Monitoring

MinIO KMS provides API endpoints for monitoring deployment health, status, and performance, as well as Prometheus-compatible metrics for integration with monitoring systems.

Health endpoints

Endpoint	Description	Use case
`/version`	Returns the version of the MinIO KMS node	Version verification
`/v1/health/live`	Returns `200 OK` for liveness checks	Kubernetes liveness probe, load balancer health
`/v1/health/ready`	Returns `200 OK` for operational readiness checks	Kubernetes readiness probe
`/v1/health/metrics`	Returns Prometheus-compatible metrics	Prometheus scraping

Prometheus metrics

The /v1/health/metrics endpoint exposes metrics in OpenMetrics v1.0 format, compatible with Prometheus text format.

Scrape configuration

Add the following job to your Prometheus configuration:

scrape_configs:
  - job_name: minkms
    scheme: https
    tls_config:
      ca_file: /path/to/ca.crt
      cert_file: /path/to/client.crt
      key_file: /path/to/client.key
    static_configs:
      - targets:
          - MINKMS_HOST_1:7373
          - MINKMS_HOST_2:7373
          - MINKMS_HOST_3:7373

Replace the targets with your MinIO KMS node addresses.

Available metrics

Client API:

Metric	Type	Description
`http`	counter	Total number of client API requests
`http_request`	counter	Request bytes sent and received
`http_request_inflight`	gauge	Client API requests currently being processed
`http_request_duration`	histogram	Time to process client API requests
`http_request_size`	histogram	Request body sizes
`http_response_size`	histogram	Response body sizes
`http_request_canceled`	counter	Client API requests canceled by the client
`http_request_timeout`	counter	Client API requests that timed out

Internode RPC:

Metric	Type	Description
`rpc_http`	counter	Total number of internode API requests
`rpc_http_request`	counter	Internode request bytes sent and received
`rpc_http_request_duration`	histogram	Time to process internode API requests
`rpc_http_request_size`	histogram	Internode request body sizes
`rpc_http_response_size`	histogram	Internode response body sizes
`rpc_http_request_canceled`	counter	Internode API requests canceled
`rpc_http_request_timeout`	counter	Internode API requests that timed out

Network connections:

Metric	Type	Description
`net_conn`	counter	Total established network connections
`net_conn_bytes`	counter	Bytes sent and received over the network
`net_conn_duration`	histogram	Time network connections remain open
`net_conn_open`	gauge	Currently open network connections

Consensus:

Metric	Type	Description
`consens_heartbeats`	counter	Heartbeats this node performed
`consens_elections`	counter	Elections this node started
`consens_leader_stepdown`	counter	Times this node stepped down from leader role

Runtime:

Metric	Type	Description
`runtime`	info	Version and build information
`runtime_cpu_time`	gauge	Total CPU time in seconds
`runtime_gc_time`	gauge	Total GC time in seconds
`runtime_gc_cycles`	gauge	Total GC cycles
`runtime_mem`	gauge	Total memory in bytes
`runtime_mem_heap`	gauge	Heap memory in bytes
`runtime_goroutines`	gauge	Number of active goroutines
`runtime_threads`	gauge	Number of OS threads

Key metrics to monitor

What to watch	Metric	Alert condition
API errors	`http_request_timeout` or `http_request_canceled` increasing	Clients unable to reach KMS
Internode health	`rpc_http_request_timeout` increasing	Cluster communication issues
Leader stability	`consens_elections` increasing rapidly	Frequent leader elections indicate instability
Resource pressure	`runtime_goroutines` growing unbounded	Possible goroutine leak
Memory	`runtime_mem_heap` approaching system limits	Memory pressure