topic Re: Failover candidate in Management & Governance

Failover candidate

vajid4525 — Fri, 15 Nov 2024 22:28:03 GMT

Hi Team,

I need your help to clarify a query on Failover candidate.

As far as I understood a failover node should be a scheduler node also, else if all services are enabled, tasks will start running in that node which can't be restricted without any additional custom load balancing rules. Even though proxy service is enabled in failover node, end users can be restricted to another node by virtual proxy load balancing. .

Please let me know whether my understanding is correct or not.

Thank you

Re: Failover candidate

Chip_Matejowsky — Thu, 13 Oct 2022 14:38:37 GMT

Hi @vajid4525,

For a node to be a failover candidate, it must run the following services:

Qlik Sense Repository Service
Qlik Sense Engine Service
Qlik Sense Proxy Service
Qlik Sense Scheduler Service

If the failover candidate becomes the active central node, it will also automatically become the manager scheduler. Because the scheduler service must be active, the failover candidate will function as a worker scheduler and will receive reload jobs.

Correct, if you have VP load balancing rules in place, you can restrict end users to specific nodes.

See the following Qlik documentation from the online Help and Knowledgebase:

Best Regards

Re: Failover candidate

Steve_Nguyen — Fri, 14 Oct 2022 12:16:16 GMT

For Replicate install on Cluster, the cluster need to be Active/Passive .

Re: Failover candidate

Giovanni_Civardi — Thu, 09 Nov 2023 10:09:54 GMT

My 2 cents on this, as I'm facing that situation these days...

In an typical enterprise-scaled scenario, I tend to have dedicated Central / Failover nodes and use them just for management purpose (= I do not want reload task being started on failover central node)

Now, with default LB Rules you cannot exclude the failover central node from running reload task

Something like isFailoverCandidate = "true" would be useful for this purpose, but unfortunately is not a valid condition for a node.

To achieve this I create I Custom Property @isFailoverCandidate with value True and assigned this to both central and failover. Then I created the LB Rule and configured the Active Central as Scheduler Manager. In case of failover, the LB Rule will exclude the new failovered central node from reload tasks.

Best,

Re: Failover candidate

TusharDev_Qlik — Fri, 09 Feb 2024 02:24:33 GMT

@Giovanni_Civardi can't we set it failover node as only Manager and not (Manager and Worker) similar to Central so that it will only be used to manage reload tasks and not run them.

Re: Failover candidate

TusharDev_Qlik — Fri, 09 Feb 2024 02:28:31 GMT

Thank you for your solution here. Only one doubt I have is how do we acheive the QMC url failover? When the central server goes down, the fqdn used to access qmc will not be functional so how can we manage same fqdn for both Central and Failover node so the end user will not face any issue if central goes down.

Re: Failover candidate

anily86 — Wed, 30 Oct 2024 13:58:29 GMT

I am facing the same problem with URLs.

Can someone assist me please ?

Re: Failover candidate

anily86 — Wed, 30 Oct 2024 14:31:48 GMT

I have done failover setup and tested by failing the central node manually to take the control of FailOver which is working fine.

But one last task I am trying to achieve here is URLs.

For example:

Central node URL is - www.centralnode.com

RIM node (FailOver) URL is - www.rimnode.com

Postgredb is sitting on a different standalone server (not with Qlik Sense Central) in Linux maintaining by DBA

In My Scenario -

Before Fail over users will be using - www.centralnode.com/hub

Now I am mapping the IPs of Central to RIM node (Fail Over) so users will be using same URL www.centralnode.com/hub without changing the URL. The moment I switch the IPs QMC is going down.

Though I have copied the Central Certficate to RIM node still it did not work.

Any suggestion where I am missing and what I am missing.

Any link or document for Automatic Failover to work ?

Your help is much appreciated and thanks in advance.

Thanks,

Anil