Tenant Load Distribution Server (LDS) metrics and alerts

Find the metrics Tenant Load Distribution Server (LDS) exposes and the alerts defined for Tenant Load Distribution Server (LDS).

Related documentation:

• How Reporting works

Service	CRD or annotations?	Port	Endpoint/Selector	Metrics update interval
Tenant Load Distribution Server (LDS)	PodMonitor	9091	selector: matchLabels: app.kubernetes.io/name: {{include "common.util.chart.name" . }} app.kubernetes.io/instance: {{include "common.util.chart.fullname" . }} service: {{.Release.Namespace }} servicename: {{include "common.util.chart.name" . }} tenant: {{.Values.tenant.sid }} Endpoints to query: /metrics/	30 seconds

See details about:

• Tenant Load Distribution Server (LDS) metrics
• Tenant Load Distribution Server (LDS) alerts

Metrics

Metric and description	Metric details	Indicator of
pulse_monitor_check_duration_seconds The duration in seconds of the last health check performed by Monitor Agent.	Unit: seconds Type: Gauge Label: tenant Sample value:	Error
pulse_lds_uptime_seconds The LDS container uptime in seconds.	Unit: seconds Type: Gauge Label: tenant Sample value:	Error
pulse_lds_senders_number The number of upstream servers to which the LDS is connected.	Unit: Type: Gauge Label: tenant Sample value: 2	Error
pulse_lds_receivers_number The number of clients connected to the LDS.	Unit: Type: Gauge Label: tenant Sample value: 2	Error
pulse_lds_sender_connected_seconds Duration in seconds of connection to the upstream server.	Unit: seconds Type: Gauge Label: tenant, sender Sample value:	Error
pulse_lds_sender_disconnected_seconds Duration in seconds of disconnection from the upstream server.	Unit: seconds Type: Gauge Label: tenant, sender Sample value:	Error
pulse_lds_sender_registered_dns_number The number of DNs registered on the upstream server.	Unit: Type: Gauge Label: tenant, sender Sample value: 1000	Saturation
pulse_lds_sender_registration_errors_number The number of failed registrations of DNs on the upstream server.	Unit: Type: Gauge Label: tenant, sender Sample value: 0	Error
pulse_lds_receiver_connected_seconds Duration in seconds of client connection to the LDS.	Unit: seconds Type: Gauge Label: tenant, receiver Sample value:	Error
pulse_lds_receiver_registered_dns_number The number of DNs registered by the client.	Unit: Type: Gauge Label: tenant, receiver Sample value: 1000	Saturation
pulse_lds_receiver_registration_errors_number The number of failed registrations of DNs received from the client.	Unit: Type: Gauge Label: tenant, receiver Sample value: 0	Error

Alerts

Alerts are based on LDS and Kubernetes cluster metrics.

The following alerts are defined for Tenant Load Distribution Server (LDS).

Alert	Severity	Description	Based on	Threshold
pulse_lds_monitor_data_unavailable	Critical	Pulse LDS Monitor Agents do not provide data.	pulse_monitor_check_duration_seconds, kube_statefulset_replicas	for 15 minutes
pulse_lds_critical_nonrunning_instances	Critical	Triggered when Pulse LDS instances are down.	kube_statefulset_status_replicas_ready, kube_statefulset_status_replicas	for 15 minutes
pulse_lds_too_frequent_restarts	Critical	Detected too frequent restarts of LDS Pod container.	kube_pod_container_status_restarts_total	2 for 1 hour
pulse_lds_critical_cpu	Critical	Detected critical CPU usage by Pulse LDS Pod.	container_cpu_usage_seconds_total, kube_pod_container_resource_limits	90%
pulse_lds_critical_memory	Critical	Detected critical memory usage by Pulse LDS Pod.	container_memory_working_set_bytes, kube_pod_container_resource_limits	90%
pulse_lds_no_connected_senders	Critical	Pule LDS is not connected to upstream servers.	pulse_lds_senders_number	for 15 minutes
pulse_lds_no_registered_dns	Critical	No DNs are registered on Pulse LDS.	pulse_lds_sender_registered_dns_number	for 30 minutes