DM迁移-卡在Load阶段

🪐 TiDB 技术问题
1

【 TiDB 使用环境】测试
【 TiDB 版本】v8.1.1
【复现路径】
【遇到的问题：问题现象及影响】

构造了一个30G左右大小的MySQL实例
想使用DM迁移到TIDB8.1.1

迁移任务过了dump阶段后，卡在Load阶段快一个小时了
表结构已经同步到目标库了，但是数据还是空的

【资源配置】进入到 TiDB Dashboard -集群信息 (Cluster Info) -主机(Hosts) 截图此页面
【附件：截图/日志/监控】

Starting component dmctl: /home/tidb/.tiup/components/dmctl/v8.1.1/dmctl/dmctl --master-addr 10.106.47.32:8261 query-status sync_task
{
    "result": true,
    "msg": "",
    "sources": [
        {
            "result": true,
            "msg": "",
            "sourceStatus": {
                "source": "mysql-source",
                "worker": "dm-10.106.47.32-8262",
                "result": null,
                "relayStatus": null
            },
            "subTaskStatus": [
                {
                    "name": "sync_task",
                    "stage": "Running",
                    "unit": "Load",
                    "result": null,
                    "unresolvedDDLLockID": "",
                    "load": {
                        "finishedBytes": "0",
                        "totalBytes": "0",
                        "progress": "0.00 %",
                        "metaBinlog": "(mysql-bin.000022, 653114714)",
                        "metaBinlogGTID": "4f684da2-9340-11ef-b912-fefcfeaae986:1-36050",
                        "bps": "0"
                    },
                    "validation": null
                }
            ]
        }
    ]
}
2

大无语
刚发完帖子，它就有动静了

Starting component dmctl: /home/tidb/.tiup/components/dmctl/v8.1.1/dmctl/dmctl --master-addr 10.106.47.32:8261 query-status sync_task
{
    "result": true,
    "msg": "",
    "sources": [
        {
            "result": true,
            "msg": "",
            "sourceStatus": {
                "source": "mysql-source",
                "worker": "dm-10.106.47.32-8262",
                "result": null,
                "relayStatus": null
            },
            "subTaskStatus": [
                {
                    "name": "sync_task",
                    "stage": "Running",
                    "unit": "Load",
                    "result": null,
                    "unresolvedDDLLockID": "",
                    "load": {
                        "finishedBytes": "48093128",
                        "totalBytes": "11780317302",
                        "progress": "0.41 %",
                        "metaBinlog": "(mysql-bin.000022, 653114714)",
                        "metaBinlogGTID": "4f684da2-9340-11ef-b912-fefcfeaae986:1-36050",
                        "bps": "1216483"
                    },
                    "validation": null
                }
            ]
        }
    ]
}
1 个赞
3

也不白发帖，像这种大批量导入，对应的表你要注意写入热点。

https://docs.pingcap.com/zh/tidb/stable/troubleshoot-hot-spot-issues

注意观察所有的tikv的负载是否均衡上升直到cpu跑满。
如果是1个tikv忙死，其他的围观，就要考虑是否是写入热点，跟着上面的文档看看流量可视化图，解决一下。

2 个赞
4

感谢提醒
我这是用sysbench 构造的100个database
每个database里面100张表
每张表大小10000

1 个赞
5

此话题已在最后回复的 7 天后被自动关闭。不再允许新回复。