English | [Русский](CONFIG_REFERENCE.ru.md)

# Monlet: Configuration Reference

Monlet consists of agents running on hosts and a central server. The agent generates its local key in `state_dir/agent.key`; the server sees a new key as `pending` until an operator accepts the agent in the UI.

## Agent TOML

File: usually `/etc/monlet/agent.toml`.

```toml
# Optional. If omitted, the OS hostname is used.
# agent_id = "host-01"

state_dir = "/var/lib/monlet-agent"

[labels]
env = "prod"
role = "api"

[server]
enabled = true
url = "https://monlet.example.com"
heartbeat_interval = "10s"
batch_interval = "10s"

[metrics]
enabled = true
listen = "127.0.0.1:9465"

[[checks]]
id = "disk_root"
name = "Root disk usage"
command = '''
/usr/local/lib/monlet/check_disk_root.sh --warn 80 --crit 90
'''
interval = "60s"
timeout = "10s"
notifications_enabled = true
resource_limits = { cpu_time = "5s", memory = "256MiB", open_files = 128 }

[[checks]]
id = "daily_backup"
name = "Daily backup freshness"
command = "/usr/local/lib/monlet/check_backup_freshness.sh"
cron = "CRON_TZ=UTC 15 6 * * *"
timeout = "2m"
```

Keys:

- `agent_id` - stable agent ID; optional, defaults to hostname.
- `state_dir` - local state, agent key, and spool.
- `[labels]` - inventory labels; do not put secrets here.
- `server.enabled` - send heartbeats/events to the server.
- `server.url` - server base URL.
- `metrics.enabled` - expose Prometheus `/metrics`.
- `metrics.listen` - `/metrics` listen address.
- `checks[].id` - stable check ID.
- `checks[].command` - shell command string; one-line `"..."` and multi-line `'''...'''` forms are supported.
- `checks[].interval` - run period (`60s`, `5m`, `1h`); required when `checks[].cron` is not set.
- `checks[].cron` - cron-style schedule; required when `checks[].interval` is not set. Standard 5-field specs, `@hourly`/`@daily`, and `CRON_TZ=...` are supported; seconds-format specs, `@every`, and `@reboot` are rejected.
- `checks[].timeout` - timeout; timed-out checks return `critical`.
- `checks[].notifications_enabled` - allow server-side notifications for this check.
- `checks[].resource_limits` - optional process limits: CPU time, virtual memory, open files.

Each check must define exactly one schedule mode: `interval` or `cron`. Interval checks run immediately after start/reload and then on the configured period. Cron checks wait for the next cron slot and do not run immediately on start.

Exit code: `0=ok`, `1=warning`, `2=critical`, `3+=unknown`.

Agent env:

- `MONLET_AGENT_MAX_CHECKS` - maximum checks in config (default 256). This bounds `{check_id}` metric cardinality. Must be a positive integer; exceeding the limit is a config load error.

Reload:

- `systemctl reload monlet-agent` reloads the local TOML via `SIGHUP`.
- Invalid config is not applied; the old config keeps running.
- A running check is not killed on reload; it runs until completion or its `timeout`.
- When a schedule changes, an interval check runs after the current in-flight run; a cron check waits for the next cron slot.
- Changing `agent_id`, `state_dir`, `server.*`, or `metrics.*` requires restart.

