The DDL owner status of all nodes in the cluster is 0

Translated
1

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: 集群中所有节点的ddl owner 的状态都为0

| username: dxss-lee

In TiDB v6.1.0, the system encountered an issue where it was unable to create tables. The SHOW DDL JOB command revealed that the status was “queue” and the table name was empty. After canceling the job, the status changed to “cancelling,” but it still could not be completed. All DDL operations were unable to proceed. Using the SELECT tidb_is_ddl_owner() statement showed that all TiDB instances returned 0. The issue was resolved by restarting all TiDB nodes using tiup restart.

2
| username: 我是咖啡哥 | Original post link

What did you do to him? :joy:

3
| username: dxss-lee | Original post link

Normal create table and truncate table operations don’t have any special procedures. The frequency is slightly higher, with a few DDLs per minute.

4
| username: tidb菜鸟一只 | Original post link

TiDB isn’t that fragile, is it? It crashes after a few DDL operations in one minute?

5
| username: dxss-lee | Original post link

I don’t know why the DDL owner hasn’t been elected. It should be a small bug.

6
| username: Min_Chen | Original post link

Can you provide the logs and monitoring data from that time?

7
| username: dxss-lee | Original post link

It happened again today.

8
| username: Billmay表妹 | Original post link

【TiDB Usage Environment】Production Environment / Testing / PoC
【TiDB Version】
【Reproduction Path】What operations were performed when the issue occurred
【Encountered Issue: Issue Symptoms and Impact】
【Resource Configuration】
【Attachments: Screenshots / Logs / Monitoring】

A more detailed description would help in making a better judgment~

9
| username: xingzhenxiang | Original post link

Let’s see the result of ADMIN SHOW DDL JOBS;.