Monitoring and Alerting

Administration UI

Silo provides an administration UI through minio_admin_port, which defaults to 9001. Access it directly at https://<node-ip>:9001.

Some configuration templates also expose the administration entry point at m.pigsty. Login credentials come from minio_access_key and minio_secret_key.

Collection Paths

Silo retains the stable job="minio", cls, ins, ip, and instance identity labels and uses flavor="silo":

Each instance’s FileSD target is written to /infra/targets/minio/<minio_cluster>-<minio_seq>.yml.

Silo registers one Metrics V3 root endpoint, which exposes cluster, system, API, and aggregated usage metrics. Pigsty drops samples with a non-empty bucket label and does not register separate per-bucket or replication endpoints, keeping time-series cardinality under control.

Grafana Dashboards

Pigsty provides the compatibility-named MinIO Overview / MinIO Instance dashboards for Silo Metrics V3 data, system logs, and instance state.

Alert Rules

The current files/victoria/rules/minio.yml defines five alerts for Silo:

Key expressions use Metrics V3 names:

minio_up < 1

max by (cls) (
  avg_over_time(minio_cluster_health_nodes_offline_count{job="minio"}[5m])
) > 0

max by (cls) (
  avg_over_time(minio_cluster_health_drives_offline_count{job="minio"}[5m])
) > 0

min by (cls) (
  minio_cluster_erasure_set_overall_health{job="minio"}
  or (minio_cluster_erasure_set_overall_write_quorum{job="minio"} * 0)
) < 1

max by (cls) (
  1 - (
    (minio_cluster_health_capacity_usable_free_bytes{job="minio"}
     or (minio_cluster_health_capacity_usable_total_bytes{job="minio"} * 0))
    / minio_cluster_health_capacity_usable_total_bytes{job="minio"}
  )
) > 0.90