Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.Original topic: dml提交卡在Binlog Prewrite 等待耗时
|
username: xxxxxxxx
To improve efficiency, please provide the following information. A clear problem description will help resolve the issue faster:
[TiDB Usage Environment]
Version 4.0.13, 3 TiKV, 3 TiDB, 3 PD, 3 Pump
[Overview] Scenario + Problem Description
Received an alert TiDB_binlog_error_total, and there is also feedback from the business that the cluster is slow in writing. After testing, it was indeed found to be very slow. Deleting a record based on the primary key takes a long time, taking more than ten seconds to complete the deletion.
There is no disk pressure or CPU load on the machine where the pump is located, and there are no error logs in the three pump components (grep -i err log/pump.log). I would like to ask what should be checked in this situation.
Additionally, disabling the pump component (binlog.enable: false) resolves the issue, so the problem should be with the pump component, but I am not sure how to locate it.
Another issue is that executing DDL operations, such as creating a new database and creating a new table, is extremely slow.
|
username: 
|
username: