Traffic / mTLS / Circuit breaking
Traffico e routing
Section titled “Traffico e routing”Richieste per coppia source→destination (matrice di traffico)
Section titled “Richieste per coppia source→destination (matrice di traffico)”sum by (source_workload, destination_workload) ( rate(istio_requests_total{reporter="destination"}[5m]))Richieste per response_flags (capire se sono drop/retry/timeout lato Envoy)
Section titled “Richieste per response_flags (capire se sono drop/retry/timeout lato Envoy)”sum by (response_flags) ( rate(istio_requests_total{reporter="destination"}[5m]))Flag comuni da monitorare: UO (upstream overflow — circuit breaker aperto), UF (upstream connection failure), UT (upstream timeout), NR (no route configurata).
Traffico non cifrato/non-mTLS nel mesh (verifica PeerAuthentication STRICT)
Section titled “Traffico non cifrato/non-mTLS nel mesh (verifica PeerAuthentication STRICT)”sum by (source_workload, destination_workload, connection_security_policy) ( rate(istio_requests_total{reporter="destination"}[5m]))Filtrare poi connection_security_policy!="mutual_tls" per isolare traffico non-mTLS: sintomo di workload fuori mesh o PeerAuthentication troppo permissiva (PERMISSIVE/DISABLE).
Circuit breaking / Connection pool
Section titled “Circuit breaking / Connection pool”Connessioni upstream attive per destinazione (saturazione pool)
Section titled “Connessioni upstream attive per destinazione (saturazione pool)”envoy_cluster_upstream_cx_activeRichieste rifiutate per overflow del circuit breaker
Section titled “Richieste rifiutate per overflow del circuit breaker”rate(envoy_cluster_upstream_rq_pending_overflow[5m])Retry effettuati per destinazione
Section titled “Retry effettuati per destinazione”sum by (destination_service_name) ( rate(istio_requests_total{reporter="destination", response_flags=~".*RR.*"}[5m]))Connessioni resettate (RST) verso upstream
Section titled “Connessioni resettate (RST) verso upstream”rate(envoy_cluster_upstream_cx_destroy_with_active_rq[5m])Timeout
Section titled “Timeout”Richieste terminate per timeout (UT flag)
Section titled “Richieste terminate per timeout (UT flag)”sum by (destination_service_name) ( rate(istio_requests_total{reporter="destination", response_flags=~".*UT.*"}[5m]))