br 备份报错

lipinbo · 2022 年7 月 6 日 08:42

nebula 版本：3.1.0
部署方式：单机
安装方式：RPM
是否为线上版本：Y

br 下载的是官网 → br 资源

问题描述：启动 graph meta storage
启动agent: ./agent --agent=“127.0.0.1:8888” --meta=“127.0.0.1:9559”
数据备份：./bin/br backup full --meta “192.168.1.195:9559” --storage “local:///mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/br/bakup” 报错

相关的日志：
{“file”:“github.com/vesoft-inc/nebula-br/pkg/clients/utils.go:20",“func”:“github.com/vesoft-inc/nebula-br/pkg/clients.connect”,“level”:“info”,"meta address”:“192.168.1.195:9559”,“msg”:“Try to connect meta service.”,“time”:“2022-07-06T01:28:36.849Z”}
{“file”:“github.com/vesoft-inc/nebula-br/pkg/clients/utils.go:44",“func”:“github.com/vesoft-inc/nebula-br/pkg/clients.connect”,“level”:“info”,"meta address”:“192.168.1.195:9559”,“msg”:“Connect meta server successfully.”,“time”:“2022-07-06T01:28:36.862Z”}
{“file”:“github.com/vesoft-inc/nebula-br/pkg/clients/utils.go:20",“func”:“github.com/vesoft-inc/nebula-br/pkg/clients.connect”,“level”:“info”,"meta address”:“127.0.0.1:9559”,“msg”:“Try to connect meta service.”,“time”:“2022-07-06T01:28:36.872Z”}
{“file”:“github.com/vesoft-inc/nebula-br/pkg/clients/utils.go:44",“func”:“github.com/vesoft-inc/nebula-br/pkg/clients.connect”,“level”:“info”,"meta address”:“127.0.0.1:9559”,“msg”:“Connect meta server successfully.”,“time”:“2022-07-06T01:28:36.875Z”}
{“file”:“github.com/vesoft-inc/nebula-br/pkg/utils/hosts.go:76",“func”:"github.com/vesoft-inc/nebula-br/pkg/utils.(*NebulaHosts).LoadFrom”,“host info”:“map[127.0.0.1:127.0.0.1:9779[STORAGE]: (data: /mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/install_dir/nebula/data/storage, root: /mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/install_dir/nebula) | 127.0.0.1:8888[AGENT]: (data: , root: ) | 127.0.0.1:9669[GRAPH]: (data: , root: /mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/install_dir/nebula) | 127.0.0.1:9559[META]: (data: /mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/install_dir/nebula/data/meta, root: /mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/install_dir/nebula)]”,“level”:“info”,“msg”:“Get cluster topology from the nebula.”,“time”:“2022-07-06T01:28:36.876Z”}
Start to backup cluster…
{“backup info”:“map[all spaces:false backup name: backups: created time:1969-12-31 16:00:00 -0800 PST full backup:false meta files:]”,“file”:“github.com/vesoft-inc/nebula-br/pkg/backup/backup.go:161",“func”:"github.com/vesoft-inc/nebula-br/pkg/backup.(*Backup).Backup”,“level”:“info”,“msg”:“Create backup in nebula machine’s local.”,“name”:"",“time”:“2022-07-06T01:28:36.983Z”}
{“file”:“github.com/vesoft-inc/nebula-br/pkg/backup/backup.go:172",“func”:"github.com/vesoft-inc/nebula-br/pkg/backup.(*Backup).Backup”,“level”:“info”,“msg”:“Ensure backup root dir.”,“name”:"",“root”:“local:///mnt/hgfs/Vmware_Share_Dir/nebulagraph/centros7/br/bakup”,“time”:“2022-07-06T01:28:36.986Z”}
Backup failed, will try to clean the remaining garbage…
Error: there is no meta files in backup info
Usage:
br backup full [flags]

Flags:
-h, --help help for full

Global Flags:
–debug Output log in debug level or not
–log string Specify br detail log path (default “br.log”)
–meta string Specify meta server
–s3.access_key string S3 Option: set access key id
–s3.endpoint string S3 Option: set the S3 endpoint URL, please specify the http or https scheme explicitly
–s3.region string S3 Option: set region or location to upload or download backup
–s3.secret_key string S3 Option: set secret key for access id
–spaces stringArray (EXPERIMENTAL)space names.
By this option, user can specify which spaces to backup. Now this feature is still experimental.
If not specified, will backup all spaces.

  --storage string         backup target url, format: <SCHEME>://<PATH>.
                               <SCHEME>: a string indicating which backend type. optional: local, s3.
                               now only s3-compatible is supported.
                               example:
                               for local - "local:///the/local/path/to/backup"
                               for s3  - "s3://example/url/to/the/backup"

panic: there is no meta files in backup info

goroutine 1 [running]:
main.main()
./main.go:16 +0x105

wey · 2022 年7 月 6 日 09:28

这里做了一个隐含的假设 agent 的 host 和服务要匹配，这里 meta 用的 192 的网段，agent 也要用这个网段哈。

弄好之后确定

show hosts meta
show hosts graph
show hosts storage

这三个结果的所有host 的 host 部分在

show hosts agent 里都有才行，比如一个是127.0.0.1 另一个是192.x.y.z 就不行了哈。

lipinbo · 2022 年7 月 7 日 06:02

三个服务的配置文件相关的ip 我把127.0.0.1 都改成了 192.168.x.x, agent 启动也改成了实际192.168.x.x
还是报同样的错误

日志信息：

lipinbo · 2022 年7 月 7 日 06:05

show agent 那个里面有个 127.0.0.1 不知道那个是不是有问题， drop hosts drop 不掉

lipinbo · 2022 年7 月 7 日 07:02

卸载了重新安装了一下，全都统一成 192.168.x.x 这样的了，还是不行

执行日志

kqzh · 2022 年7 月 7 日 07:47

你好，看日志这是另外一个错误了，想问下集群上存在space吗，如果没有的话可以先创建一个再执行backup

lipinbo · 2022 年7 月 7 日 09:01

有了

执行还是报错

wey · 2022 年7 月 7 日 09:18

那个 --storage 的local path 是正确存在的路径的哈？看你写成了 bakup，meta文件不存在的报错我还不熟悉，好像是 meta 创建备份完了还是空的造成的？ @kqzh

	if len(backupInfo.GetMetaFiles()) == 0 {
		return backupName, fmt.Errorf("there is no meta files in backup info")
	}

func (m *NebulaMeta) CreateBackup(spaces []string) (*meta.CreateBackupResp, error) {
	req := meta.NewCreateBackupReq()

	req.Spaces = make([][]byte, 0, len(spaces))
	for _, space := range spaces {
		req.Spaces = append(req.Spaces, []byte(space))
	}

	for {
		resp, err := m.client.CreateBackup(req)
		if err != nil {
			return nil, err
		}

		if resp.GetCode() == nebula.ErrorCode_E_LEADER_CHANGED {
			err := m.reconnect(resp.GetLeader())
			if err != nil {
				return nil, err
			}
			continue
		}

		return resp, nil
	}

}