Description
I've been digging into instrumenting our project with opentelemetry, and the tracing you've provided works great for getting metrics for queries already run, but I wonder what your appetite would be for adding traces in right before a query is executed? I'd be happy to jump in and add it in if you're open to the feature. It's helpful in our trace visualization tool to see where and when a query starts and end to see how it affects application performance in relation to concurrently running code. Also knowing the best place to add it in would be helpful. I see that the current traces happen in the check_for_error
function, which won't run if you don't get an {:ok, cmd, response}
response from DBConnection.execute
. I wonder if DBConnection.execute
would be the best place to put that? Or perhaps in something that gets called everywhere before DBConnection.execute
gets called?