How to monitor MinIO server with Prometheus Slack

Prometheus is a cloud-native monitoring platform, built originally at SoundCloud. Prometheus offers a multi-dimensional data model with time series data identified by metric name and key/value pairs. The data collection happens via a pull model over HTTP/HTTPS. Targets to pull data from are discovered via service discovery or static configuration.

MinIO exports Prometheus compatible data by default as an authorized endpoint at /minio/prometheus/metrics. Users looking to monitor their MinIO instances can point Prometheus configuration to scrape data from this endpoint.

This document explains how to setup Prometheus and configure it to scrape data from MinIO servers.

Table of Contents

Prerequisites

To get started with MinIO, refer MinIO QuickStart Document. Follow below steps to get started with MinIO monitoring using Prometheus.

1. Download Prometheus

Download the latest release of Prometheus for your platform, then extract it

tar xvfz prometheus-*.tar.gz
cd prometheus-*

Prometheus server is a single binary called prometheus (or prometheus.exe on Microsoft Windows). Run the binary and pass --help flag to see available options

./prometheus --help
usage: prometheus [<flags>]

The Prometheus monitoring server

. . .

Refer Prometheus documentation for more details.

2. Configure authentication type for Prometheus metrics

MinIO supports two authentication modes for Prometheus either jwt or public, by default MinIO runs in jwt mode. To allow public access without authentication for prometheus metrics set environment as follows.

export MINIO_PROMETHEUS_AUTH_TYPE="public"
minio server ~/test

3. Configuring Prometheus

3.1 Authenticated Prometheus config

If MinIO is configured to expose metrics without authentication, you don't need to use mc to generate prometheus config. You can skip reading further and move to 3.2 section.

The Prometheus endpoint in MinIO requires authentication by default. Prometheus supports a bearer token approach to authenticate prometheus scrape requests, override the default Prometheus config with the one generated using mc. To generate a Prometheus config for an alias, use mc as follows mc admin prometheus generate <alias>.

The command will generate the scrape_configs section of the prometheus.yml as follows:

scrape_configs:
- job_name: minio-job
  bearer_token: <secret>
  metrics_path: /minio/prometheus/metrics
  scheme: http
  static_configs:
  - targets: ['localhost:9000']

3.2 Public Prometheus config

If Prometheus endpoint authentication type is set to public. Following prometheus config is sufficient to start scraping metrics data from MinIO.

scrape_configs:
- job_name: minio-job
  metrics_path: /minio/prometheus/metrics
  scheme: http
  static_configs:
  - targets: ['localhost:9000']

4. Update scrape_configs section in prometheus.yml

To authorize every scrape request, copy and paste the generated scrape_configs section in the prometheus.yml and restart the Prometheus service.

5. Start Prometheus

Start (or) Restart Prometheus service by running

./prometheus --config.file=prometheus.yml

Here prometheus.yml is the name of configuration file. You can now see MinIO metrics in Prometheus dashboard. By default Prometheus dashboard is accessible at http://localhost:9090.

6. Configure Grafana

After Prometheus is configured, you can use Grafana to visualize MinIO metrics. Refer the document here to setup Grafana with MinIO prometheus metrics.

List of metrics exposed by MinIO

MinIO server exposes the following metrics on /minio/prometheus/metrics endpoint. All of these can be accessed via Prometheus dashboard. The full list of exposed metrics along with their definition is available in the demo server at https://play.min.io:9000/minio/prometheus/metrics

These are the new set of metrics which will be in effect after RELEASE.2019-10-16*. Some of the key changes in this update are listed below.
- Metrics are bound the respective nodes and is not cluster-wide. Each and every node in a cluster will expose its own metrics.
- Additional metrics to cover the s3 and internode traffic statistics were added.
- Metrics that records the http statistics and latencies are labeled to their respective APIs (putobject,getobject etc).
- Disk usage metrics are distributed and labeled to the respective disk paths.

For more details, please check the Migration guide for the new set of metrics.

The list of metrics and its definition are as follows. (NOTE: instance here is one MinIO node)

NOTES:
> 1. Instance here is one MinIO node.
> 2. s3 requests exclude internode requests.

