e-mhui

Results 5 comments of e-mhui

:+1: This PR can improve the speed of querying tableid, but the select operation during the snapshot is still slow.

@gong Take a look at this, I saw that you have addressed it in another PR. https://github.com/apache/flink-cdc/pull/2976

Hi,I used the method in this PR to close the connection during the snapshot, but the effect is still not obvious. Synchronizing 50 million pieces of data occupied 5000 sessions.

能否和 hadoop 一样,在同一个级别下不允许创建同名的文件和目录。 或者是提供其他办法来判断是否是文件还是目录,而不是使用 listStatus 方法来判断。

> > 能否和 hadoop 一样,在同一个级别下不允许创建同名的文件和目录。 或者是提供其他办法来判断是否是文件还是目录,而不是使用 listStatus 方法来判断。 > > 如果方便的话,也欢迎提修复 PR。 好的,之前简单看了下,使用到这个方法的地方有很多,在原有方法上改动会比较大。