Can a failover node work as scheduling node

[rohitk1609]

Hi,

I have configured a failover node and now I need to configure one node which will handle reloading tasks, My Questions is , Can I use the same Failover node as reloading node , If yes , how ? As reloading node need to set SLAVE and if I set it to SLAVE on central and on the time when my central node stop working , will it automatically set to MASTER and SLAVE which manage and reload the tasks ?

[rohitk1609]

How would I check when I central node down, Can I open the QMC on Failover node after getting down of Central node ?

[Vincenzo_Esposito]

Yes of course you can. Since you have all service (including proxy) you can access to it

[rohitk1609]

Yes I saw we can open QMC on Rim node too.

One quick question:

I have added one RIM to CENTRAL node and created an application, added =computername() to see application is consuming which server resource, central or rim and published to a stream.

then I went to virtual proxy => Central => added my RIM in load balancing nodes

then I went to virtual proxy of RIM node and added CENTRAL in load balancing.

So now both virtual proxy has both CENTRAL and RIM

When I open my application on central node it is showing computer name CENTRAL and when I open my application on RIM it is showing RIM in computer name.

Then I disabled ResourcesOnNonCentralNodes and created a new rule and check load balancing option and filter my application id in resource filter tab and select the node as RIM

Now when I preview it is showing RIM node "B" Application

I don't know what B signifies exactly

now when I opening the application on central node it is still showing CENTRAL in computer name and it suppose to show RIM because i have written a rule for it.

Can you please help me how to run a application on CENTRAL node but consume RIM node, what I am missing ?

[Vincenzo_Esposito]

The rule "ResourcesOnNonCentralNodes"  make (as the description) "All resources need to be accessible on the non-central nodes, except the monitoring apps". This means resources will be shared across all other nodes. If you disable it you expect to use only the central node. Actually, you can't deny having resource on the central node due to  the first rule "ResourceOnTheCentralNode" which is ReadOnly therefore unmodifiable.  

try to create a new Balancing rule as:

((node.name="<RIM>" and resource.id="<APP_ID>"))

[rohitk1609]

I have disabled ResourcesOnNonCentralNodes and created the rule as you have stated but even if I open the application on master it show master and if I open the application on RIM it show RIM,

Do I need to do some configuration on virtual proxies for both central and rim node?

secondly,

Can you please tell me what if the benefit of create repository database and app storage out  of the central node in shared persistence ?

[vlad_komarov]

Vincenzo,

To add a little bit more complexity to the question rohitk1609 was asking: we need to have a system that works behind our own Network’s Load Balancer (F5).

And I need to setup 2 (at least) Proxy nodes that will handle the users’ traffic in this configuration. I do not want to separate users for “App 1” go to “Proxy 1” and “App 2” go to “Proxy 2”… The F5 has no control over Qlik and just redirects users to “next available” node in my case, so all Proxy Nodes are better be capable to handle all apps from all users…

I was not able to do it in the previous configuration (with QS 3.2) and I was hoping the Sept 2017 release is more flexible to handle that. Appreciate your help and suggestions…

Regards,

Vlad

[Vincenzo_Esposito]

To achieve your goal should be enough to set a balanancing rule as earlier defined. Check if it works looking the resources on the server or using the operation monitor. You don't need to link users to proxys.

[Vincenzo_Esposito]

You want to move db and share persistence out ou the central node to avoid any single point of failure. Of course you need to manage an HA for both (e.g. NAS/HA DB or HA VMs). In that case all the QlikSense site can still work even if the central node fails

[rohitk1609]

We are asking, what is the benefit of mode db and share location out of the central node. We already created one failover candidate. So can you please help me to know how taking DB and apps out of the central node?

what will happen if my DB or apps storage server will get down ??

[Vincenzo_Esposito]

Rohit, seems this thread is getting a bit confusing and out of the original post's subject. For the benefit of all, I suggest you to close this post if the original question has been answered and open one more specific on HA configuration if this is your real need.

Just to conclude on the topic, if the node running the DB or Share folder will fail, Qlik Sense is no longer available.