Track SignalR traffic
Right now, SignalR and WebSockets messages are invisible to AI. Worse yet, they count as a really long HTTP request. It would be nice to have a separate diagram in the Live Metrics Stream blade and everywhere else where successful or failed request are being displayed.
Anonymous
shared this idea
Thanks for your valuable feedback. Could you please share about your scenario which would help us to better understand the scope. We do have other feedback request from customers to ignore few parts with in SignalTraffic . https://feedback.azure.com/forums/357324-application-insights/suggestions/8395065-server-response-time-needs-to-ignore-signalr#{toggle_previous_statuses}.
So want to better understand the two requirements and prioritize it accordingly with our existing feature backlog. Please share additional details on your scenario. Thank you
3 comments
-
Kevin Griffin
commented
I'd also appreciate this! Our hubs register as 6-10 minutes in the performance counter. It's horrible.
-
Francis Moore
commented
We would greatly benefit from this feature as well.
Here's our scenario. We are moving some of the communications between our web app frontend and backend (.Net Core) from HTTP to SignalR. Doing that, we loose all the request tracking Application Insights comes with. We end up with one very long HTTP call, but no detail about the individual messages going back and forth between our frontend and backend.
What we are looking for is the same level of functionnality with SignalR than what you get out of the box for HTTP calls:
- Request tracking for individual SignalR messages, with resultCode, duration, etc
- Correlation with dependancies, traces, events, exceptions and metrics being tracked during those requests -
Nick Muller
commented
At least don't log the call as a really long http request.