webPage context emitting many ids?

For engineers GCP pipeline
1

Hi,

In our Google setup we are getting huge blobs in the webpage context which causes the our page_view table to become extremely big (as we are unnesting the webpage context we get duplicates).

The picture below shows three events that happens in the same session. It looks like for each event a new id gets appended. Would it make sense to always return the latest value? I think I need some help in understanding what is happening here.

The effect of this is that our page_view model gets extremely big and throwing many duplicated page view events since it recognizes several UUIDs in the webpage context

image
image1094×990 75.3 KB

Any guidance is much appreciated!

2

That’s certainly not the expected behaviour. A few questions to dig into it:

  1. What version of the tracker are you using?
  2. What table is in your screenshot?
  3. What does the web page context look like in the atomic events table? are there multiple values in there?
1 Like
3
  1. js 3.1.1
  2. the atomic table
  3. each event row has several values in the web page context
4

Hey @brajjany, a couple further questions from me as I haven’t seen this before.

  1. Could you share your tracker initialisation code? i.e. your newTracker call?
  2. Any other tracker calls you make? This might make it clearer what could be going on here.
  3. Are you using the JS Tag Tracker, or the NPM Module Browser Tracker?
  4. Lastly, what sort of application are you using the tracker in? React, Angular? Deployed with a Tag Manager?
1 Like
5

Hi, sorry for the late reply Paul!

  1. I currently dont have the script but its an inline js tag. Quite basic initialization.
  2. Only tracker installed is Snowplow if this is what you mean
  3. It’s written directly as an inline js, so not the browser tracker is used
  4. The application is created in a CMS. So a ‘regular’ site with a DOM. No SPA setup.

Will get some more information from the dev who takes care of the script.

6

Nothing immediately springs to mind unfortunately @brajjany

I think we’d need to see the initialisation code and an example of an event track which leads to this behaviour, I’ve certainly never seen anything like this before.

I’ll ask a couple of colleagues too in case they have any ideas!