Merge pull request #213 from StackVista/STAC-23146

[STAC-23146] Expose TCP connections and protocol metrics with OTEL

Author: Andreagit97

checks/net.go

@@ -59,6 +59,8 @@ type ConnectionsCheck struct {
 
 	// Use this as the network relation cache to calculate rate metrics and drop short-lived network relations
 	cache *NetworkRelationCache
+
+	podCorrelation *podCorrelationInfo
 }
 
 // Name returns the name of the ConnectionsCheck.
@@ -146,6 +148,21 @@ func (c *ConnectionsCheck) Run(cfg *config.AgentConfig, groupID int32, currentTi
 		protocolMap[k] = config.PostgresProtocolName
 	}
 
+	log.Debugf("Protocol map: %v", protocolMap)
+	log.Debugf("collected %d connection data", len(connectionStats))
+	for key, metrics := range connectionStats {
+		log.Debugf("connection data for %s", key)
+		for _, metric := range metrics {
+			log.Debugf("\t%v", metric)
+		}
+	}
+	// Here we have all we need to export OTEL metrics
+	// We cannot filter short-lived connections because we need them for the security use case.
+	if c.podCorrelation != nil {
+		c.podCorrelation.processConnections(conns.Conns, connectionStats)
+		// In the future we could return here and disable all the rest of the logic, also the process Check
+	}
+
 	httpObservations := aggregateHTTPTraceObservations(conns.HTTPObservations)
 
 	dnsMap := map[string][]dns.Hostname{}
@@ -160,15 +177,6 @@ func (c *ConnectionsCheck) Run(cfg *config.AgentConfig, groupID int32, currentTi
 		containerToPod = c.podsCache.GetContainerToPodMap(context.TODO())
 	}
 
-	log.Debugf("Protocol map: %v", protocolMap)
-	log.Debugf("collected %d connection data", len(connectionStats))
-	for key, metrics := range connectionStats {
-		log.Debugf("connection data for %s", key)
-		for _, metric := range metrics {
-			log.Debugf("\t%v", metric)
-		}
-	}
-
 	formattedConnections, connsPods := c.formatConnections(cfg, conns.Conns, connectionStats, httpObservations, containerToPod, protocolMap)
 	c.prevCheckTime = currentTime
 

checks/net_linux.go

@@ -33,6 +33,14 @@ func (c *ConnectionsCheck) Init(cfg *config.AgentConfig, _ *model.SystemInfo) er
 		return fmt.Errorf("failed to create network tracer: %s.  Set the environment STS_NETWORK_TRACING_ENABLED to false to disable network connections reporting", err)
 	}
 
+	if cfg.NetworkTracer.PodCorrelation.Enabled {
+		cfg.NetworkTracer.PodCorrelation.ObserverLogLevel = cfg.LogLevel
+		c.podCorrelation, err = newPodCorrelationInfo(&cfg.NetworkTracer.PodCorrelation)
+		if err != nil {
+			return err
+		}
+	}
+
 	c.podsCache = pods.MakeCachedPods(60 * time.Second)
 	c.localTracer = t