chore(schema, telemetry): add and update schema telemetry events COMPASS-8818 #6723
Conversation
Anemy
added
no release notes
|
I wonder if we should distinguish between copy and download? |
| if (abortSignal?.aborted) { | ||
| throw new Error(abortSignal?.reason || new Error('Operation aborted')); | ||
| } | ||
|
|
There was a problem hiding this comment.
Do we want this to be an error even though it was an user action? Especially since now it's difficult to get to a re-try, the user basically needs to close&reopen the tab:
There was a problem hiding this comment.
Sorry that I didn't notice this sooner
There was a problem hiding this comment.
Note: I'll be changing the error handling in #6733, I can include check for this case and show a warning toast, if we some notification for the user.
There was a problem hiding this comment.
Good callout, my main reason for changing it to this flow was to align it with the find/documents page:
Folks can still re-try by clicking the Analyze button. That being sad I do agree it's not as discoverable or intuitive as having the big button there. I'm happy going either way here. Would you prefer not having it be an error when someone aborts the operation and return them to the zero state?
Sending you a ping on slack to chat on errors and toasts.
There was a problem hiding this comment.
Toast vs in-tab error TBD. For the question on whether intentional abort is an error - aggregation goes to "no results" state if you cancel.
There was a problem hiding this comment.
Looks like both this and the errors in general don't have a clear strategy :/
paula-stacho
left a comment
paula-stacho
left a comment
There was a problem hiding this comment.
Looks good! We can figure out the errors with the next PR.
COMPASS-8818
Adds a new telemetry events:
Schema Analysis Cancelled.Updates the existing
Schema ExportedandSchema Analyzedto have new events.Updated the way we gather schema metadata for the events to make it so that we only traverse the tree once instead of twice, also makes it so that we unblock the thread occasionally while traversing this tree in case of a very large tree.
We're only tracking the export when folks click the copy button, or click the export button, see discussion in https://mongodb.slack.com/archives/C07JL273J06/p1739389537801429
We only track the
optional_field_countandvariable_type_countproperties for top level fields re https://mongodb.slack.com/archives/C07JL273J06/p1739398926141009