Skip to content

Logging

Glint uses Go's structured slog library. All log output goes to stderr.

Log Format

Glint supports two output formats, configured via log_format in the config file or the GLINT_LOG_FORMAT environment variable.

Human-readable key=value format. Best for local development and interactive debugging.

time=2026-02-16T12:00:00.000Z level=INFO msg="starting glint" version=0.1.0 commit=abc1234 listen=:3800
time=2026-02-16T12:00:00.015Z level=INFO msg="collector started" name=pve:main interval=15s
time=2026-02-16T12:00:15.032Z level=DEBUG msg="PVE collection complete" instance=main nodes=1 guests=8

Structured JSON, one object per line. Best for systemd/journald, Docker json-file driver, and log aggregators.

{"time":"2026-02-16T12:00:00.000Z","level":"INFO","msg":"starting glint","version":"0.1.0","commit":"abc1234","listen":":3800"}
{"time":"2026-02-16T12:00:00.015Z","level":"INFO","msg":"collector started","name":"pve:main","interval":"15s"}

Set via config:

log_format: "json"

Or environment variable:

export GLINT_LOG_FORMAT=json

Log Levels

Level What it logs
error Failures that affect functionality (DB errors, API failures)
warn Degraded behavior (alerts fired, cluster detection fallback)
info Lifecycle events (startup, shutdown, prune results)
debug Per-poll collection details, HTTP request logs

Set via config (log_level: "debug") or env var (GLINT_LOG_LEVEL=debug).

Production recommendation

Use log_level: "info" and log_format: "json" for production. This gives you structured logs that are easy to filter while keeping volume manageable.

Docker Logging

Docker captures container stdout/stderr automatically. The json-file logging driver (default) stores logs on disk. The recommended docker-compose.yml includes rotation settings:

logging:
  driver: json-file
  options:
    max-size: "10m"
    max-file: "3"

Viewing Logs

# Follow logs
docker compose logs -f glint

# Filter by level (requires json log format)
docker compose logs glint | jq 'select(.level == "ERROR")'

# Show last 100 lines
docker compose logs --tail=100 glint

systemd / journald

When running as a systemd service (bare metal or Podman Quadlet), set log_format: "json" for best integration. systemd captures stderr and stores it in the journal.

Viewing Logs

# Follow logs
journalctl -u glint -f

# View logs since last boot
journalctl -u glint -b

# Filter by priority (requires json format for structured fields)
journalctl -u glint --priority=err

# Export as JSON
journalctl -u glint -o json

# Show logs from last hour
journalctl -u glint --since="1 hour ago"

log_level: "info"
log_format: "json"

This gives you structured logs that journald can index and filter, while keeping the volume manageable.

Log Fields

All log entries include these base fields:

Field Description Example
time ISO 8601 timestamp 2026-02-16T12:00:00.000Z
level Log level INFO, ERROR
msg Human-readable message "PVE collection complete"

Collector logs include additional context:

Field Description Example
instance PVE/PBS instance name "main"
name Collector identifier "pve:main"
interval Poll interval "15s"
nodes Number of nodes polled 1
guests Number of guests found 8
duration Collection duration "1.234s"

Alert logs include:

Field Description Example
alert_type Rule that fired "node_cpu_high"
severity Alert severity "warning"
subject What triggered it "main/pve"