2 Replies Latest reply: Feb 7, 2018 10:47 AM by Mónica Alejandra Yedra Almazán RSS

    Server (failover) Qlik Sense don't start

    Mónica Alejandra Yedra Almazán

      Hi,

       

      I'm doing redundancy tests, with the central node and the second node (failover).

       

      The test is the following:
      Download services from central node and activate the second node (failover)
      Restart central server and activate the second node (failover)

       

      When we perform those activities and want to enter the hub (through balanced IP) send error: Internal Server Error o The web site is not available

       

      Change the timeout of the Failover TimeOut: 1 minute

      But we still can not see the second node coming into action, the second node configured as failover never takes action

       

      Specifications:
      1 Node Center (Active)
      1 Node Failover (Master 2) (Pasive)
      2 Proxys
      2 Scheduler
      1 Storage

      S.O Windows Server 2012 R2
      Qlik Sense Junio 2017
      Load balancing with F5

       

      Help!

        • Re: Server (failover) Qlik Sense don't start
          Levi Turner
          • Where does PostgreSQL live? Is it alive during the outage of the Primary (Central) node? It needs to be.
          • Where do the apps live? Is it alive during the outage of the Primary (Central) node? It needs to be.
          • How is failover occurring? Using Qlik Sense natively doing this? Or via API calls? (Reference https://www.youtube.com/watch?v=Y5IblMGwrLU)
          • What virtual proxy is in use in this test? What kind of authentication is in use? Stop the Proxy service on the Central and you'd get an accurate test of what occurs with the Central node being down if the Central node proxy is in use on the front end.

           

          I'd personally break things down to basics.

          • Failover without taking services down via API
          • Validate
          • Confirm the log signature in Log > Repository > Trace > System_Repository of the failover event
          • Then do the test of automatic failover (ideally using the default 10 minute value rather than 1 minute)
          • Validate
          • Adjust the timeout value via API then test the automatic failover

           

          Cheers.

            • Re: Server (failover) Qlik Sense don't start
              Mónica Alejandra Yedra Almazán

              Hi Levi

               

              I will try to explain, an apology my English is not very good:


              • WHere  does PostgreSQL live? Is it alive during the outage of the Primary (Central) node? It needs to be.

              postgresql is in the central node.

              How can I validate if it is executed correctly?

              • Where do the apps live? Is it alive during the outage of the Primary (Central) node? It needs to be.

              the applications are in a 'storage (isilon)'


              Using Qlik Sense natively


              When I try to make the api call, I frame certificate error or other error. I've never really done this, and I do not really know what I can be doing wrong (attach image) 9740D38F-FA67-4430-8C04-073D22F429B6.jpeg


              • What virtual proxy is in use in this test? What kind of authentication is in use? Stop the Proxy service on the Central and you'd get an accurate test of what occurs with the Central node being down if the Central node proxy is in use on the front end.

              I already did this test, when I turn off the proxy service of the central proxy nodes take that role and allow me to continue entering the hub