## Server env

File: `.env` or container environment.

```env
MONLET_AUTH_TOKEN=replace-with-random-ui-token
MONLET_DATABASE_URL=postgresql+asyncpg://monlet:monlet@postgres:5432/monlet
MONLET_LOG_LEVEL=INFO

MONLET_HOST=0.0.0.0
MONLET_PORT=8000

MONLET_BODY_LIMIT_BYTES=1048576
MONLET_OUTPUT_MAX_BYTES=8192
MONLET_MAX_BATCH_EVENTS=200

MONLET_ENABLE_DETECTOR=true
MONLET_STALE_AFTER_SEC=20
MONLET_DEAD_AFTER_SEC=30
MONLET_DETECTOR_TICK_SEC=5

MONLET_DB_POOL_SIZE=10
MONLET_DB_MAX_OVERFLOW=10
MONLET_DB_POOL_TIMEOUT_SEC=30
MONLET_DB_POOL_RECYCLE_SEC=1800

MONLET_EVENTS_RETENTION_MONTHS=36
MONLET_EVENTS_FUTURE_PARTITIONS=3
MONLET_PARTITION_MAINTENANCE_INTERVAL_SEC=3600
MONLET_OUTBOX_RETENTION_MAX_ROWS=50000
MONLET_OUTBOX_PRUNE_BATCH_SIZE=5000

MONLET_EVENT_DEDUP_RETENTION_DAYS=30
MONLET_EVENT_DEDUP_PRUNE_BATCH_SIZE=5000

MONLET_MAX_PENDING_AGENTS=10000
MONLET_PENDING_AGENT_TTL_DAYS=7
MONLET_PENDING_AGENT_PRUNE_BATCH_SIZE=1000

MONLET_ENABLE_NOTIFIER_WORKER=true
MONLET_NOTIFIER_TICK_SEC=5
MONLET_NOTIFIER_BATCH_SIZE=20
MONLET_NOTIFIER_CONCURRENCY=4
MONLET_NOTIFIER_MAX_ATTEMPTS=8
MONLET_NOTIFIER_LEASE_SEC=300
MONLET_NOTIFIER_HTTP_TIMEOUT_SEC=10
MONLET_NOTIFIER_INCLUDE_OUTPUT=true
MONLET_NOTIFIER_MAX_OUTPUT_BYTES=1024
MONLET_NOTIFICATION_TIME_ZONE=UTC

MONLET_NOTIFIER_DEBUG_ENABLED=true
MONLET_NOTIFIER_TELEGRAM_ENABLED=false
MONLET_NOTIFIER_WEBHOOK_ENABLED=false
MONLET_NOTIFIER_ALERTMANAGER_ENABLED=false
```

Basics:

- `MONLET_AUTH_TOKEN` - Bearer token(s) for UI/operator API, not for agents. Space/comma-separated lists support overlap rotation.
- `MONLET_DATABASE_URL` - PostgreSQL DSN (`postgresql+asyncpg://...`).
- `MONLET_LOG_LEVEL` - log level.
- `MONLET_HOST` / `MONLET_PORT` - HTTP server host and port (default `0.0.0.0:8000`).

Ingestion limits:

- `MONLET_BODY_LIMIT_BYTES` - maximum request body size (default 1 MiB = 1048576).
- `MONLET_OUTPUT_MAX_BYTES` - stored event `output` limit (default 8192).
- `MONLET_MAX_BATCH_EVENTS` - maximum events in one batch (default 200).

Liveness detector:

- `MONLET_ENABLE_DETECTOR` - enable stale/dead background recalculation (default `true`).
- `MONLET_STALE_AFTER_SEC` - seconds without heartbeat before an agent becomes `stale`.
- `MONLET_DEAD_AFTER_SEC` - seconds without heartbeat before an agent becomes `dead`.
- `MONLET_DETECTOR_TICK_SEC` - liveness recalculation period.

DB pool:

- `MONLET_DB_POOL_SIZE` - persistent pool connections (default 10).
- `MONLET_DB_MAX_OVERFLOW` - extra connections above the pool (default 10).
- `MONLET_DB_POOL_TIMEOUT_SEC` - wait time for a free connection (default 30).
- `MONLET_DB_POOL_RECYCLE_SEC` - recycle connections by age (default 1800).

Storage and maintenance:

- `MONLET_EVENTS_RETENTION_MONTHS` - event partition retention (default 36).
- `MONLET_EVENTS_FUTURE_PARTITIONS` - future monthly partitions to pre-create (default 3).
- `MONLET_PARTITION_MAINTENANCE_INTERVAL_SEC` - partition maintenance and prune period (default 3600).
- `MONLET_OUTBOX_RETENTION_MAX_ROWS` - upper bound for outbox rows; extra terminal rows are pruned (default 50000).
- `MONLET_OUTBOX_PRUNE_BATCH_SIZE` - terminal outbox prune batch size (default 5000).
- `MONLET_EVENT_DEDUP_RETENTION_DAYS` - `event_id` idempotency record retention (default 30).
- `MONLET_EVENT_DEDUP_PRUNE_BATCH_SIZE` - dedup prune batch size (default 5000).
- `MONLET_MAX_PENDING_AGENTS` - maximum unaccepted new agents (default 10000).
- `MONLET_PENDING_AGENT_TTL_DAYS` - old pending-key cleanup TTL (default 7).
- `MONLET_PENDING_AGENT_PRUNE_BATCH_SIZE` - old pending-key prune batch size (default 1000).

