Conversation analytics used to work before, but from last two weeks we are getting the below error for all conversations, since we don’t have tech support purchased, unable to reach out to anyone for what looks like a platform issue.
Examining the Data Request \n\n
I'm starting with the user's intent: to learn about the data. The initial step is to look for relevant explores. This should give me a solid foundation – a map of available models and explores – which I'll then use to fulfill the request. I'm keen to see the structure revealed.
Considering the Failure I've hit a snag. The function explore schema discovery threw an INTERNAL error. The immediate next step is clear. I'll attempt a retry, hoping it was just a blip. If the second attempt fails, I'll have to inform the user about the persistent issue.
Rethinking My Approach I'm now recalibrating. The persistent INTERNAL error in explore schema discovery throws a wrench into my plans. My initial strategy to relay information based on the schema is now impossible. Considering the "no context" reality and the user prompt, I must now focus on communicating the tool failure. I'm going to make one final call to explore schema discovery.
Hi @Elango_Balusamy This looks like a platform issue, not a configuration problem. The INTERNAL error during explore schema discovery means Conversational Analytics cannot read your model metadata. Since it fails for all conversations and worked before, it is unlikely to be caused by user permissions.
First confirm that your database connections work and that explores open normally in the regular Explore UI. If everything works there but Conversational Analytics still fails, it is most likely a backend issue. In that case you need to report it through Google Cloud console or your Looker account contact, as it cannot be fixed locally.
Hi @marout, There are a lot of these errors happening in the conv analytics - “I’m still learning and can’t help with that. Try a different request.”
What solution will be implemented for this or what needs to be done on our side? - It is giving this response for simple questions as well.
Hi there! The “I’m still learning and can’t help with that. Try a different request.” can be indicative of multiple errors under the hood. We’re hoping to make the agent message more clear soon!
In the meantime, you should be able to see the error if you expand, as shown below. Internal error usually means it’s something on our side - we’re continuously working to fix the errors behind the scenes.
So basically your conversation analytics were working fine, and now suddenly every request is throwing that internal error during schema discovery… feels like something broke on their side, not your setup.
I faced something similar a few months back. In my case, the issue just started randomly and retries didn’t fix it either. Turned out it was a backend glitch from the platform.
What worked for me:
Try running a very basic explore / fresh query (sometimes cache gets weird)
Switch project / model if you have multiple (just to test)
Wait it out for a bit… mine fixed itself after a few days
Not sure if it helps but even while working on my Dooflix data stuff, I’ve seen these random internal errors come and go.
Are you seeing this across all projects or just one?
