLUSTERMAP             = "bsClusterMap"


lowercase

all lowercase is ambiguous

Cyber-SiKu · 2023-01-04T03:08:13Z

tools-v2/pkg/config/bs.go

+	VIPER_CURVEBS_PORT             = "curvebs.port"
+	CURVEBS_CHUNKSERVER_ID         = "csId"
+	VIPER_CURVEBS_CHUNKSERVER_ID   = "curvebs.csId"
+	CURVEBS_CHUNKSERVER_ADDR       = "csAddr"


Cyber-SiKu · 2023-01-04T03:08:21Z

tools-v2/pkg/config/bs.go

+	CURVEBS_CHUNKSERVER_ID         = "csId"
+	VIPER_CURVEBS_CHUNKSERVER_ID   = "curvebs.csId"
+	CURVEBS_CHUNKSERVER_ADDR       = "csAddr"
+	VIPER_CURVEBS_CHUNKSERVER_ADDR = "curvebs.csAddr"


ditto

all lowercase is ambiguous

Cyber-SiKu · 2023-01-04T03:08:27Z

tools-v2/pkg/config/bs.go

+	VIPER_CURVEBS_CHUNK_SIZE       = "curvebs.chunkSize"
+	CURVEBS_PORT                   = "port"
+	VIPER_CURVEBS_PORT             = "curvebs.port"
+	CURVEBS_CHUNKSERVER_ID         = "csId"


ditto

all lowercase is ambiguous

Cyber-SiKu · 2023-01-04T03:09:32Z

tools-v2/pkg/cli/curvecli.go

@@ -47,7 +47,7 @@ func setupRootCommand(cmd *cobra.Command) {
 }

 func newCurveCommand() *cobra.Command {
-	cmd := &cobra.Command{
+	rootCmd := &cobra.Command{


why rename？

why rename？

for test , I will change it back

Cyber-SiKu · 2023-01-04T03:11:41Z

tools-v2/pkg/cli/command/curvefs/create/topology/zone.go

@@ -166,7 +166,7 @@ func (tCmd *TopologyCommand) scanZones() *cmderror.CmdError {

 func (tCmd *TopologyCommand) removeZones() *cmderror.CmdError {
 	tCmd.deleteZoneRpc = &DeleteZoneRpc{}
-	tCmd.deleteZoneRpc.Info = basecmd.NewRpc(tCmd.addrs, tCmd.timeout, tCmd.retryTimes, "DeleteServer")
+	tCmd.deleteZoneRpc.Info = basecmd.NewRpc(tCmd.addrs, tCmd.timeout, tCmd.retryTimes, "DeleteZone")


good job！

Cyber-SiKu · 2023-01-04T03:12:12Z

tools-v2/pkg/cli/command/curvebs/status/etcd/etcd.go

@@ -189,26 +189,26 @@ func (eCmd *EtcdCommand) ResultPlainOutput() error {
 }

 func NewStatusEtcdCommand() *EtcdCommand {
-	etcdCmd := &EtcdCommand{


why rename？

why rename？

coding error. fix

Cyber-SiKu · 2023-01-04T03:13:40Z

tools-v2/pkg/cli/command/curvebs/query/file/fileinfo.go

 	// fmt.Println("user:", owner)
 	// fmt.Println("viper:", viper.GetString(config.VIPER_CURVEBS_USER))


you can delete this

you can delete this

fix

Cyber-SiKu · 2023-01-04T03:20:12Z

tools-v2/pkg/cli/command/curvebs/create/cluster/topo.go

+func (ctCmd *ClusterTopoCmd) RunCommand(cmd *cobra.Command, args []string) error {
+	err := ctCmd.updateTopology()
+	if err.TypeCode() != cmderror.CODE_SUCCESS {
+		//return fmt.Errorf(err.Message)


remove

fix

Cyber-SiKu · 2023-01-04T03:20:50Z

tools-v2/pkg/cli/command/curvebs/list/maydamagedvolumes/chunkserver.go

+}
+
+func (mdvCmd *MayDamVolCmd) ListChunkServersOnServers() *cmderror.CmdError {
+	// fmt.Println("ListChunkServersOnServers() in")


Cyber-SiKu · 2023-01-04T03:20:58Z

tools-v2/pkg/cli/command/curvebs/list/maydamagedvolumes/chunkserver.go

+				mdvCmd.isRetired = true
+				continue
+			}
+			//fmt.Println(chunkServer)


Cyber-SiKu · 2023-01-04T03:21:07Z

tools-v2/pkg/cli/command/curvebs/list/maydamagedvolumes/chunkserver.go

+}
+
+func (mdvCmd *MayDamVolCmd) CheckIfChunkServerOnline(csAddr string) bool {
+	// fmt.Printf("CheckIfChunkServerOnline() in, csAddr is csAddr %s", csAddr)


