Adding a schedular node to multinode environment

[ArifShah]

Hi,

We have recently tried to use of the existing nodes as a schedular. I went to the nodes and enabled the schedular service on the node (The node is already having a schedular service running in it)

In morning, all the reloads that went to this node failed. I tried to check the log script but it did not even generate the log files. I went to the engine trace logs and found this

Server: App Load: Beginning open of app bcfd5dea-8849-4a97-beb4-c0de51c158bd with user
DOC loading: Beginning load of document BCFD5DEA-8849-4A97-BEB4-C0DE51C158BD
Exception raised in AppSerializer::Load: 'int': -500	0	INTERNAL
ServerDocumentEntry: DoLoad caught extended exception 1003: App not found
Document Load: The document bcfd5dea-8849-4a97-beb4-c0de51c158bd failed to load
StdServer: OnRemoteClosed entered for socket
Session stop reason: Socket closed by client
DOC loading: Unloading document bcfd5dea-8849-4a97-beb4-c0de51c158bd, state UNLOADING

Is there something that I have missed in order to reuse this node as a schedular?

[Udo]

Hi,

I'm just curious, if this has been resolved meanwhile? I'm facing the same problem, when I just added another scheduler to our multi node environment.:
- ServerDocumentEntry: DoLoad caught extended exception 1003: App not found

followed by

-Document Load: The document 1b1fd006-ac3c-4bb4-8ee0-7350287e3893 failed to load.

I made sure the shared folder is available for the service user, and the app is also there.

Thanks,
Udo

[Bastien_Laugiero]

Hi @Udo 

Could you try to link the engine on that node to a Virtual Proxy and open the app 1b1fd006-ac3c-4bb4-8ee0-7350287e3893 from the hub? Do you also get an "App not found" error?

Also is it happening to every apps you are trying to reload on that node?

Thank you!

[Udo]

Hi @Bastien_Laugiero 

Thanks for your reply! How would I link the engine to a virtual proxy? The node itself already has a virtual proxy, so would I need to create another one? Or do I need to enable the Proxy service on that node to test this?

Thanks,
Udo

[Bastien_Laugiero]

Hi, 

If possible the simplest is to reuse the virtual proxy on that node.
Make sure the load balancing is configured with the affected node and that the proxy is attached. 

[Udo]

Hi @Bastien_Laugiero ,
we enabled the Proxy and were able to access the hub on that nod.
Anyway, when trying to open an app it's saying "The app you specified could not be found". That's happening for all apps.

It seems to be like in this article. https://support.qlik.com/articles/000037978
We went through all resolution steps except (3) and made sure they were not causing the problem. We are issuing the "max connection" proposal with our DBA. But I'm not sure if this could really cause this...

Thanks,
Udo

[Bastien_Laugiero]

Hi, 

Step 3 would be applicable if it was happening randomly which is not the case here. 

This really looks like the engine on that node is struggling getting access to the shared. 

Could you try to restart the services Qlik Sense Dispatcher and Qlik Sense Engine. The under C:\Programdata\Qlik\Sense\Log\Engine\Trace, open the log System_Engine.txt and check at line 8 if you see a line looking like below or an error:

8	20210315T085135.012+0100	INFO	QlikServer1	Mount: Found Mount  at location \\dc1\Share\Apps  browsable

[Udo]

Hi @Bastien_Laugiero ,

Thanks for your help! Finally, we managed to get the node working, by enabling the proxy functionality (for this node in the QMC) and restarting it. Do you know, if it is necessary to have the proxy running on a dedicated scheduler? I was thinking, engine and scheduler would be sufficient?

Thanks,
Udo

[Bastien_Laugiero]

Hi,

Thank you for the feedback. So you shouldn't need to proxy to be enable to run tasks from the QMC. 

I am guessing that it's perhaps the fact that you restarted the services that fixed the issue. 

Could you try to disable the proxy service and see if it still works?

[Udo]

Thanks, Bastien. We were able to stop the proxy on this node. Everything is working now.
Maybe the root cause was that, when adding the node, engine and scheduler were not activated at the same time. Not sure. When adding another node last week, we dirctly activated engine and scheduler and it worked out of the box. Thanks!