Reporting off QSDA collections

QSDA
#1

Hi Rob- Is it possible to get the details of collections for the app’s so that we can report off of that?
Is it stored some where?

Thanks,
John

#2

Hi John,
When viewing an analysis, you can use the “Export Analysis to CSV” function in the Tools menu.

-Rob

#3

Thanks Rob, I guess you mentioned about repository in our call last time.
Are these information stored somewhere so we can use qliksense to report on such as which app’s were analyzed, when they were analyzed, resolved flags etc.,

Thanks,
John

#4

You can use the endpoint
http://qsdahost:5000/api/v1/collections

to get a list of collections and their status. If you want to process that directly into Qlik Sense you should be able to load with the qlik REST connector. But I just tested and parsing failed. I’ll look into it and post an update here.

-Rob

#5

Thanks Rob. The link that you provided isn’t working
Rather, when I click on a single collection and do an export, the url is like below.
Is it possible to get the complete list?

http://qsdahoost:5003/api/v1/admin/51898534-a926-43-9aed-9c4391c/export?

So when I look at a single collection, the information is very less (for eg., the performance flags are not available)

As you mentioned, please look into the REST connector and let me know.

Thanks,
John

#6

Hi John,
In the url you provided, did you change the hostname from qsdahost to your host? Try using “localhost”.

There are a number of exports available depending on your needs.

  1. The “Export” available from the Admin row menu (the one you posted) is for exporting the raw collection data. The export file is used to move a collection from one QSDA instance to another. The export file is not meant for loading elsewhere.

  2. The “Export Analysis as CSV” from the tools menu, is used to export a complete analysis, including the Flags.
    image

  3. When viewing the Admin → Collection List, you can use the grid Export menu to export the list to excel or csv.

Please tell me a bit more about your use case and I’ll try to provide you with details or potentially modify the product to support your case,

-Rob

#7

Sorry, I posted the wrong REST url for collection list. It should be:
http://localhost:5000/api/v1/collectors

-Rob

#8

Hi Rob - Thanks for all the information. Very helpful.
The use case is, we wanted to have an QSDA report to include in the production deployment process for an given qliksense app.
The QSDA app provides all the information for an given app in a tabular format with summary, Flag information and what not. This qsda app will also have info from other app (eg., App metadata analyzer)
So before publishing an qliksense dashboard to production, this qsda app will be an gating mechanism checking all the threshold and ensure that the dashboard don’t have much flags. If there are flags, then the developer cleans up the dashboard, runs the QSDA again, validates it and then publish the dashboard to production.
So this is the overall idea.

Please let me know your suggestions Rob.

Thank you,
John

#9

The error in the Qlik REST parsing is a Qlik Issue, Auto-Detect incorrectly guesses the data to be XML You can fix this by turning off auto detect in the Connection def.
image

And then in the Select Data dialog, select JSON in the Response Type dropdown.

image
image935×350 61.9 KB

-Rob

#10

Thanks Rob, the rest connector works. Is it possible to get the flag count by category using the rest?
I think we can also get the historical data as well.

I hope you saw my usecase.

Thanks,
John

#11

You can load the flags for an analysis with
http://localhost:5000/api/v1/collectors/collectorId/views/flags

The Qlik REST connector allows a WITH CONECTION(Url xxx) parameter so you can vary your script without modifying the connection.

You can explore and test the available QSDA API on your QSDA instance at:
http://localhost:5000/swagger/index.html

#12

Thanks Rob, will take a look.
In the mean time, I have an quick question.
We have analyzed around 75 app’s and today all of a sudden everything shows as - Blue badges( indicate a count of in-process collectors, or collectors that have been evicted from memory but are available in the QSDA repository.)

Do you know the cause for this?

Thanks,
John

#13

If the services were restarted all the collections would be in an unloaded state. You just need to press View when you want to view one of them.

If you did not restart the services intentionally, they may have auto restarted after a crash or machine reboot.

#14

Ok, Thanks Rob!
I assume, Load and then view. (Please see the attached)
So the report that I have from the endpoint lost all of the flag count. Is there a way to load all the apps again in one shot if unload happens again ?

Thanks,
John

QSDA
QSDA1276×618 49.1 KB

#15

Yes, Load and then view. There is not a single endpoint to load all the apps, but you could loop through a list and load.
/collectors. – will give you the list, those with status = “Collected” are not loaded.
id/load – load a collection
id/evict – unload the collection when done.

I think this is a fairly “heavy” way to get what you want, due to the way I’ve built it. One challenge is going to be waiting for load to complete. Note that as of v1.5.1, flagCount is a property of collectors. So for anything analyzed (or loaded) from 1.5.1, you should see a total flag count in /collectors without having to load. I can expand that count to include counts by category if you want.

-Rob

#16

Thanks Rob. If you can provide us with counts by category that would be really helpful!

#17

Hi Rob - Do you have an quick example to use WITH CONECTION(Url xxx) to get the flag details for all the collector ID’s?
Thanks,
John

#18

Hi John,
Attached is a QVF example of using REST to load the collectors, filtering out those that are currently loaded (in memory) and loading the flags for the in-memory analysis.

-Rob
QSDA Collection.qvf (180 KB)

#19

Thank you much Rob!! Will take a look.

#20

Hi Rob - Happy New year!!
A quick question. Is it possible to bring Qliksense App published date in QSDA?

Thank you,
John