Merge pull request #240 from golisai/worklooad-uuid

Introducing  workload.uid metric attribute for kubernetes workloads

docs/metrics/dimensions.yaml

@@ -29,6 +29,19 @@ dest.workload.name:
   brief: See workload.name
   description: See workload.name
 
+workload.uid:
+  brief: Kubernetes workload or application UUID.
+  description: If the associated metric comes from a monitored Kubernetes cluster running the k8-collector, the metric may note the UUID of the source and destination workload that the traffic being measured moved between. This UUID represents the unique identifier of the associated Deployment or DaemonSet. 
+  example: 0485c36b-283e-496e-af47-1a69545cd887
+
+source.workload.uid:
+  brief: See workload.uid
+  description: See workload.uid
+
+dest.workload.uid:
+  brief: See workload.uid
+  description: See workload.uid
+
 image_version:
   brief: Kubernetes pod version label of the workload.
   description: If the associated metric comes from a monitored Kubernetes cluster running the k8-collector, the metric may note the source and destination versions that the traffic being measured moved between.  The version label is optionally set by the administrator configuring Kubernetes.  Sometimes referred to as the 'service version.'  May be found in the docker (pod) image or Kubernetes metadata.

reducer/aggregation/agg_root_span.cc

@@ -57,6 +57,9 @@ void AggRootSpan::update_node(
   using role_t = ::ebpf_net::aggregation::spans::role::s_t;
   auto role = as_short_string<role_t>(msg->role, stat_counters.trunc_role);
 
+  using uid_t = ::ebpf_net::aggregation::spans::role::uid_t;
+  auto role_uid = as_short_string<uid_t>(msg->role_uid, stat_counters.trunc_role_uid);
+
   using version_t = ::ebpf_net::aggregation::spans::role::version_t;
   auto version = as_short_string<version_t>(msg->version, stat_counters.trunc_version);
 
@@ -84,6 +87,7 @@ void AggRootSpan::update_node(
   if (!role_ref.valid()) {
     return;
   }
+  role_ref.modify().uid(role_uid);
 
   auto az_ref = index.az.by_key({az, role_ref.loc()});
   if (!az_ref.valid()) {

reducer/aggregation/labels.h

@@ -20,6 +20,7 @@ struct NodeLabels {
   std::string ip;
   std::string az;
   std::string role;
+  std::string role_uid;
   std::string version;
   std::string env;
   std::string ns;

reducer/aggregation/labels.inl

@@ -9,6 +9,7 @@
 
 #define FOREACH_NODE_LABEL(FUNC)                                                                                               \
   FUNC("workload.name", role)                                                                                                  \
+  FUNC("workload.uid", role_uid)                                                                                               \
   FUNC("availability_zone", az)                                                                                                \
   FUNC("id", id)                                                                                                               \
   FUNC("ip", ip)                                                                                                               \
@@ -39,6 +40,7 @@ inline NodeLabels::NodeLabels(::ebpf_net::aggregation::weak_refs::az az_ref) : N
 inline NodeLabels::NodeLabels(::ebpf_net::aggregation::weak_refs::role role_ref)
 {
   role = role_ref.s().to_string();
+  role_uid = role_ref.uid().to_string();
   version = role_ref.version().to_string();
   env = role_ref.env().to_string();
   ns = role_ref.ns().to_string();

reducer/aggregation/stat_counters.h

@@ -17,7 +17,8 @@
   FN(pod_name, 6)                                                                                                              \
   FN(process, 7)                                                                                                               \
   FN(role, 8)                                                                                                                  \
-  FN(version, 9)
+  FN(version, 9)                                                                                                               \
+  FN(role_uid, 10)
 
 namespace reducer::aggregation {
 

reducer/ingest/agent_span.cc

@@ -610,6 +610,7 @@ void AgentSpan::bpf_lost_samples(
 void AgentSpan::set_pod_new(
     std::string_view uid,
     std::string_view owner_name,
+    std::string_view owner_uid,
     std::string_view pod_name,
     std::string_view ns,
     std::string_view version,
@@ -640,7 +641,7 @@ void AgentSpan::set_pod_new(
       version);
 
   // populate the pod span with pod metadata
-  k8s_pod.set_pod_detail(jb_blob(owner_name), jb_blob{pod_name}, jb_blob(ns), jb_blob(version));
+  k8s_pod.set_pod_detail(jb_blob(owner_name), jb_blob{pod_name}, jb_blob(ns), jb_blob(version), jb_blob(owner_uid));
 
   // Add pod to k8s_pod_set if it isn't already there.
 
