Telemetry: investigate why demo.kedro.org is not flowing to Snowflake

[Huongg]

Description

At the moment, we’re not seeing any data flowing from demo.kedro.org to Snowflake, even though it appears to pass through Heap. This issue is to investigate the cause of this issue.

Checklist

• Include labels so that we can categorise your feature request

[rashidakanchwala]

Weird, I can do this, when I use demo.kedro.org, and turn on live data feed.

[tynandebold]

Adding another data point: Heap does send events from the demo.kedro.org page.

[Huongg]

hmm so it seems like it does send to heap, but somehow nothing shows in snowflake. As i just see the data from the last 2 months, and they are mainly just localHost or github page. Any idea why this would happened @rashidakanchwala @tynandebold 🤔

[rashidakanchwala]

hey I think this is just the data preview, and it doesn't show all the rows. Maybe you need to write filter query to see when landing page is 'demo.kedro.org'

[Huongg]

hey i think @astrojuanlu also confirmed that he couldn't see the demo page from the landing page when he did the query #2140

I'm not seeing https://demo.kedro.org/ in the top list of landing pages, which is interesting to say the least:
LANDING_PAGE,COUNT
127.0.0.1/,54765
localhost/,1822
127.0.0.1/experiment-tracking,357
anasaito-alpha-lineage-x6qqx4593p59q-4141.githubpreview.dev/,299
deepyaman.github.io/jaffle-shop/,162
172.19.113.195/,139
0.0.0.0/,108
133.127.13.9/,108
4141-cs-697995841852-default.cs-us-central1-pits.cloudshell.dev/,107
4141-cs-200f5089-bab4-4372-bef8-0b4fd0d92189.cs-us-east1-pkhd.cloudshell.dev/,65

[rashidakanchwala]

@Huongg , i think in that case we need to open an issue to investigate why isn't the heap data reflected in snowflake and close this one, or update it correctly.

cc - @astrojuanlu

[Huongg]

yup i can update the description on this ticket if it's easier?

[astrojuanlu]

One surprising thing we're observing is that there's lots of pageviews associated with demo.kedro.org ✔️ but zero "interaction events" 🤔

[rashidakanchwala]

@Huongg , @astrojuanlu

Could you confirm if the table missing demo.kedro.org details was Viz_interactions_table? I’ve identified the reason for the missing information: the event was set up to track clicks on elements with data-heap-event, but demo.kedro.org uses data-test attributes instead, as it’s the latest version of Kedro.

By default, Heap generates a generic click event for every click. In my opinion that 'Generic Click' Heap table is the best to follow, and we can use the hierarchy variable within it to identify which specific event was clicked.

let me know what you think? Here's a screenshot of what viz_interactions_table tracks

[Huongg]

hey great work @rashidakanchwala , i can confirm the table is called VIZ_INTERACTION_EVENT_2 in snowflake. and I can now see your new sync table called KEDRO_VIZ_USER_INTERACTIONS_GENERIC_CLICK_EVENT, which is what we need ❤️

[rashidakanchwala]

For the GENERIC_CLICK_EVENTS table, we need to create two custom columns, either in Tableau or Snowflake, as follows:

Extract Data Heap Events (on Tableau) :

REGEXP_EXTRACT([Heap Hierarchy], "\[data-heap-event=([^;]+)\]")

Extract Data Test Events (on Tableau) :

REGEXP_EXTRACT([Heap Hierarchy], "\[data-test-event=([^;]+)\]")

The union of these columns will capture all relevant events, though the naming conventions may vary and require some cleanup. We can also consider @astrojuanlu 's approach by focusing specifically on newer Kedro-Viz versions, using only data-test events for consistency.

also cc : @lid-rs

[rashidakanchwala]

@lid-rs , @astrojuanlu -- let me know if this is fine, and I can close the issue. I looked more into details - the above and the data is same as HEAP.

[astrojuanlu]

Amazing, thanks @rashidakanchwala!

Can we then unsync VIZ_INTERACTION_EVENT_2 and focus on KEDRO_VIZ_USER_INTERACTIONS_GENERIC_CLICK_EVENT?

About creating custom columns,

Creating a dynamic column in Tableau is a good quick and dirty solution but eventually the Single Source of Truth should be a "gold" table in Snowflake, see https://lakshmanok.medium.com/what-goes-into-bronze-silver-and-gold-layers-of-a-medallion-data-architecture-4b6fdfb405fc

That's out of the scope for this ticket anyway.

Originally posted by @astrojuanlu in #2140 (comment)

[rashidakanchwala]

I have done so. I reckon the tables will still exist in snowflake just not updated. This will create some noise, should we manually delete/archive/hide them or just ignore them?

[astrojuanlu]

For now let's ignore them, I'd say.

This now seems to be working:

Closing!