Intermittent "No Connection" Message When Opening Document Via Ajax Client With Clustered QlikView Servers

Users may experience intermittent issues opening documents using the Ajax client from the AccessPoint page, resulting in a "No Connection" message/dialog.

This article outlines steps to take in a clustered QlikView Server environment experiencing the above symptom, which could be a result of corrupt TicketData.pgo or .meta file.

Another potential explanation of the message is the QlikView Server service(s) have run out of physical memory.

The TicketData.pgo or .meta file has become corrupted or out of sync, and a new version needs to be created.

There is nothing in the file that will cause a loss of anything by recreating new versions. You may also notice in the QlikView Server Event logs that one server is creating a ticket and the other server is attempting to consume it but the ticket cannot be found.  The reason is the 'shared' TicketData.pgo file is not updating correctly so the clustered servers will not have access to all tickets generated, or the .meta file may contain old server references which may prevent the application from opening properly. 

The most likely cause of the QlikView Server running out of physical memory would be an app that has suddenly increased in size due to a change in the data loaded to the application, or perhaps an expression that is creating a calculation condition that is using far more resources than usual, but there will need to be further troubleshooting of the environment to potentially determine which app(s) are in play and then review things.  There has been an Ideation request submitted to have the QlikView Management Console show real-time the memory footprint of each application in the environment, which would allow the administrator to more quickly determine which app(s) may be outside their baseline expected footprint, but we do not have confirmation of when or if this will be accepted and included in a future release.  

Follow the steps below to confirm TicketData.pgo file is working properly:

1. Go to C:\ProgramData\QlikTech\QlikViewServer and confirm the TicketData.pgo file timestamp is updating with each user login.
2. Go to the location of your QVServer Root Folder path as specified in the Management Console under System\Setup\QlikView Servers\QVS resource\Folders tab (default is C:\ProgramData\QlikTech\Documents). Verify the copy of TicketData.pgo is also updating with each user login or use NotePad to open the .meta file associated with the qvw file in question and look for old server names that are no longer part of the environment.
3. If either of the above TicketData.pgo files is not updating (check timestamp), stop the QVS services on each server running in the clustered configuration, or if you have old server names in the .meta file, simply rename the .meta file(s) in question and allow new ones to be created.  Be sure to reset any document preloading on specific cluster nodes etc. 
4. Remove all copies of the TicketData.pgo files.  
5. Restart the QVS services. 
6. Verify all copies of TicketData.pgo are now updating as expected or that a new .meta file has been created.

For the secondary issue of running out of physical memory, restart the QlikView Server service(s).