Default set of information

name description
go_ all standard go runtime metrics
process_ all process level metrics
promhttp_ all prometheus scrape metrics

MinIO node specific information

name description
minio_version_info Current MinIO version with its commit-id
minio_disks_offline Total number of offline disks on current MinIO instance
minio_disks_total Total number of disks on current MinIO instance

Disk metrics are labeled by 'disk' which indentifies each disk

name description
disk_storage_total Total size of the disk
disk_storage_used Total disk space used per disk
disk_storage_available Total available disk space per disk

S3 API metrics are labeled by 'api' which identifies different S3 API requests

name description
s3_requests_total Total number of s3 requests in current MinIO instance
s3_errors_total Total number of errors in s3 requests in current MinIO instance
s3_requests_current Total number of active s3 requests in current MinIO instance
s3_rx_bytes_total Total number of s3 bytes received by current MinIO server instance
s3_tx_bytes_total Total number of s3 bytes sent by current MinIO server instance
s3_ttfb_seconds Histogram that holds the latency information of the requests

Internode metrics only available in a distributed setup

name description
internode_rx_bytes_total Total number of internode bytes received by current MinIO server instance
internode_tx_bytes_total Total number of bytes sent to the other nodes by current MinIO server instance

Apart from above metrics, MinIO also exposes below mode specific metrics

Bucket usage specific metrics

All metrics are labeled by bucket, each metric is displayed per bucket. buckets_objects_histogram is additionally labeled by object_size string which is represented by any of the following values

name description
bucket_usage_size Total size of the bucket
bucket_objects_count Total number of objects in a bucket
bucket_objects_histogram Total number of objects filtered by different sizes

Cache specific metrics

MinIO Gateway instances enabled with Disk-Caching expose caching related metrics.

Global cache metrics

name description
cache_hits_total Total number of cache hits
cache_misses_total Total number of cache misses
cache_data_served Total number of bytes served from cache

Per disk cache metrics

| cache_usage_percent | Total percentage cache usage |
| cache_usage_state | Indicates cache usage is high or low, relative to current cache 'quota' settings |

cache_usage_state holds only two states

Gateway specific metrics

MinIO Gateway instance exposes metrics related to Gateway communication with the cloud backend (S3, Azure & GCS Gateway).

<gateway_type> changes based on the gateway in use can be 's3', 'gcs' or 'azure'. Other metrics are labeled with method that identifies HTTP GET, HEAD, PUT and POST requests to the backend.

name description
gateway_<gateway_type>_requests Total number of requests made to the gateway backend
gateway_<gateway_type>_bytes_sent Total number of bytes sent to cloud backend (in PUT & POST Requests)
gateway_<gateway_type>_bytes_received Total number of bytes received from cloud backend (in GET & HEAD Requests)

Note that this is currently only support for Azure, S3 and GCS Gateway.

MinIO self-healing metrics - self_heal_*

MinIO exposes self-healing related metrics for erasure-code deployments only. These metrics are not available on Gateway or Single Node, Single Drive deployments. Note that these metrics will be exposed only when there is a relevant event happening on MinIO server.

name description
self_heal_time_since_last_activity Time elapsed since last self-healing related activity
self_heal_objects_scanned Number of objects scanned by self-healing thread in its current run. This will reset when a fresh self-healing run starts. This is labeled with the object type scanned
self_heal_objects_healed Number of objects healing by self-healing thread in its current run. This will reset when a fresh self-healing run starts. This is labeled with the object type scanned
self_heal_objects_heal_failed Number of objects for which self-healing failed in its current run. This will reset when a fresh self-healing run starts. This is labeled with disk status and its endpoint

Migration guide for the new set of metrics

This migration guide applies for older releases or any releases before RELEASE.2019-10-23*

MinIO disk level metrics - disk_*

The migrations include

MinIO disk level metrics - disk_storage_*

These metrics have one label.

The migrations include

MinIO network level metrics

These metrics are detailed to cover the s3 and internode network statistics.

The migrations include

Some of the additional metrics added were