[rabbitmq-discuss] RabbitMQ Windows 2008r2 Setup

[steve flitcroft]

Hi Rob,

Not really the shared storage but the way the RabbitMq windows
installer appends a guid to the service name means there is a
difference between the nodes in the cluster therefore they do not play
nice with the Cluster and Failover manager and everything has to be
identical. I have got around this by manually creating the service
using the earl service add.
However I have moved on in my thoughts since I believe we need an
Active/Active setup rather than Active/Passive.
I think I am going to go with the use of HaProxy to load balance
requests to a windows rabbitmq cluster.


On Nov 22, 11:49 am, Robert Raschke <rtrli... at googlemail.com> wrote:
> Hi Steve,
>
> On Tue, Nov 22, 2011 at 8:10 AM, steve flitcroft
> <steve.flitcr... at gmail.com>wrote:
>
>
>
>
>
>
>
>
>
> > I am currently investigating the best way to setup RabbitMQ on our
> > infrastructure. Over the last few days I have been playing with the
> > idea of using Cluster and Failover feature of Win2008 r2 however this
> > would only give me an single Active/Passive cluster and I would have
> > to use shared storage for the rabbitmq instances and the nature in
> > which the windows services install make this a bit of a nightmare to
> > configure. Another option to enable a load balanced cluster is to use
> > NLB however this does not really give the resilience that i need since
> > NLB only listens for the NIC and not configured to ensure that
> > rabbitmq service is actually in a fit state. I really need the
> > consumers to use the same dns entry without any intelligence to be
> > dynamically switching addresses.
>
> > Could I draw upon anybody else's experience in configuring RabbitMQ on
> > a windows infrastructure. I have considered using some linux boxes and
> > ha proxy to load balance a windows rabbitmq cluster however our
> > corporate IT are reluctant to introduce different OS's onto the
> > estate!
>
> What exactly makes shared storage so difficult?
>
> Personally, I always recommend that the data for my Windows hosted apps
> reside on a network storage device. This allows for "easy" passive
> failover, log file access, and backup. I have also found that most
> companies frown very heavily on machine local data storage, mostly for the
> same reasons. The only reason to have data stored on the local machine,
> would be performance. But I've yet to see network storage be a bottleneck.
> YMMV.
>
> Robby
>
> _______________________________________________
> rabbitmq-discuss mailing list
> rabbitmq-disc... at lists.rabbitmq.comhttps://lists.rabbitmq.com/cgi-bin/mailman/listinfo/rabbitmq-discuss