Configuration¶

Glint supports two configuration methods: a YAML config file (recommended for multi-instance setups) and environment variables (convenient for single-instance deployments).

Config File¶

Pass a config file with --config:

glint --config /etc/glint/glint.yml

Full Reference¶

Server Settings¶

listen: ":3800"           # (1)!
db_path: "/data/glint.db" # (2)!
log_level: "info"         # (3)!
log_format: "text"        # (4)!
history_hours: 48         # (5)!
worker_pool_size: 4       # (6)!

1. Address and port to bind the HTTP server. Default: :3800
2. Path to the SQLite database file. Must be writable. Default: glint.db
3. Log verbosity: debug, info, warn, error. Default: info
4. Log output format: text (human-readable) or json (structured). Default: text
5. Hours of metric history to retain for sparkline charts. Default: 48
6. Maximum concurrent API calls across all collectors. Default: 4

PVE Instances¶

At least one PVE instance is required.

pve:
  - name: "main"                    # (1)!
    host: "https://pve:8006"        # (2)!
    token_id: "glint@pam!monitor"   # (3)!
    token_secret: "xxx-xxx-xxx"     # (4)!
    insecure: true                  # (5)!
    poll_interval: "15s"            # (6)!
    disk_poll_interval: "1h"        # (7)!
    ssh:                            # (8)!
      host: "192.168.1.215"
      user: "root"
      key_path: "/config/ssh/id_ed25519"

1. Unique label for this instance. Used in UI and alerts.
2. PVE API endpoint URL (include port).
3. API token in user@realm!tokenname format.
4. Token secret from pveum user token add.
5. Skip TLS certificate verification. Set true for self-signed certs. Default: false
6. How often to poll node and guest metrics. Default: 15s
7. How often to poll S.M.A.R.T. disk data (slow operation). Default: 1h
8. Optional SSH connection for CPU temperature monitoring.

PBS Instances¶

PBS monitoring is optional.

pbs:
  - name: "main-pbs"                 # (1)!
    host: "https://pbs:8007"          # (2)!
    token_id: "glint@pbs!monitor"     # (3)!
    token_secret: "xxx-xxx-xxx"       # (4)!
    insecure: true                    # (5)!
    datastores: ["homelab"]           # (6)!
    poll_interval: "5m"               # (7)!

1. Unique label for this PBS instance.
2. PBS API endpoint URL (include port).
3. API token in user@realm!tokenname format.
4. Token secret from proxmox-backup-manager user generate-token.
5. Skip TLS certificate verification. Default: false
6. Datastores to monitor. Empty list = monitor all discovered datastores.
7. How often to poll backup data. Default: 5m

Notifications¶

notifications:
  - type: ntfy                        # (1)!
    url: "http://ntfy:8080"           # (2)!
    topic: "homelab-alerts"           # (3)!

  - type: webhook                     # (4)!
    url: "https://hooks.example.com/glint"
    method: "POST"                    # (5)!
    headers:                          # (6)!
      Authorization: "Bearer xxx"

1. Provider type: ntfy or webhook.
2. ntfy server URL.
3. ntfy topic name.
4. Generic webhook --- POSTs the full alert as JSON.
5. HTTP method: POST (default) or PUT.
6. Custom headers for authentication.

Alert Rules¶

All alert rules are optional. Defaults are applied if omitted.

alerts:
  node_cpu_high:
    threshold: 90           # Percent CPU usage
    duration: "5m"          # Must sustain for this long
    severity: "warning"

  guest_down:
    grace_period: "2m"      # Ignore brief restarts
    severity: "critical"

  backup_stale:
    max_age: "36h"          # Alert if last backup older than this
    severity: "warning"

  disk_smart_failed:
    severity: "critical"

  datastore_full:
    threshold: 85           # Percent datastore usage
    severity: "warning"

Environment Variables¶

For a single PVE + PBS instance, you can skip the config file entirely:

PVE¶

PBS¶

Notifications¶

Server¶

Cluster Setup¶

Multi-Node PVE Cluster¶

Point Glint at any one node in your cluster. It auto-discovers all nodes via the /nodes endpoint and deduplicates guests by cluster ID.

pve:
  - name: "prod-cluster"
    host: "https://ANY_CLUSTER_NODE:8006"
    token_id: "glint@pam!monitor"
    token_secret: "xxx"
    insecure: true

Multiple Independent PVE Instances¶

For separate (non-clustered) Proxmox hosts, add multiple entries. Each needs its own API token.

pve:
  - name: "homelab"
    host: "https://192.168.1.215:8006"
    token_id: "glint@pam!monitor"
    token_secret: "xxx"
    insecure: true

  - name: "remote"
    host: "https://10.0.0.50:8006"
    token_id: "glint@pam!monitor"
    token_secret: "yyy"
    insecure: true

Version Info¶

Glint embeds build metadata in the binary:

glint --version

glint v0.1.0
  commit:    abc1234def5678 (clean)
  built:     2026-02-16T12:00:00Z
  go:        go1.26
  platform:  linux/amd64

At startup, the same info is logged:

level=INFO msg="starting glint" version=v0.1.0 commit=abc1234def5678 built=2026-02-16T12:00:00Z dirty=clean go=go1.26 listen=:3800

Security Considerations¶

• Tokens are read-only. PVEAuditor and Audit roles cannot modify anything.
• Use insecure: true only for self-signed certificates. If you have proper TLS certs, set it to false.
• Store tokens securely. Use Docker secrets, environment variables from a secrets manager, or file-based secrets for production deployments.
• Network isolation. Glint only needs access to the PVE/PBS API ports (8006/8007). It does not need SSH access unless you enable temperature monitoring.
• Revoke tokens if compromised: pveum user token remove glint@pam monitor