Hi Dai - I will do my best - after reviewing the thread I am a little unclear at what the particular issue is - we have a playlist in our Help YouTube channel that introduces multi-node have you taken a look at this:
if so and you are still having a problem, can you please provide me with specific bullet points at what the problems are, possibly even include screen captures if necessary. This info will help guide us better towards a solution.
Hi Mike, thanks a lot for your answer. I'll Try to be as clear as I can.
We have a multinode site with several components:
1. Primary Node (all services, purpose: Production)
2. Secondary Node (all services as well, failover candidate, purpose: Production)
3. Consumer Node 1 (only Repository, Engine and Proxy services, purpose: Production)
4. Consumer Node 2 (only Repository, Engine and Proxy services, purpose: Production)
5. Development Node (all services, purpose: Development)
We're aiming to:
* Having only Consumer Nodes to take care of displaying dashboards.
* Having only Primary and Secondary Nodes to take care of scheduling and reloads.
* Having everything developed on the Development node.
Default Virtual proxies are setup as follows:
* Primary and Secondary: Load balancing nodes --> Primary and Secondary.
* Consumer 1 and 2: Load balancing nodes --> Consumer 1 and 2
* Development: Load balancing nodes --> Development
Now what we need:
One and only one url for everybody to access to the site, even when services are down on Primary, since we have a failover candidate on Secondary. If clients access to hub (dashboards), Consumer nodes should respond. If we access to qmc then Primary and Secondary should respond. Through that url, adding /dev for example, to access to Development node.
Is any of this possible with only Qlik Sense? or should we buy an external network balancer?
Can we configure Virtual Proxys and whitelists, etc in order to access only one address for the whole site? In this case, should we configure extra Virtual Proxies than default ones?
Thanks a lot for your guidance and help!!
You *can* do all of this with a single proxy, however it has been my experience that implementations with this many servers and are aiming for the availability you are looking for has a external load balancer in front. Steps:
- You need two virtual proxies:
- the default on serving /hub and /qmc - connect this to consumer 1 and consumer 2 engines
- /dev - connect to dev node
you can LINK those two virtual proxies to the Proxy on Consumer 1 at a minimum. You can also LINK those two virtual proxies to Consumer 2 as well. If you do not have an external load balancer, you can have qlik.company.com DNS point to Proxy 1, however if Consumer 1 fails, you will have manual steps to point qlik.company.com to Consumer 2 proxy.
With a load balancer then, qlik.company.com points to the load balancer and load balances to Proxy 1 and 2 on Consumer 1 and 2. From there the virtual proxies handle the routing to the right nodes / engines.
Hope this helps -