Skip to content

Latest commit

 

History

History
45 lines (32 loc) · 2.38 KB

monitoring.md

File metadata and controls

45 lines (32 loc) · 2.38 KB
Raw

Monitoring the NATS-Replicator

The nats-replicator provides optional HTTP/s monitoring. When configured with a monitoring port the server will provide two HTTP endpoints:

You can also just navigate to the monitoring port, i.e. http://localhost:9090, and a page will point you at these two paths.

/varz

The /varz endpoint returns a JSON encoded set of statistics for the server. These statistics are wrapped in a root level object with the following properties:

  • start_time - the start time of the replicator, in the replicator's timezone.
  • current_time - the current time, in the replicator's timezone.
  • uptime - a string representation of the replicator's up time.
  • http_requests - a map of request paths to counts, the keys are /, /varz and /healthz.
  • connectors - an array of statistics for each connector.

Each object in the connectors array, one per connector, will contain the following properties:

  • name - the name of the connector, a human readable description of the connector.
  • id - the connectors id, either set in the configuration or generated at runtime.
  • connects - a count of the number of times the connector has connected.
  • disconnects - a count of the number of times the connector has disconnected.
  • bytes_in - the number of bytes the connector has received, may differ from received due to headers and encoding.
  • bytes_out - the number of bytes the connector has sent, may differ from received due to headers and encoding.
  • msg_in - the number of messages received.
  • msg_out - the number of messages sent.
  • count - the total number of requests for this connector.
  • rma - a running moving average of the time required to handle each request. The time is in nanoseconds.
  • q50 - the 50% quantile for response times, in nanoseconds.
  • q75 - the 75% quantile for response times, in nanoseconds.
  • q90 - the 90% quantile for response times, in nanoseconds.
  • q95 - the 95% quantile for response times, in nanoseconds.

Pass the URL property pretty=true to get formatted JSON. For example, http://localhost:8080/varz?pretty=true.

/healthz

The /healthz endpoint is provided for automated up/down style checks. The server returns an HTTP/200 when running and won't respond if it is down.