QVS -Is There a way to get detailed information for each user resources consumption?

[kevinalvino]

Dear All,

As Governance dashboard can be used to track the overall views of condition of what happen in our environment, but we cannot get the detailed information for each user accessing what dashboard and how many that activities eats up the resource (Like RAM or CPU Usage ) .

For scenario example :

Let say that on the same timestamp, there was 5 Opened Dashboard , and each dashboard has 4 user accessing it at the same time.

And suddenly CPU goes up, and the server crashed.

Based on the data, we cannot pinpoint which "exact" user caused the problem to occur.

The Question is :

Is there any way to get this related information, as we already tried to check the Governance dashboard and exhaust our resource to search a way to achieve this ? Maybe we miss up several things like possible logs from QVS that can be used for the governance dashboard, or there are specific version of QVS that generated this.

Thank you for your attention,

Best Regards,

Kevin

[Miguel_Angel_Baeyens]

To get accurate data on CPU and RAM usages you need to set up performance counters in Windows, I think there is an article in the Qlik Support portal on those most used for troubleshooting Qlik issues.

You can then load that data on QlikView and compare to the errors you get from the QlikView services or application logs.

[kevinalvino]

Hi Miguel,

Thanks for the reply.

Is using the logs provided from performance counters can be used to trace who is the user causing the server to crash ? As i know ( correct me if i'm wrong), from current governance dashboard, we can know what dashboard used, and how much the related dashboard eats up the resource( because this is possible, then CPU and RAM usage are already viewed here). But, we don't know , which one , exactly, the user causing it.

So to put it simply : Finding exactly the user who causing high usage of RAM and CPU in a dashboard(s) available.

Thank you for your attention,

Best Regards,

Kevin

[Miguel_Angel_Baeyens]

Windows Perf Counters follow processes, not dashboards. The Performance logs in QlikView could help you in identifying that, but I don't think you can do it straightforward: you will need to see the timestamp with highest usage, then see who was online at the time, which documents were open and of those which one was using the largest amount of resources.

But let's wait for others to chime in. I know that Michael Terenzi created the System Monitor and Rob Wunderlich also has some monitoring tools, and they may have better ideas.

[kevinalvino]

Hi Miguel,

Thank for the reply.

I see....but the suggested idea seems take many step and need do manual checking/tracing, as it's preferable that we can use something alike Governance dashboard so we just need to reload and see.

I agree, let's see if others can help or have better idea with this problem.

Thanks,

Best Regards,

Kevin

[Miguel_Angel_Baeyens]

Ideally, the Governance Dashboard + Windows performance counters should give what you want, but as far as I know, there is no such application just out of the box. For example, the performance logs in QlikView are written, by default, every 5 minutes, although this value can be changed in the QMC. Likewise, the Session logs are written when the session finishes, either by timeout or manually closed by the user.

That's why you need to triangulate, unless someone already did already.

[Peter_Cammaert]

One technique I often resorted to after doing such an analysis is this: as soon as you know which users were working in this dashboard that probably crashed the server (argh, those @#!§!!! server objects with bad expressions...), ask them what they were doing at the time of the accident, and of those that you suspect may have had a hand in the crash, individually ask them to perform those tasks again. All the while monitoring cpu and ram load. The result will be obvious. The tricky thing is speed: if you wait too long for a quiet moment in the use of QlikView server, they may have forgotten already what they were doing...

Good luck!

[rwunderlich]

You can merge Performance and Audit logs by timestamp to get an rough idea about what was happening when resource consumption spiked or crashed.

-Rob

http://masterssummit.com

http://qlikviewcookbook.com

[kevinalvino]

Hi Peter,

Seems a good idea, but i think it is hard to implement in the site for my case, cause the user of the dashboard not only little as the example i used, but ranging dozens of user. This mean that to track a little bit hard to ask one by one the user what they are doing in the certain time range based by logs. As you already stated, the server crashed already past 1 week...and i think everybody might already forgot, and we must check from beginning again.

But, appreciate this answer and the efforts.

Thanks,

Best Regards,

Kevin

[kevinalvino]

Hi Rob,

May i know if there are any other logs beside Audit logs, that generated that also could be used along with the performance monitor counter? It will be helpful if there are multiple logs that can be used to check to make the result even more detailed.

Thanks,

Best Regards,

Kevin