Configuring Prometheus Metrics
Configure metrics endpoints and ServiceMonitor resources for Connectware components using the connectware and connectware-agent Helm charts.
The connectware and connectware-agent Helm charts can expose Prometheus metrics for their components and create Kubernetes ServiceMonitor custom resources so a Prometheus Operator installation can discover and scrape them automatically.
Prometheus Operator is a Kubernetes controller that manages Prometheus instances declaratively. Instead of maintaining a static scrape configuration, you create ServiceMonitor resources that describe where and how Prometheus should scrape a service. It is commonly installed as part of the kube-prometheus-stack Helm chart.
Service Monitors
A ServiceMonitor is a Prometheus Operator custom resource that tells Prometheus which service to scrape and how. Rather than editing the Prometheus scrape configuration by hand, you let the charts generate these resources for you.
Both charts can create ServiceMonitor resources for the components and agents that expose a metrics endpoint. Set serviceMonitor.enabled to true on a component, an agent, or globally, and the chart generates a ServiceMonitor that Prometheus Operator discovers and scrapes automatically. Every metrics values table under Configuration Values includes a serviceMonitor.* block.
If your Prometheus Operator installation restricts which ServiceMonitor resources it selects through a serviceMonitorSelector, set the matching labels via serviceMonitor.labels. Otherwise the generated ServiceMonitor is ignored.
If Prometheus Operator is not available in your cluster, see Scraping Connectware Metrics for static configuration and Kubernetes Service Discovery alternatives.
Components with Metrics Support
connectware Helm Chart
The connectware Helm chart configures metrics on the components listed below.
All components (global)
global.metrics
Auth Server
authServer.metrics
Broker
broker.metrics
NATS
nats.metrics
Protocol Mapper
protocolMapper.metrics
Resource Status Tracking
resourceStatusTracking.metrics
Service Manager
serviceManager.metrics
Auth Server, Protocol Mapper, Resource Status Tracking, and Service Manager expose a metrics endpoint directly on containerPort 9184. Broker exposes a metrics endpoint directly on containerPort 8888. NATS does not expose its own metrics endpoint. Instead, nats.metrics.enabled deploys a Prometheus exporter as a sidecar next to the NATS container.
connectware-agent Helm Chart
The connectware-agent Helm chart configures metrics on protocol-mapper agents. Each agent exposes a metrics endpoint on containerPort 9184, the same as the in-cluster Protocol Mapper.
Defaults for all agents
protocolMapperAgentDefaults.metrics
Override for a single agent
protocolMapperAgents[*].metrics
A per-agent metrics value always takes precedence over the defaults set in protocolMapperAgentDefaults.metrics.
Configuration Values
The following Helm values control metrics for each chart.
connectware Helm Chart
The connectware Helm chart groups metrics values by component. Every metrics endpoint also accepts a serviceMonitor.* block for creating a Prometheus Operator ServiceMonitor resource.
Global Metrics
Values under global.metrics apply as defaults to every component that supports metrics. A component-level metrics value always takes precedence over the global value.
global.metrics.enabled
Enables metrics in every component that supports it. Default: true.
global.metrics.collectIntervalMs
How often collected values are stored into Prometheus counters, in milliseconds.
global.metrics.rateLimit.count
Maximum number of scrape requests allowed within the rate limit window.
global.metrics.rateLimit.windowMs
Duration of the rate limit window, in milliseconds.
global.metrics.serviceMonitor.enabled
Creates a ServiceMonitor for every component that supports it. Default: false.
global.metrics.serviceMonitor.namespace
Namespace for the generated ServiceMonitor resources. Default: release namespace.
global.metrics.serviceMonitor.labels
Labels applied to every generated ServiceMonitor.
global.metrics.serviceMonitor.jobLabel
Label whose value replaces the job name in the scraped metrics.
global.metrics.serviceMonitor.interval
Prometheus scrape interval for the ServiceMonitor endpoints.
global.metrics.serviceMonitor.scrapeTimeout
Prometheus scrape timeout for the ServiceMonitor endpoints.
global.metrics.serviceMonitor.metricRelabelings
Metric relabeling rules applied to samples before ingestion.
global.metrics.serviceMonitor.relabelings
Relabeling rules applied to scrape targets before scraping.
