[Feature] Agent tracing Implementation #652
yourbuddyconner
started this conversation in
Ideas
Replies: 1 comment 4 replies
-
Implementation Details and Crate Deps Notes: @prestwich Found this killer twitter thread that helps shed some light on the complex Rust Tracing ecosystem. |
Beta Was this translation helpful? Give feedback.
4 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
We need a overarching implementation strategy to correlate agent spans together. In a web programming/microservices context this is often the
Request_ID
orCorrelation_ID
.Summarized:
Correlation_ID
header (i.e. this request comes from the user) generate one and add it to downstream requests.Correlation_ID
header (i.e. this request comes from another microservice), use this value in downstream requests.In this way, all spans can be correlated/loaded together in a tracing UI to get the full picture of request propagation and latency.
We have a slightly harder time of it because the system we're building has an entirely different architecture but we can still borrow this
Correlation_ID
concept and use it to our advantage.In an ideal world, we would be able to implement some
Correlation_ID
that connects all spans relating to a particularmessage
together. However, due to "The Way It Is" this proves difficult to do in practice, in particular because messages are batched together when relayed and potentially processed out of order.Ideas for ways to correlate spans together:
CommittedMessage
(Link)Beta Was this translation helpful? Give feedback.
All reactions