Skip to content

Monitoring and Alert Management ​

EMQX Neuron provides metric monitoring and alert events, for tracking the state of an instance and being notified when something goes wrong.

TIP

Both are disabled by default and are configured and queried through the HTTP API only; there is no configuration screen in the console. Whether a configuration has taken effect is also checked through the API.

Monitoring metrics ​

The metrics that can be collected fall into three categories:

CategoryContents
Collection engineNumber of southbound and northbound drivers, driver link states, number of drivers in error
Data processing engineRecords in and out per rule, number of stopped rules
System resourcesCPU, memory, and so on

The monitoring configuration selects which metrics are wanted. They can be retrieved two ways:

MethodDescription
Push to PushgatewayWith a Pushgateway address configured, EMQX Neuron pushes metrics to it for Prometheus to scrape
API queryRead the current metrics directly from the API

For the configuration and query endpoints, see the Monitor API.

Alert events ​

EMQX Neuron evaluates alert conditions by polling. The alert configuration selects which rules are enabled and how many consecutive polls are required to raise and to clear an alert — N is the number of consecutive abnormal polls before the event is raised, and P the number of consecutive normal polls before it clears.

Alert typeSubjectTrigger conditionRaised afterCleared after
Driver node abnormal (southbound and northbound)A single driverThe driver is running but not connectedN consecutive pollsP consecutive normal polls
Stream processing rule abnormalA single ruleThe error count of any source, operator, or sink in the rule increasesN consecutive pollsP consecutive normal polls
EMQX Neuron restartThe current instanceEMQX Neuron restarts1 pollNot applicable

Alert events can be retrieved two ways:

MethodDescription
Push to a webhookWith a webhook address configured, EMQX Neuron pushes alert events to it
API queryQuery the most recent alert events

For the configuration and query endpoints, see the Alert Rule API.