fyuan1316

@joe-elliott Thanks for the quick reply! >Currently I believe we add the database name to the server label and set the connection_type label equal to database. In your example we...

Thanks @joe-elliott ! I'm glad you like the proposal and are helping to confirm and move it forward. Yes, I tend to split it in two to represent the latency...

Sorry, I may have misunderstood. According to the database implementation, it is, and it looks like it sets ServerLatencySec to the same value as ClientLatencySec. Do you think this is...

Thanks for your reply! I agree with your approach. It's a great idea to use existing information to render the middle node of the asynchronous call relationship. I also noticed...

Great! I have created a Grafana issue and referenced this one. Thank you for your support!

Based on the current implementation, I agree with @kvrhdn that is the best Grafana could do. Adding a label representing the specific implementation of `messaging_system` would be better. What do...

One small question, do we want the edge to represent the direction of the instruction flow or the data flow? If it represents the instruction flow, I think the `consumer`...

Hi @t00mas, This PR https://github.com/grafana/tempo/pull/3453 introduces a new metric `traces_service_graph_request_messaging_system_seconds` for messaging system, which has recently been merged. I wonder if this will help?

Thanks for your quick reply. Consider if the istio community will not change the way the metrics are generated. I think considering swapping metrics without changing the direction of the...

I don't know if this is a bug. just don't quite understand the significance of the calculations. As I said above. When the sent traffic is selected for the query,...