[rabbitmq-discuss] Queue replication (I know, I know)

Mon Oct 18 17:28:06 BST 2010

On 10/18/2010 09:39 AM, Simon MacMullen wrote:
> I don't think there's really a reason for having RabbitMQ duplicate file 
> writes - the Windows software RAID or hardware RAID or a SAN can do 
> that. Real active-active is of course another story.

Let me clarify things a bit here. These are retail locations with
commodity PCs running Windows XP. There is one SOHO-class server at each
site and a handful of desktop machines in various roles. The server does
have hardware RAID, but the scenario we're trying to provide recovery
for is when the server itself dies. This could be for any number of
reasons, including people spilling drinks in the power supply (this has
happened, multiple times). The people at the sites are *extremely*
non-technical, so we're looking for (ideally) automated failover
solutions, or when that doesn't exist we would at least like a setup
that would allow our help desk to remotely bring up a service on another
machine in the site so that they can continue to function. We're not
looking for clustering, or HA even. We just want the queue and config
data to get copied somewhere else so that in the event that the server
is fried, we haven't lost it. Like I said, DRBD would make this
stupid-simple, but it doesn't exist for Windows (I don't know of
anything like it for Windows). The next best thing as far as I can tell
would be to simply have RabbitMQ write everything twice. If anyone on
the list has a better idea (outside of purchasing equipment) I'm all ears :)

Thanks,

Derek