Hi -- we got a "possible ambiguous connection" err...
# gooddata-cloud
p
Hi -- we got a "possible ambiguous connection" error for some reason in a workspace, which wiped out a bunch of custom dashboards/metrics we had created. What caused this?
m
Hi Phillip, would you please be so kind as to provide more details on the steps you took to receive the “possible ambiguous connection” error?
Also, can you please provide more details on how this resulted in loosing some custom dashboards/metrics?
Can you provide us with a timeframe when this happened too please?
p
@Josh Kladis ☝️ can you provide this when convenient?
@Donald Worley CC also
j
I brought in a new data set to our model and made connections to the current data sets. I saw a warning label that said i should alter the data sets to avoid ambiguous paths. I figured I could work around any ambiguity in the insights and saved the model without error. However when I went to check my current dashboard I noticed most of my insights were not functional. I checked my insights and found that about 15 insights were missing altogether. This was about 9;30-9:45 am PST.
for clarity, The ambiguous paths were my fault and I have since rebuilt the data model correctly and recreated most of the insights I lost, but the curious part is why so much of my work was deleted regardless of any broken data connections.
j
Hi Josh, I didn’t see the exact changes you’ve made in your LDM, but based on your description, it seems like you’ve interrupted some dataset connections. In this case, this behavior is expected. Insights can no longer be computed once the connections to related datasets are broken. The same thing would happen, for example, if you were to delete some facts or attributes from your LDM. All nested objects used by them would cease to function. Therefore, please be careful with the changes you are doing within your LDM.
j
You may have misunderstood the timeline of events. No data sets were deleted or broken when my insights were deleted, only an ambiguous path for one attribute within the data sets. There also wasn't any warning or confirmation to delete so much of my work, it was just gone. This does not seem like something that should be expected behavior.
m
Hi Josh, thanks for getting back to us with this additional information - please let us review this and get back you once we have more details.
👍 1
d
Hello, May I know if the warning you saw in the LDM was exactly:
Alternative paths found in model
? Also, I see you mentioned that you got an error, was it like this? or it was a mix of this and deleted insights? And the changes in the LDM, was it 1:N or M:N? Mainly asking this last one, since I was able to reproduce this issue partially, but I would like to check if it happens under other circumstances
j
I did not receive any error messages, it allowed me to save the model without issue, and upon doing so it removed my work. I do not remember what the exact warning label that I saw said besides that i should alter the data sets to avoid ambiguous paths. Yes i had created ambiguous paths to an attribute by selecting M:N. The ambiguous paths is not an issue though, the issue is that so much of my work was deleted without any confirmation, regardless of available data. I would have expected it to just show the insights as you show, 'unable to display'.
d
Hi Josh, I totally agree, that it shouldn’t have deleted all the work. I’m trying to reproduce the issue, to be able to check were this is triggering and how to avoid it. This is why I was asking about the behaviour. I will check a bit deeper on this and get back to you soon. I’m very sorry that this happened