Standard Metrics Endpoints
Auth Server, Protocol Mapper, Resource Status Tracking, and Service Manager share the same metrics endpoint configuration. Replace ${COMPONENT_NAME} with authServer, protocolMapper, resourceStatusTracking, or serviceManager.
${COMPONENT_NAME}.metrics.enabled
Enables the metrics endpoint, a dedicated metrics service, and containerPort 9184. Inherits global.metrics.enabled when unset.
${COMPONENT_NAME}.metrics.collectIntervalMs
How often collected values are stored into Prometheus counters, in milliseconds.
${COMPONENT_NAME}.metrics.rateLimit.count
Maximum number of scrape requests allowed within the rate limit window.
${COMPONENT_NAME}.metrics.rateLimit.windowMs
Duration of the rate limit window, in milliseconds.
${COMPONENT_NAME}.metrics.serviceMonitor.enabled
Creates a ServiceMonitor for the component.
${COMPONENT_NAME}.metrics.serviceMonitor.namespace
Namespace for the ServiceMonitor.
${COMPONENT_NAME}.metrics.serviceMonitor.labels
Labels applied to the ServiceMonitor.
${COMPONENT_NAME}.metrics.serviceMonitor.jobLabel
Label whose value replaces the job name in the scraped metrics.
${COMPONENT_NAME}.metrics.serviceMonitor.interval
Prometheus scrape interval for the ServiceMonitor endpoint.
${COMPONENT_NAME}.metrics.serviceMonitor.scrapeTimeout
Prometheus scrape timeout for the ServiceMonitor endpoint.
${COMPONENT_NAME}.metrics.serviceMonitor.metricRelabelings
Metric relabeling rules applied to samples before ingestion.
${COMPONENT_NAME}.metrics.serviceMonitor.relabelings
Relabeling rules applied to the scrape target before scraping.
NATS
NATS exposes metrics through a Prometheus exporter sidecar rather than through a metrics endpoint of its own, so its values differ from the other components.
nats.metrics.enabled
Deploys a Prometheus exporter sidecar next to the NATS container.
nats.metrics.resources.requests.cpu
CPU request for the exporter sidecar.
nats.metrics.resources.requests.memory
Memory request for the exporter sidecar.
nats.metrics.resources.limits.cpu
CPU limit for the exporter sidecar.
nats.metrics.resources.limits.memory
Memory limit for the exporter sidecar.
nats.metrics.serviceMonitor.enabled
Creates a ServiceMonitor for the NATS exporter.
nats.metrics.serviceMonitor.namespace
Namespace for the ServiceMonitor.
nats.metrics.serviceMonitor.labels
Labels applied to the ServiceMonitor.
nats.metrics.serviceMonitor.jobLabel
Label whose value replaces the job name in the scraped metrics.
nats.metrics.serviceMonitor.interval
Prometheus scrape interval for the ServiceMonitor endpoint.
nats.metrics.serviceMonitor.scrapeTimeout
Prometheus scrape timeout for the ServiceMonitor endpoint.
nats.metrics.serviceMonitor.metricRelabelings
Metric relabeling rules applied to samples before ingestion.
nats.metrics.serviceMonitor.relabelings
Relabeling rules applied to the scrape target before scraping.
