QV Scalability Tools (QV10,QV11,QV12)

[Daniel_Larsson]

This package (referred to as QVScalabilityTools) contains a complete set of tools for easy creation, execution and analyzing of load/performance tests and regression tests.

QlikView documents to help analyze result and log files(This was previously included in this package):Scalability Results Analyzer

Older versions of the tool can be found here: Tool for easy creation of load/performance tests of QlikView (v.10 and 11)

Tools for Qlik Sense can be found here: Qlik Sense Scalability Tools

Note: Using windows built-in unzip functionality will not allow the .exe to launch unless unblocking zipped package first (properties -> unblock).

Note: Regarding Apache Log4j, reference CVE-2021-44228 (also referred to as Log4Shell), these set of tools are not directly vulnerable to the exploit, however JMeter, the tool required as a separate download, is. How to mitigate the vulnerability for JMeter has been documented here : https://community.qlik.com/t5/Knowledge/Qlik-Scalability-tools-and-log4j/ta-p/1872009 

Included parts are:

• Standalone application for creating a JMeter script
• Support files for launching the script by utilizing the JMeter engine (JMeter installed is a pre-requisite)
• Documentation on how to use the package

Version 1.2.7->1.2.8

• Changes Section Access requests

Changelog 1.2.6->1.2.7

• NTLM support removed due current solution broken and lack of a new secure solution.
• Small updates for static content, for future product releases.

2019-08-26

Documentation updates due to:
- New recommended JMeter version (4.0)
- Changes to the licensing for the Oracle Java implementation (seehttps://www.oracle.com/technetwork/java/javase/overview/oracle-jdk-faqs.html)
- Open-source Java Development Kits (JDKs) have been verified for use with JMeter and the QlikView Scalability Tools

Changelog 1.2.5->1.2.6

• Added support for QlikView 12.50 and 12.20SR10

Changelog 1.2.4->1.2.5

• Added support for QlikView 12.30

Changelog 1.2.3->1.2.4

• Added support for JMeter 4.0 and Java 9.
• Bugfixes for regression testing
• Added actions: Select Excluded, Select All, Select Possible
• Added support for 12.20 SR3

Changelog 1.2.2->1.2.3

• Bugfix for export to excel action
• Added transfer state (bookmark) support for document chaining
• GUI: Template selection more accurately show supported Qv versions

Changelog 1.2.1->1.2.2

• Bugfixes, make sure minimum amount of selections are respected in certain corner cases

Changelog 1.2->1.2.1

• Improved performance for high throughput scenarios

Changelog 1.1 -> 1.2

• Improved assertions logic to show faulty actions as errors
• Change JMeter template scripts to use groovy instead of beanshell
• New prerequisites:
  • JMeter version 3.0
  • Java 64-bit JRE 8

Your use of QVScalabilityTools will be subject to the same license agreement between you and QlikTech for QlikView Server. QlikTech does not provide maintenance and support services for QVScalabilityTools, however please check QlikCommunity for additional information on use of these products.

PLEASE NOTE:

The supported versions of Java are 8 and 9. Supported version of JMeter are 3.0 and 4.0, but not 3.2

[jerrysvensson]

Thanks

We are really looking forward to using this.

[#3]

Hey just wondering we are using NTLM Auth and using NTFS Security, Is there anyway possible to test memory,  I can have 20 concurrent users on an app but the memory does not change .  however CPU Spikes high because it is intensive.  Thanks.

[Sebastian_Fredenberg]

Hi,

I can think of a few reasons why you don't see memory rising.

• If the memory has already grown to working set min (normally at 70%) then it will stay at that limit.
• The selections made by the simulated users are all cached, more randomization will mitigate this
• The app or expressions does not require much of ram so it looks like it is hardly growing

Spiking CPU for some cores is normal for any calculation - but note that if it is just short spikes then it is not necessarily intense. So to get the full picture of how CPUis loaded that it needs to be looked at for a longer duration, taking the average over all cores.

My best bet if you do have high average CPU, is that the selections you simulate are already in the cache. A good idea is also to check for any errors in the simulation - if users cannot get in then nothing will be selected.

cheers

/sebastian

[#5]

I think its closing the connection since I am only using one user.  Authentication is NTLM and Security is NTFS.     I believe the error is Killed because Named User Cal was needed from another client".

[Sebastian_Fredenberg]

Hi,

That explains it then - Named User Cals & NTLM is not a good match for simulating users with the tools. From the documentation:

"The default JMeter behavior is to connect to the QlikView Server with the NTLM-user that started it. This will work in some cases but Named CALs does not accept multiple sessions with different session-cookies for the same username and QlikView document. Ensuring that other licenses are used instead might solve this, but there are other options as well. Configuring to use DMS-mode and header-authentication is an option. The test scenario must then be configured to use Header-authentication, with the usernames in a text file. This will ensure that unique users will connect and as long as each user is assigned a CAL the simulation will work."

So with that, if you can use header for the simualtion, I strongly recomend that.

/sebastian

[#7]

Are there any drawbacks or things I need to look out for changing from NTFS TO DMS and NTLM TO Header.  I understand there is more flexibility using this method but does the conversion process require a lot more maintenance ?

[Sebastian_Fredenberg]

Hi,

No I would not think so, My suggestion would be to change to header/dms in order to run the test and then change back to your preference once done with that. That would mean no maintenance.

/Sebastian

[#9]

Just wondering on the format of giving header information to the qlikview scalability tools how would that work and if I do decide to change it to header information how would I go about doing maintence on that on the qlikview side or the jscript side  . what credentials do I need . Thank you so much for your help and sorry for the bombarding of questions

[Daniel_Larsson]

I would suggest that you use header authentication only during the testing. To use header auth you need to change both the loadscript and QlikView. For changes on the QlikView side please refer to the QlikView documentation, the configuration on the tool side is well described in QVScalabilityTools.pdf which can be found in the documentations folder.

[#11]

Hi All,

I am using QlikViewTestScriptGeneratorGUI for regression testing on two different versions of my qlikview document. The only difference between the 2 documents is the addition of extra text box, in the latest version. I have created and run scenarios for both the versions and on comparing the 2 results it says 0% difference. How is this possible when an additional text box is added?

And

What does the percentage difference signifies?

And

When will I see a difference in the compared results?

And

Can I also compare the data in the tables and input box using QVScalability tool?