关于rac环境下flush redolog情况

aquarius

当发生global cache current blocks的传输时，会发生flush redo log,这点比较好理解！但是，为什么构造cr块，发生global cache cr blocks也会触发flush redo log呢？

aquarius

恩  对的   先构造好了再传到request instance上的

gdpr-dba

原帖由 aquarius 于 2012-4-19 22:09 发表
好 谢谢  明白了  就是说被修改过的块要将当前的image传输到request 的instance上，以便在request instance上构造cr block,此时就要将相关dml先flush下

我觉得应该是先构造cr block再传过去的，而不是要把current block传输到request 的instance上再构造吧？

aquarius

好 谢谢  明白了  就是说被修改过的块要将当前的image传输到request 的instance上，以便在request instance上构造cr block,此时就要将相关dml先flush下

gdpr-dba

原帖由 maclean 于 2012-4-19 21:44 发表
A log flush for a cr block transfer is required when a cr buffer is cloned from a current buffer which has redo
pending. A high percentage is indicative of hot blocks with frequent read after write ac ...

不明白，cr block不都是从current block构造的吗？

Maclean Liu(刘相兵

PS:

可以通过 设置_cr_server_log_flush to false(LMS are/is waiting for LGWR to flush the pending redo during CR fabrication. Without going too much in to details, you can turn off the behaviour by setting   _cr_server_log_flush to false.) 来禁止cr server flush redo log，但是该参数对于current block的flush time无效， 也强烈不推荐使用。

_cr_server_log_flush         TRUE         TRUE         if TRUE, flush redo log before serving a CR buffer

Maclean Liu(刘相兵

A log flush for a cr block transfer is required when a cr buffer is cloned from a current buffer which has redo
pending. A high percentage is indicative of hot blocks with frequent read after write access.


consistent read 中  cr block 可能来源于 current block 的 clone ， 而current block 的相关redo 可能会未写出 ， 需要这些 redo flush后 cr block才能被send 。