On Fri, Aug 17, 2007 at 04:19:57PM +0200, Hannes Dorbath wrote:
> On 17.08.2007 15:59, Tom Lane wrote:
> >On the other side of the coin, I have little confidence in DRBD
> >providing the storage semantics we need (in particular guaranteeing
> >write ordering). So that path doesn't sound exactly risk-free either.
>
> To my understanding DRBD provides this. I think a discussion about that
> with the DRBD developers would be very useful for many users searching
> for a solution to replicate PostgreSQL, so I'm cross posting this to
> DRBD list. Maybe you can make clear in detail what requirements
> PostgreSQL has.
It does, AFAIK, if yuo configure it properly. I think it's the "protocol"
parameter you need to set to C which is the slowest, but it's the only one
that waits for the block to hit *both* disks.
//Magnus