compute_ctl: Add OTEL tracing to incoming HTTP requests and startup (#10971)

We lost this with the switch to axum for the HTTP server. Add it back.

In addition to just resurrecting the functionality we had before, pass
the tracing context of the /configure HTTP request to the start_postgres
operation that runs in the main thread. This way, the 'start_postgres'
and all its sub-spans like getting the basebackup become children of the
HTTP request span. This allows end-to-end tracing of a compute start,
all the way from the proxy to the SQL queries executed by compute_ctl as
part of compute startup.
This commit is contained in:
Heikki Linnakangas
2025-02-26 19:27:16 +00:00
committed by GitHub
parent 643a48210f
commit 5cfdb1244f
7 changed files with 65 additions and 5 deletions
+6 -1
View File
@@ -45,13 +45,18 @@ pub(in crate::http) async fn configure(
return JsonResponse::invalid_status(state.status);
}
// Pass the tracing span to the main thread that performs the startup,
// so that the start_compute operation is considered a child of this
// configure request for tracing purposes.
state.startup_span = Some(tracing::Span::current());
state.pspec = Some(pspec);
state.set_status(ComputeStatus::ConfigurationPending, &compute.state_changed);
drop(state);
}
// Spawn a blocking thread to wait for compute to become Running. This is
// needed to do not block the main pool of workers and be able to serve
// needed to not block the main pool of workers and to be able to serve
// other requests while some particular request is waiting for compute to
// finish configuration.
let c = compute.clone();
+1
View File
@@ -121,6 +121,7 @@ impl From<Server> for Router<Arc<ComputeNode>> {
)
.layer(PropagateRequestIdLayer::x_request_id()),
)
.layer(tower_otel::trace::HttpLayer::server(tracing::Level::INFO))
}
}