Cyber-SiKu · 2023-01-04T03:21:15Z

tools-v2/pkg/cli/command/curvebs/list/maydamagedvolumes/chunkserver.go

+}
+
+func (mdvCmd *MayDamVolCmd) FindOffLineChunkServers() *cmderror.CmdError {
+	// fmt.Println("FindOffLineChunkServers() in")


Cyber-SiKu · 2023-01-04T03:26:08Z

tools-v2/pkg/cli/command/curvebs/list/maydamagedvolumes/maydamagedvolumes.go

+	// 列出集群中所有的server
+	err := mdvCmd.GetServersInfo()
+	if err.TypeCode() != cmderror.CODE_SUCCESS {
+		return err.ToError()
+	}
+	// 列出每台server上所有的chunkserver
+	err1 := mdvCmd.ListChunkServersOnServers()
+	if err1.TypeCode() != cmderror.CODE_SUCCESS {
+		return err1.ToError()
+	}
+	// 找出所有离线的chunkserver并记录这些chunkserver的地址到mdvCmd.offLineCsAddrs
+	err = mdvCmd.FindOffLineChunkServers()
+	if err.TypeCode() != cmderror.CODE_SUCCESS {
+		return err.ToError()
+	}
+	if len(mdvCmd.offLineCsAddrs) == 0 {
+		// fmt.Println("len(mdvCmd.offLineCsAddrs) == 0")
+		return err.ToError()
+	}
+	// 获取所有离线server上的copysets
+	err = mdvCmd.GetCopySetsOnOffLineCs()
+	if err.TypeCode() != cmderror.CODE_SUCCESS {
+		return err.ToError()
+	}
+	//统计各个copysets出现的次数,2次及以上的视为损坏的copyset，记录进mdvCmd.damagedCps


English note

English note

fix

Cyber-SiKu

delete tools-v2/pkg/cli/command/curvebs/list/topo

Cyber-SiKu · 2023-01-04T03:31:24Z

tools-v2/pkg/cli/command/curvebs/list/maydamagedvolumes/maydamagedvolumes.go

+func NewMayDamVolCmd() *cobra.Command {
+	mdvCmd := &MayDamVolCmd{
+		FinalCurveCmd: basecmd.FinalCurveCmd{
+			Use:     "may-damaged-vols",


damaged-vols may be better？

damaged-vols may be better？

no, original name is may-broken-vol, may can't be removed

Signed-off-by: jolly-sy <[email protected]>

aspirer · 2023-02-16T02:45:14Z

we should split these commands and commit them one by one.

jolly-sy force-pushed the bsTool branch 2 times, most recently from 5a31ea7 to 9d0426e Compare December 4, 2022 03:20

jolly-sy force-pushed the bsTool branch 2 times, most recently from 5be2154 to 41e77ac Compare December 13, 2022 05:50

Cyber-SiKu requested changes Dec 13, 2022

View reviewed changes

jolly-sy force-pushed the bsTool branch from 41e77ac to a8b979d Compare December 19, 2022 09:52

Cyber-SiKu reviewed Jan 4, 2023

View reviewed changes

jolly-sy force-pushed the bsTool branch from a8b979d to 521e7c6 Compare January 4, 2023 07:26

jolly-sy added 2 commits January 4, 2023 15:33

curvebs|tools:bs create topology --bs bsClusterMap filePath

b5c37e1

Signed-off-by: jolly-sy <[email protected]>

curveBs|tools-v2:check-consistency,list maybrokenvol

696ef10

Signed-off-by: jolly-sy <[email protected]>

jolly-sy force-pushed the bsTool branch 2 times, most recently from c3ed48d to 902c0e7 Compare January 4, 2023 08:25

curvebs|tools-v2|rebuild cil check-chunkserver

f910b9d

Signed-off-by: jolly-sy <[email protected]>

jolly-sy force-pushed the bsTool branch from 902c0e7 to f910b9d Compare January 6, 2023 02:38

aspirer closed this Feb 16, 2023

This was referenced Apr 7, 2023

[curve/toos-v2]: add list-may-broken-vol #2356

Closed

[curve/toos-v2]: add check-chunkserver #2353

Closed

[curve/toos-v2]: add check-consistency #2351

Open

		// fmt.Println("user:", owner)
		// fmt.Println("viper:", viper.GetString(config.VIPER_CURVEBS_USER))

CurveBs|tools-v2|rebuild check-consistency,list may-broken-vol, check-chunkserver #2132

CurveBs|tools-v2|rebuild check-consistency,list may-broken-vol, check-chunkserver #2132

Conversation

jolly-sy commented Dec 3, 2022

jolly-sy commented Dec 13, 2022

Cyber-SiKu commented Dec 13, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jolly-sy commented Dec 19, 2022

Cyber-SiKu commented Dec 19, 2022

jolly-sy commented Dec 19, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Cyber-SiKu left a comment

Choose a reason for hiding this comment

Cyber-SiKu Jan 4, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

aspirer commented Feb 16, 2023

Cyber-SiKu commented Dec 13, 2022 •

edited

Loading

Cyber-SiKu Jan 4, 2023 •

edited

Loading