CQ5 Author Clustering

Hi,

As a quick start, you might want to refer to this article: http://dev.day.com/docs/en/cq/current/deploying/cluster.html and http://dev.day.com/docs/en/crx/current/administering/cluster.html#How Clustering Works

If you still find you need more information, please reach out to us and we would be glad to assist. Thanks!

Thanks Avik for the link.

The trouble I am facing at the moment is that I want to have Active - Active Author Node Replication.

 [img]cq5.png[/img]

This is the architecture of my CQ5 cluster where publisher clustering is in place.

Do I need to follow manual slave join approach for this architecture.

Yashika - CRX Clustering creates an exact replica of a node, so the two author instances in your diagram would be identical and they would both publish to the *same* set of publish instances.

This has nothing to do with manual vs. automatic slave join.

justin_at_adobe wrote...

Yashika - CRX Clustering creates an exact replica of a node, so the two author instances in your diagram would be identical and they would both publish to the *same* set of publish instances.

This has nothing to do with manual vs. automatic slave join.

 

Justin,

I am not able to find a way to cluster these two author instances. The documentation states that Author Clustering happens in Master-Slave mode but I wish to have it in Master-Master mode.

Could you please help me with the process to cluster the two author instances, after installing them on two separate nodes.

Justin,

Wanted to clarify my use-case.

I am having two Author Instances on different datacenters. I am trying to work on devising a Disaster Recovery Strategy where I need a failover support in case any of the author instances stops working due to any datacenter related disaster.

Hi Yashika,

When you run CQ 5.x (or AEM 5.6.1) in a TarPM cluster, you will always have "master - slave" relation between the cluster nodes. There is a single master and a number of slaves. That's on the lowlevel repository view. Of course you can work on all instances (it appears to a developer/admin, that all cluster nodes are equal), but deep down it's still a master-slave relation.

Second, if you want to run a TarPM cluster over multiple datacenters, you'll likely experience bad performance because of the latency. TarPM clusters rely on a stable and low-latency cluster connection (I always say: Latency less than 10 ms, better only 2 miliseconds; if you have higher latency, you'll get performance problems). With MongoDB the situation can be different, as then it's a MongoDB thing.

And: Having author and publish in a single cluster doesn't make sense, as Justin already pointed out. You have a single /content tree, but how would you split between the content, which is prepared by editors, and the content you deliver?

Kind regards,
Jörg

I am sure this config is possible. But it sure doesnt make sense to cluster an author and publish servers together.