[afnog] NEED HELP

[Simon Vass]

If your running OpenERP 6.1, I think you need to look at Postgresql 
Enterprise Clustering, and running multiple openerp-servers load 
balanaced on the front end.

On 28/02/13 13:22, Johan Bernhardsson wrote:
> Application was openerp :) But yes a master <-> master replication is 
> needed and I'm quite unsure on how postgresql (that openerp uses as 
> backend database) handles that. I don't think it will do so out of the 
> box.
>
> I have many designs for high availability but none of the includes 
> openerp :)
>
> Best way would to do a setup with kvm/libvirt and migrate a virtual 
> host on failure.
>
> /Johan Bernhardsson
>
>
> 2013-02-28 11:15 skrev Phil Regnauld:
>> Johan Bernhardsson (johan) writes:
>>> If databases are replicated so they exists on both nodes then i
>>> would go for a HA solution with heartbeat or one virtual machine
>>> that can be booted on two nodes.
>>>
>>> The heartbeat solution would move the ip between the machines if one
>>> goes down and activate the application on it. And one can move the
>>> ip back if the master comes to live and database has been synced
>>> again.
>>>
>>> But the simplest solution is manual fail over.
>>
>>     Also, application code will need to be replicated - plus, 
>> replication of
>>     the DB needs to be tested in both directions, after the slave 
>> replica is
>>     promoted to master.
>>
>>     If we know the application's name, we might be able to come up 
>> with some
>>     useful advice about others who may have done the work before.
>>
>>     Alternatively, use lxc/linux-vserver or jails as lightweight 
>> virtualization
>>     combined with regular replication of the content from node A to 
>> node B (the
>>     entire machine).
>>
>>     Cheers,
>>     Phil
>