Broker
Broker exposes a metrics endpoint on containerPort 8888, but does not accept the shared collectIntervalMs or rateLimit.* tuning values.
broker.metrics.enabled
Enables the Broker metrics endpoint.
broker.metrics.serviceMonitor.enabled
Creates a ServiceMonitor for Broker.
broker.metrics.serviceMonitor.namespace
Namespace for the ServiceMonitor.
broker.metrics.serviceMonitor.labels
Labels applied to the ServiceMonitor.
broker.metrics.serviceMonitor.jobLabel
Label whose value replaces the job name in the scraped metrics.
broker.metrics.serviceMonitor.interval
Prometheus scrape interval for the ServiceMonitor endpoint.
broker.metrics.serviceMonitor.scrapeTimeout
Prometheus scrape timeout for the ServiceMonitor endpoint.
broker.metrics.serviceMonitor.metricRelabelings
Metric relabeling rules applied to samples before ingestion.
broker.metrics.serviceMonitor.relabelings
Relabeling rules applied to the scrape target before scraping.
connectware-agent Helm Chart
The connectware-agent Helm chart groups metrics values into defaults that apply to every agent and per-agent overrides. A per-agent value always takes precedence over the matching default.
Defaults for All Agents
Values under protocolMapperAgentDefaults.metrics apply to every agent unless overridden for a specific agent.
protocolMapperAgentDefaults.metrics.enabled
Enables the metrics endpoint, a dedicated metrics service, and containerPort 9184.
protocolMapperAgentDefaults.metrics.collectIntervalMs
How often collected values are stored into Prometheus counters, in milliseconds.
protocolMapperAgentDefaults.metrics.rateLimit.count
Maximum number of scrape requests allowed within the rate limit window.
protocolMapperAgentDefaults.metrics.rateLimit.windowMs
Duration of the rate limit window, in milliseconds.
protocolMapperAgentDefaults.metrics.serviceMonitor.enabled
Creates a ServiceMonitor for every agent.
protocolMapperAgentDefaults.metrics.serviceMonitor.namespace
Namespace for the generated ServiceMonitor resources.
protocolMapperAgentDefaults.metrics.serviceMonitor.labels
Labels applied to every generated ServiceMonitor.
protocolMapperAgentDefaults.metrics.serviceMonitor.jobLabel
Label whose value replaces the job name in the scraped metrics.
protocolMapperAgentDefaults.metrics.serviceMonitor.interval
Prometheus scrape interval for the ServiceMonitor endpoints.
protocolMapperAgentDefaults.metrics.serviceMonitor.scrapeTimeout
Prometheus scrape timeout for the ServiceMonitor endpoints.
protocolMapperAgentDefaults.metrics.serviceMonitor.metricRelabelings
Metric relabeling rules applied to samples before ingestion.
protocolMapperAgentDefaults.metrics.serviceMonitor.relabelings
Relabeling rules applied to scrape targets before scraping.
Overrides for a Single Agent
Values under protocolMapperAgents[*].metrics override the defaults for one agent. Only the values you set are overridden; the rest continue to inherit from protocolMapperAgentDefaults.metrics.
protocolMapperAgents[*].metrics.enabled
Enables the metrics endpoint, a dedicated metrics service, and containerPort 9184.
protocolMapperAgents[*].metrics.collectIntervalMs
How often collected values are stored into Prometheus counters, in milliseconds.
protocolMapperAgents[*].metrics.rateLimit.count
Maximum number of scrape requests allowed within the rate limit window.
protocolMapperAgents[*].metrics.rateLimit.windowMs
Duration of the rate limit window, in milliseconds.
protocolMapperAgents[*].metrics.serviceMonitor.enabled
Creates a ServiceMonitor for the agent.
protocolMapperAgents[*].metrics.serviceMonitor.namespace
Namespace for the ServiceMonitor.
protocolMapperAgents[*].metrics.serviceMonitor.labels
Labels applied to the ServiceMonitor.
protocolMapperAgents[*].metrics.serviceMonitor.jobLabel
Label whose value replaces the job name in the scraped metrics.
protocolMapperAgents[*].metrics.serviceMonitor.interval
Prometheus scrape interval for the ServiceMonitor endpoint.
protocolMapperAgents[*].metrics.serviceMonitor.scrapeTimeout
Prometheus scrape timeout for the ServiceMonitor endpoint.
protocolMapperAgents[*].metrics.serviceMonitor.metricRelabelings
Metric relabeling rules applied to samples before ingestion.
protocolMapperAgents[*].metrics.serviceMonitor.relabelings
Relabeling rules applied to the scrape target before scraping.
