cdc主从同步延迟

[像风一样的男子]

【 TiDB 使用环境】生产环境 /测试/ Poc
cdc同步到下游tso一直在推进，但是延迟越来越高，下游也是tidb集群，主要有几张大表延迟很高。怎么提高同步速度？
[2024/12/13 14:09:38.524 +08:00] [INFO] [replication_manager.go:759] [“schedulerv3: slow table”] [namespace=default] [changefeed=slave-tidb] [tableID=265] [tableStatus=Replicating] [checkpointTs=454574902863987182] [resolvedTs=454576092552102092] [checkpointLag=1h15m39.500514064s]
[2024/12/13 14:09:38.524 +08:00] [INFO] [replication_manager.go:759] [“schedulerv3: slow table”] [namespace=default] [changefeed=slave-tidb] [tableID=235] [tableStatus=Replicating] [checkpointTs=454574835610943668] [resolvedTs=454576092552102092] [checkpointLag=1h19m56.050514064s]
[2024/12/13 14:09:38.524 +08:00] [INFO] [replication_manager.go:759] [“schedulerv3: slow table”] [namespace=default] [changefeed=slave-tidb] [tableID=187] [tableStatus=Replicating] [checkpointTs=454574835348799609] [resolvedTs=454576092552102092] [checkpointLag=1h19m57.050514064s]
[2024/12/13 14:09:38.524 +08:00] [INFO] [replication_manager.go:759] [“schedulerv3: slow table”] [namespace=default] [changefeed=slave-tidb] [tableID=267] [tableStatus=Replicating] [checkpointTs=454574769681727561] [resolvedTs=454576092552102092] [checkpointLag=1h24m7.550514064s]

[nobody (不定时出现)]

这里需要判断下下游 tidb 以及 ticdc 到底哪个是瓶颈。
如果是 ticdc ，可以考虑将大表单独一个 changefeed , 或者把这个表调度到压力小的 ticdc 节点上，或者调整 worker-count。
如果是下游 tidb ,则按照写入优化的思路去调优即可。

[像风一样的男子]

结案了，资源管控设置太低了。调大后恢复了。