Notifier worker:

- `MONLET_ENABLE_NOTIFIER_WORKER` - enable notification delivery worker (default `true`).
- `MONLET_NOTIFIER_TICK_SEC` - outbox poll period (default 5).
- `MONLET_NOTIFIER_BATCH_SIZE` - outbox rows claimed per tick (default 20).
- `MONLET_NOTIFIER_CONCURRENCY` - parallel deliveries inside one tick (default 4).
- `MONLET_NOTIFIER_MAX_ATTEMPTS` - maximum delivery attempts before permanent failure (default 8).
- `MONLET_NOTIFIER_LEASE_SEC` - row lease duration during delivery, to prevent another worker from claiming it (default 300).
- `MONLET_NOTIFIER_HTTP_TIMEOUT_SEC` - HTTP delivery timeout (default 10).
- `MONLET_NOTIFIER_INCLUDE_OUTPUT` - include `output` text in notifications (default `true`).
- `MONLET_NOTIFIER_MAX_OUTPUT_BYTES` - notification output truncation after redaction (default 1024).
- `MONLET_NOTIFICATION_TIME_ZONE` - timezone for human-readable notification timestamps.

Notifier channels:

- Telegram: `MONLET_NOTIFIER_TELEGRAM_ENABLED=true`, `MONLET_NOTIFIER_TELEGRAM_TOKEN`, `MONLET_NOTIFIER_TELEGRAM_CHAT_ID`.
- Webhook: `MONLET_NOTIFIER_WEBHOOK_ENABLED=true`, `MONLET_NOTIFIER_WEBHOOK_URL`, optional `MONLET_NOTIFIER_WEBHOOK_TOKEN`.
- Alertmanager: `MONLET_NOTIFIER_ALERTMANAGER_ENABLED=true`, `MONLET_NOTIFIER_ALERTMANAGER_URL`.
- Debug/log: `MONLET_NOTIFIER_DEBUG_ENABLED=true`.

## Web env

The web UI reads API/auth/timezone settings from env. Without auth settings, read-only pages are available locally, but admission mutations return `403`.

```env
MONLET_API_BASE_URL=http://127.0.0.1:8000
MONLET_API_TOKEN=replace-with-server-token
MONLET_WEB_TIME_ZONE=UTC
MONLET_WEB_AUTH_USERNAME=admin
MONLET_WEB_AUTH_PASSWORD=replace-with-strong-password
MONLET_WEB_SESSION_SECRET=replace-with-random-32-bytes-min
MONLET_WEB_SESSION_TTL_SEC=604800
NEXT_PUBLIC_MONLET_POLL_MS=10000
# Or, instead of local login, trust reverse-proxy auth:
# MONLET_WEB_TRUST_PROXY_AUTH=true
```

- `MONLET_API_BASE_URL` - server base URL for server-side fetch.
- `MONLET_API_TOKEN` - Bearer token for UI requests to the server; it is not sent to the browser.
- `MONLET_WEB_TIME_ZONE` - IANA timezone used to render timestamps (default `UTC`).
- `MONLET_WEB_AUTH_USERNAME` / `MONLET_WEB_AUTH_PASSWORD` - local cookie-login credentials.
- `MONLET_WEB_SESSION_SECRET` - session signing secret, at least 32 bytes; shorter values make UI auth misconfigured.
- `MONLET_WEB_SESSION_TTL_SEC` - session lifetime (default 604800 = 7 days).
- `MONLET_WEB_TRUST_PROXY_AUTH` - `true` trusts the `X-Forwarded-User` header from a reverse proxy. Mutually exclusive with local login; configuring both modes is an error (UI returns `500`).
- `NEXT_PUBLIC_MONLET_POLL_MS` - browser polling interval in milliseconds (default 10000); for production Next.js builds, this is a build-time variable.

## Minimal Flow

1. Server starts with PostgreSQL and `MONLET_AUTH_TOKEN`.
2. Agent starts from TOML, creates `state_dir/agent.key`, and sends heartbeat.
3. The new agent appears in UI as `pending`.
4. Operator accepts the agent.
5. After accept, the server starts accounting for events, checks, incidents, and notifications.