When the connectware-agent Helm chart is deployed as a subchart of the connectware Helm chart, agents also inherit any unset metrics value from the parent chart's global.metrics.*.
Configuring Metrics
connectware Helm Chart
Enabling Metrics for All Components
Metrics are enabled by default in the connectware Helm chart. To enable metrics explicitly for every component that supports it, set global.metrics.enabled to true:
To disable metrics everywhere, set global.metrics.enabled to false.
Enabling Metrics for a Single Component
To enable or disable metrics for one component without changing the global default, set metrics.enabled under that component's own configuration section. A component-level value always takes precedence over global.metrics.enabled.
The following example disables metrics globally, then re-enables them only for Protocol Mapper:
Configuring the Scrape Rate Limit
Set rateLimit.count and rateLimit.windowMs to cap how many scrape requests a component's metrics endpoint accepts within a time window. This protects the component from excessive scrape traffic, for example, if a Prometheus instance is misconfigured with a scrape interval that is too short.
The following example allows up to ten scrape requests every minute for Protocol Mapper:
Creating a ServiceMonitor for Prometheus Operator
A ServiceMonitor is a Prometheus Operator custom resource that tells Prometheus where and how to scrape a target. Set serviceMonitor.enabled to true on a component to create one, so Prometheus Operator discovers and scrapes the component automatically.
ServiceMonitor is a Kubernetes Custom Ressource and your cluster needs to have the matching Custom Resource Definition installed, or the Helm installation will fail. Ensure you have Prometheus Operator and its Custom Resource Definitions installed before proceeding.
The following example creates a ServiceMonitor for Service Manager with a 30-second scrape interval:
Setting resources for the NATS Prometheus Exporter
NATS requires its own Prometheus exporter sidecar, which is enabled by default. Set compute resources for the sidecar under nats.metrics.resources:
Disabling NATS Prometheus Exporter
NATS requires its own Prometheus exporter sidecar, which is enabled by default. If you want to disable this sidecar, set nats.metrics.enabled to false:
connectware-agent Helm Chart (Standalone)
Use this tab if you deploy the connectware-agent Helm chart on its own. If you deploy it as a subchart of the connectware Helm chart, see the connectware-agent (As Subchart) tab instead.
Enabling Metrics for All Agents
Set defaults under protocolMapperAgentDefaults.metrics. Every agent inherits these values unless overridden per agent:
To disable metrics for every agent, set protocolMapperAgentDefaults.metrics.enabled to false.
Overriding Metrics for a Single Agent
Add a metrics block to an entry in protocolMapperAgents to override the defaults for that agent only. The following example raises the rate limit for high-frequency-agent, while bender-robots keeps the defaults:
To disable metrics for a single agent while keeping them enabled elsewhere, set metrics.enabled to false under that agent's entry.
Creating a ServiceMonitor for Prometheus Operator
Set metrics.serviceMonitor.enabled to true under protocolMapperAgentDefaults.metrics to create a ServiceMonitor for every agent. Set it under a specific agent's entry to create one for that agent only.
connectware-agent Helm Chart (As Subchart)
Use this tab if you deploy the connectware-agent Helm chart as a subchart of the connectware Helm chart. If you deploy it on its own, see the connectware-agent (Standalone) tab instead.
When deployed as a subchart, nest all connectware-agent values under the connectwareAgent key. Agents also inherit any unset metrics value from the parent chart's global.metrics.*.
Inheriting Metrics from the Parent Chart
With no per-agent overrides, every agent inherits global.metrics.* from the connectware Helm chart. The following example disables metrics globally, and every agent picks up those settings automatically:
Overriding Metrics for All Agents
To override the parent's global.metrics for every agent, set values under connectwareAgent.protocolMapperAgentDefaults.metrics:
Overriding Metrics for a Single Agent
Add a metrics block to an entry in connectwareAgent.protocolMapperAgents to override values for that agent only:
Applying Configuration Changes
After editing your values.yaml file, apply the changes by running the helm upgrade command. For details, see Applying Helm Configuration Changes.
Last updated
Was this helpful?