Hi 👋 I am trying to create some simple insights in the GUI that use an attribute filter on a sum field, however every time I try to save I get the error

Failed to save insight, please try again.

If i remove the filter, the insight saves. I was not able to find anything online regarding this. Are there any known issues regarding attribute filtering?

Hello Josh, You may want to check your LDM and make sure that the object you are using as a filter is connected to the other objects being used in the metrics or attributes.

Hi Joseph, yes I believe it is. I am able to use the object to filter data and get accurate results, export, etc. However the insight just will not save.

Is there a trace id you could provide when getting that error message?

No unfortunately it doesn't give any other details, only the failed to save message.

Hi Josh, this error really does typically come if there isn’t a connection between objects in the LDM. Would you be able to share a screenshot of the LDM and highlight exactly where each object from the insight is?

All data I am using is from the same dataset, and provides expected outputs and results. Would I be able to create visuals or insights using these objects if they were not connected? Also is there a connection between objects required when the objects all reside in the same dataset?

Hi Josh, I believe you might be hitting some limitation with how your LDM has been set up. Do you really need all attributes and facts set within a single table? I will check with our product team to see if this is the case.

If an error occurs in Analytical Designer app, you can always find more details in browser developer console. At least, there should be the traceId, but if it is "expected" kind of error, you can find there a human readable error message. Please, copy&paste it here.

Apologies for the delay - here are some error details from just now attempting to use my 'Tags' object to filter: traceId: "1e9894b8c84b98c3"

wow! The error message suggests that the definition of the insight is longer than 15000 characters, so it does not fit to internal metadata storage column, which is VARCHAR(15000). We saw such an issue in the case of Dashboards, but never in the case of Insights. Can you briefly describe how many entities did you put to insight, how long are texts you manually entered there?

for this insight i only put two items on it. one is a dollar value, and the other is a list of 'Tags' (these are key:value pairs but unfortunately they come to us in unique strings that combine all tags on a resource, per resource - the tags field is quite large) we tried to separate the key:value pairs into an array type on the backend, but GD does not support that. We are trying to find a solution to separate those key:values to make it more usable. I just found it odd that I could use the field and get expected results and outputs, but I am unable to save. I have a client waiting on multiple insights from us that must utilize these tag keys.

So if I understand it well, you create a filter on these tags and select only subset, which is still very large?

The Tags object itself is very large, currently with 7099 unique values. The filtered subsets are much smaller, in this example going down to 166 values. whats interesting is that if I bring in only the tags as row values I can successfully save that as an insight, but as soon as I filter it down to the 166 values, I can no longer save.

The thing is that selected filter values are stored as a part of the metadata definition of insights. When you add tags to rows, the definition just contains the ID of the attribute, no values.

yes, just verified if I select a small number (10) I can save.

@Jiri Schmid @Peter Plochan any kind of workaround solution for this issue? Insight filter contains a lot of values, which are quite long. The insight definition is longer than 15000 characters, so the saving to our metadata fails. If there is no workaround, could we extend the corresponding field?

Only the AD is longer (

DASHBOARD_MAX_CONTENT_SIZE = 30_000

) , while all other use the default (

DEFAULT_CONTENT_LENGTH = 15000

). We can increase the size, that means update the code and modify the database storage. The only workaround could be putting the filter to the dashboard definition. On the other hand, is not such a big filter a bug ? Would be a db-computing effective enough ?

I assume this is not a question to us...?

i see, and i don't see other option then increase the size in the metadata storage, @Jan Soubusta prioritise please

Posting to internal Slack, I think this request already exists, we need to find it, do ++ and prioritize it.

🎉 New note created.