Include pod UID and container name as optional trace labels

It's useful to include these fields in the trace span attributes to help correlate traffic within a cluster. This propagates these labels through the trace initialization, similar to how we handle the service name (and other labels that last for the entire pod lifetime). See linkerd/linkerd2#13501 for the corresponding control plane change. Signed-off-by: Scott Fleener <[email protected]>
linkerd · Dec 20, 2024 · a0ee879 · a0ee879
1 parent a1a5713
commit a0ee879
Show file tree

Hide file tree

Showing 3 changed files with 42 additions and 10 deletions.
diff --git a/linkerd/app/src/env.rs b/linkerd/app/src/env.rs
@@ -266,6 +266,9 @@ const ENV_OUTBOUND_HTTP1_CONNECTION_POOL_IDLE_TIMEOUT: &str =
 
 const ENV_SHUTDOWN_GRACE_PERIOD: &str = "LINKERD2_PROXY_SHUTDOWN_GRACE_PERIOD";
 
+const ENV_POD_UID: &str = "_pod_uid";
+const ENV_POD_CONTAINER_NAME: &str = "_pod_containerName";
+
 // Default values for various configuration fields
 const DEFAULT_OUTBOUND_LISTEN_ADDR: &str = "127.0.0.1:4140";
 pub const DEFAULT_INBOUND_LISTEN_ADDR: &str = "0.0.0.0:4143";
@@ -395,6 +398,9 @@ pub fn parse_config<S: Strings>(strings: &S) -> Result<super::Config, EnvError>
 
     let shutdown_grace_period = parse(strings, ENV_SHUTDOWN_GRACE_PERIOD, parse_duration);
 
+    let pod_uid = strings.get(ENV_POD_UID);
+    let container_name = strings.get(ENV_POD_CONTAINER_NAME);
+
     let inbound_discovery_idle_timeout =
         parse(strings, ENV_INBOUND_DISCOVERY_IDLE_TIMEOUT, parse_duration);
     let outbound_discovery_idle_timeout =
@@ -845,10 +851,14 @@ pub fn parse_config<S: Strings>(strings: &S) -> Result<super::Config, EnvError>
                 .unwrap_or_default();
 
             let trace_service_name = trace_service_name.ok().flatten();
+            let pod_uid = pod_uid.ok().flatten();
+            let container_name = container_name.ok().flatten();
 
             trace_collector::Config::Enabled(Box::new(trace_collector::EnabledConfig {
                 attributes,
                 hostname: hostname?,
+                pod_uid,
+                container_name,
                 service_name: trace_service_name,
                 control: ControlConfig {
                     addr,

diff --git a/linkerd/app/src/trace_collector.rs b/linkerd/app/src/trace_collector.rs
@@ -1,11 +1,13 @@
-use linkerd_app_core::http_tracing::{CollectorProtocol, SpanSink};
-use linkerd_app_core::metrics::ControlHttp as HttpMetrics;
-use linkerd_app_core::svc::NewService;
-use linkerd_app_core::{control, dns, identity, opencensus, opentelemetry};
+use linkerd_app_core::{
+    control, dns,
+    http_tracing::{CollectorProtocol, SpanSink},
+    identity,
+    metrics::ControlHttp as HttpMetrics,
+    opencensus, opentelemetry,
+    svc::NewService,
+};
 use linkerd_error::Error;
-use std::collections::HashMap;
-use std::future::Future;
-use std::pin::Pin;
+use std::{collections::HashMap, future::Future, pin::Pin};
 
 pub mod oc_collector;
 pub mod otel_collector;
@@ -24,6 +26,8 @@ pub struct EnabledConfig {
     pub control: control::Config,
     pub attributes: HashMap<String, String>,
     pub hostname: Option<String>,
+    pub pod_uid: Option<String>,
+    pub container_name: Option<String>,
     pub service_name: Option<String>,
     pub kind: CollectorProtocol,
 }
@@ -95,6 +99,8 @@ impl Config {
                     CollectorProtocol::OpenTelemetry => otel_collector::create_collector(
                         addr.clone(),
                         inner.hostname,
+                        inner.pod_uid,
+                        inner.container_name,
                         svc_name,
                         inner.attributes,
                         svc,

diff --git a/linkerd/app/src/trace_collector/otel_collector.rs b/linkerd/app/src/trace_collector/otel_collector.rs
@@ -4,10 +4,15 @@ use linkerd_app_core::{
 };
 use linkerd_opentelemetry::{
     self as opentelemetry, metrics,
-    proto::proto::common::v1::{any_value, AnyValue, KeyValue},
-    proto::transform::common::ResourceAttributesWithSchema,
+    proto::{
+        proto::common::v1::{any_value, AnyValue, KeyValue},
+        transform::common::ResourceAttributesWithSchema,
+    },
+};
+use std::{
+    collections::HashMap,
+    time::{SystemTime, UNIX_EPOCH},
 };
-use std::{collections::HashMap, time::SystemTime, time::UNIX_EPOCH};
 use tokio::sync::mpsc;
 use tokio_stream::wrappers::ReceiverStream;
 use tonic::{body::BoxBody, client::GrpcService};
@@ -16,6 +21,8 @@ use tracing::Instrument;
 pub(super) fn create_collector<S>(
     addr: ControlAddr,
     hostname: Option<String>,
+    pod_uid: Option<String>,
+    container_name: Option<String>,
     service_name: String,
     attributes: HashMap<String, String>,
     svc: S,
@@ -53,6 +60,15 @@ where
         .attributes
         .0
         .push(hostname.unwrap_or_default().with_key("host.name"));
+    if let Some(pod_uid) = pod_uid {
+        resources.attributes.0.push(pod_uid.with_key("k8s.pod.uid"));
+    }
+    if let Some(container_name) = container_name {
+        resources
+            .attributes
+            .0
+            .push(container_name.with_key("k8s.container.name"));
+    }
 
     resources.attributes.0.extend(
         attributes