@@ -661,18 +662,18 @@ void AgentSpan::set_pod_new(
 
 void AgentSpan::pod_new_legacy(::ebpf_net::ingest::weak_refs::agent span_ref, u64 timestamp, jsrv_ingest__pod_new_legacy *msg)
 {
-  set_pod_new(msg->uid, msg->owner_name, {}, msg->ns, {}, msg->ip);
+  set_pod_new(msg->uid, msg->owner_name, msg->owner_uid, {}, msg->ns, {}, msg->ip);
 }
 
 void AgentSpan::pod_new_legacy2(::ebpf_net::ingest::weak_refs::agent span_ref, u64 timestamp, jsrv_ingest__pod_new_legacy2 *msg)
 {
-  set_pod_new(msg->uid, msg->owner_name, {}, msg->ns, msg->version, msg->ip);
+  set_pod_new(msg->uid, msg->owner_name, msg->owner_uid, {}, msg->ns, msg->version, msg->ip);
 }
 
 void AgentSpan::pod_new_with_name(
     ::ebpf_net::ingest::weak_refs::agent span_ref, u64 timestamp, jsrv_ingest__pod_new_with_name *msg)
 {
-  set_pod_new(msg->uid, msg->owner_name, msg->pod_name, msg->ns, msg->version, msg->ip);
+  set_pod_new(msg->uid, msg->owner_name, msg->owner_uid, msg->pod_name, msg->ns, msg->version, msg->ip);
 }
 
 void AgentSpan::pod_container(::ebpf_net::ingest::weak_refs::agent span_ref, u64 timestamp, jsrv_ingest__pod_container *msg)

reducer/ingest/agent_span.h

@@ -107,6 +107,7 @@ class AgentSpan : public ::ebpf_net::ingest::AgentSpanBase {
   void set_pod_new(
       std::string_view uid,
       std::string_view owner_name,
+      std::string_view owner_uid,
       std::string_view pod_name,
       std::string_view ns,
       std::string_view version,

reducer/matching/flow_span.cc

@@ -230,7 +230,8 @@ void FlowSpan::update_node(::ebpf_net::matching::weak_refs::agg_root agg_root, F
       jb_blob(n.address),
       jb_blob(n.comm),
       jb_blob(n.container_name),
-      jb_blob(n.pod_name));
+      jb_blob(n.pod_name),
+      jb_blob(n.role_uid));
 }
 
 void FlowSpan::create_agg_root(::ebpf_net::matching::weak_refs::flow flow, NodeData const &node_a, NodeData const &node_b)
@@ -316,6 +317,7 @@ FlowSpan::NodeData FlowSpan::resolve_node(::ebpf_net::matching::weak_refs::flow
   auto [id, az, is_autonomous_system] = get_id_az(side);
 
   std::string role;
+  std::string role_uid;
   std::string version;
   std::string env;
   std::string ns;
@@ -364,6 +366,7 @@ FlowSpan::NodeData FlowSpan::resolve_node(::ebpf_net::matching::weak_refs::flow
     // enrich
     node_type = NodeResolutionType::K8S_CONTAINER;
     role = pod.owner_name();
+    role_uid = pod.owner_uid();
     version = pod.version();
     ns = pod.ns();
 
@@ -477,6 +480,7 @@ FlowSpan::NodeData FlowSpan::resolve_node(::ebpf_net::matching::weak_refs::flow
       .id = id,
       .az = az,
       .role = role,
+      .role_uid = role_uid,
       .version = version,
       .env = env,
       .ns = ns,

reducer/matching/flow_span.h

@@ -97,6 +97,7 @@ class FlowSpan : public ::ebpf_net::matching::FlowSpanBase {
     std::string id;
     std::string az;
     std::string role;
+    std::string role_uid;
     std::string version;
     std::string env;
     std::string ns;

reducer/matching/k8s_pod_span.cc

@@ -20,6 +20,7 @@ void K8sPodSpan::set_pod_detail(
 {
   span_ref.modify()
       .owner_name({short_string_behavior::truncate, msg->owner_name})
+      .owner_uid({short_string_behavior::truncate, msg->owner_uid})
       .pod_name({short_string_behavior::truncate, msg->pod_name})
       .ns({short_string_behavior::truncate, msg->ns})
       .version({short_string_behavior::truncate, msg->version});

render/ebpf_net.render

@@ -1098,6 +1098,9 @@ app ingest {
     // can be obtained from docker label `io.kubernetes.pod.name`??
     string<64> owner_name   // name of the Deployment, DaemonSet, ReplicaSet or StatefulSet owning the pod
 
+    // owner uid
+    string<64> owner_uid
+
     // can be obtained from docker label `io.kubernetes.pod.namespace`
     string<64> ns
 
@@ -1343,6 +1346,7 @@ app matching {
     u64 uid_hash
 
     string<64> owner_name   // name of the Deployment, DaemonSet, ReplicaSet or StatefulSet owning the pod
+    string<64> owner_uid
     string<64> pod_name
     string<64> ns
     string<64> version
@@ -1357,6 +1361,7 @@ app matching {
       2: string pod_name
       3: string ns
       4: string version
+      5: string owner_uid
     }
   } /* span k8s_pod */
 
@@ -1425,6 +1430,7 @@ app aggregation {
     pool_size 70000
     index (s, version, env, ns, node_type, process, container)
     string<80> s
+    string<80> uid
     string<64> version
     string<32> env
     string<64> ns
@@ -1498,6 +1504,7 @@ app aggregation {
       11: string process
       12: string container
       13: string pod_name
+      14: string role_uid
     }
     14: msg update_tcp_metrics {
       1: u8 direction