On Wed, May 09, 2018 at 09:04:04AM +0900, Tatsuo Ishii wrote:
> I have simulated the idea and I wonder how to implement the query
> result cache on the streaming standby servers because no DML/DDL are
> executed on standby servers, that makes it impossible to invalidate
> the query cache. Probably the only way to allow to use the query cache
> is,
>
> 1) Let invalidate the query cache on the primary server.
>
> 2) The cache storage needed to be on the external cache server like
> memcached.
Or a hook within the REDO loop which can be processed for each record?
You could take any actions needed with that by tracking mostly heap
records.
--
Michael
