Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi,
Recently we tried to upgrade our Qlik Sense from May 2023 to August 2023 (IR). The upgrade went well for DEV and QA servers. But it failed on production servers. We had to restore production servers with backed up AMI. Now we are investigating the reason why it failed. We have realized that we made a mistake when upgrading QS production which has two nodes: We stopped services on both nodes and then ran installation on two nodes at the same time. We know it is incorrect. But why did it cause problem? I am writing a report and need to gather deeper information.
Based on installation log files, I can tell the failure occurred at ServiceDispatcher step. Error occurred when it tried to create schema 'qlik_notifier_service'. I checked the log file for the successful installation on DEV server. When it tried to create schema 'qlik_notifier_service' there, it skipped this step because "Schema 'qlik_notifier_service' already exists" (the sentence in log). Why did the installation not skip it during production installation?
The attached ServiceDispatcher_installation_logs.txt contains related content from log files.
We need to make sure there is nothing else which caused installation failure. Please advise.
Thanks
Hi @Bo_PT,
Thanks for reaching out to us.
I would request you to go through the below article and see if you can resolve the issue.
Qlik Sense Enterprise on Windows Upgrade Failed due to MobilityRegistrarService or qlik_notifier_ser...
If the issue still persists, then please refer the below article and provide the complete set of Installation Logs.
How to troubleshoot failed Qlik Sense installations
I hope it helps!
Best Regards,
Deeksha
Technical Support Engineer
Thanks for your reply. We have already restored the Qlik Sense server. But my question is: why would it fail if we do installation on two nodes at the same time? Is there other reason which caused the failure? I understand the installation on node 2 (rim node) would fail because it could not connect to center node. But would the installation on node 1 (center node) fail too?
We will do another upgrade in later weeks. But first we need to know the reason why it failed on first try.
I know it would be better to attach those two ServiceDispatcher log files. But they are too big to attach.
Thanks
Hello @Bo_PT
If you have more than one node in a site, you must upgrade each node separately.
Each node in a multi-node site must run the same version of Qlik Sense.
You should upgrade the central node first.
Generally, in any failed upgrade it is always recommend to go through the installation logs and identify which component as failed.
In this case, it is the service dispatcher with the error: ERROR: schema "qlik_notifier_service" already exists so upgrading both the nodes at the same time might have caused this issue.
Refer Considerations about multi-node deployments
If this comment clarifies your query then please mark it as accepted solution to give further visibility to other community members.
Have a good day😊
I have one question about 'All nodes should be offline when you upgrade'. I know all Qlik services must be stopped before upgrade. But should rim node be removed in QMC before upgrade?
You should stop the services from all the nodes and no need to remove the nodes. Thank you!