Broken event samples?


there are always a few broken events (typically 1 or 2%) in the breakdown by type whenever I check my “Actions this month” under “Account → Subscription”. And I never was lucky to find one in the Tracker debugger when I filter for broken events in the last 6 hours. It would be great if there was something like a log for the last x broken events in order to debug and fix them, even if only a few might occur. I failed to find data about broken events using the API. And even if I could access the number of broken events there, I would still not get samples of raw broken event requests.

Any idea how to catch those rare errors?


1 Like

It’s a good feature request and I’m pretty sure I’ve seen it before. I’ll file another one anyways.

If you don’t see the broken event in Tracker Debugger of any of your Piwik PRO sites then it could be that the tracking request doesn’t contain the site UUID and therefore ends up in neither of those. I think this case is also in our backlog.

When it comes to catching regular broken events that are simply appearing outside of your working hours - you could use some kind of an automation tool (e.g. n8n, Zapier, Make) to poll the debugger API every few hours.

I’ve encountered this exact case before. I’ve used n8n for data polling, extraction and transformation and then pushed a notification to Slack. Got a video tutorial on this but never got to the point of uploading it :sweat_smile:.

Not an ideal solution for sure but gets the job done.

1 Like

I just tried to quickly build something like that and used the event_type 8 for goal conversions in order to see what I can get from the stream of sessions.

But as it seems, the filter for event_type does not really work. Even if I set a limit of 1 and a filter for only sessions with goals and then list all events of the first session received, I seem to get sessions without a goal conversion? How could that even be possible?

However, if I catch a session with a goal conversion, extract the uuid of that certain event and then try to query the stream of logs for that specific event, the response is a 500.

I give up for now :expressionless:

Update: solved using the API and a scheduled GCP Cloud Function, waiting for Slack to tell me what’s wrong with those events :wink:. Thanks for the idea!