Re: better architecture?

combining replies for the list:

On 11/19/14, Charles Zaffery <charlesz(at)focusschoolsoftware(dot)com> wrote:
> 2 and 3 can be covered by this:
> http://clusterlabs.org/wiki/PgSQL_Replicated_Cluster
does something similar exist for windows?

On 11/20/14, Michael Paquier <michael(dot)paquier(at)gmail(dot)com> wrote:
> On Thu, Nov 20, 2014 at 10:58 AM, zach cruise <zachc1980(at)gmail(dot)com> wrote:
>> 2. what happens if master-slave are rebooted at different times?
> What do you mean by that? If replication is impacted?
eg if i were to reboot the vmware running the master in the evening,
and reboot the vmware running the slave in the night, how would they
sync up?

On 11/20/14, Adrian Klaver wrote:
> What replication method are you using?
> The built in methods, Slony. Bucardo, etc?
built in

> The production users cannot enter or update records?
they can't. slave is read-only.

> If you have replication set up master -> slave, how can there be a difference between the two?
there isn't. both contain dev and prod databases. users connect to the
dev databases from the dev web server, and to the prod databases from
the prod web server.

> Not sure where the mssql databases into this?
our corporate partners use them. when i need to query against them, i import.

>> 2. what happens if master-slave are rebooted at different times?
>>
>> 3. i also need to auto-promote slave to master if master fails (without using repmgr or postgres-r or even postgres-xl). how?
> Answers for 2 & 3 are dependent on answers to the above questions.