Monitors

What a monitor is, the eight types Glowo can run, and the settings every monitor shares.

A monitor is one thing you want watched: a URL, a host and port, a DNS record, or a job that is supposed to check in. Glowo runs the check on a fixed interval from one or more regions, records every result, and decides when the result is bad enough to tell someone about.

The eight types

HTTP(S)

An HTTP request, its status, timing and body. Websites, APIs, health endpoints.

TCP

A TCP connection opens on a host and port. Databases, message brokers, SMTP.

Ping

ICMP echo reaches a host. Servers, routers, anything without a port to speak to.

DNS

A record resolves, optionally to an expected value. Delegation, MX records, failover records.

DNS Diff

Whether a DNS answer has changed since last time. Detecting hijacks and unplanned record edits.

WebSocket

A WebSocket connects, and optionally echoes a message. Realtime backends, chat, live feeds.

gRPC

The standard gRPC health protocol answers SERVING, or the endpoint speaks gRPC at all. Microservices, internal APIs.

Push

That something called you within a grace period. Cron jobs, backups, batch pipelines.

The type is fixed when you create the monitor. Everything else can be edited later.

Statuses

Status	Meaning
`UP`	The last confirmed result was good.
`DOWN`	The check failed, confirmed. Alerts fire on entry.
`DEGRADED`	Reachable but not healthy — a `warn` assertion failed, or a response-time threshold was crossed.
`CHANGED`	DNS Diff only: the answer changed. Needs acknowledging.
`MAINTENANCE`	Inside a maintenance window. Checks still run; alerts are suppressed.
`PENDING`	Created but not yet checked.

A failing check does not immediately mean DOWN — see Confirmations and flapping.

Settings every monitor shares

Setting	Allowed values	Default
`name`	1–255 characters	—
`interval`	one of 30, 60, 120, 300, 600, 1800, 3600 seconds	—
`retries`	0–10	3
`timeout`	1000–60000 ms	10000
`tags`	up to 20 tags, 50 characters each	none
`description`	up to 1000 characters	none
`regions`	up to 32 selected from the available regions	plan default
`regionStrategy`	`all`, `round_robin`, `random`	`all`
`confirmationsRequired`	1–5	2
`recoveryConfirmationsRequired`	1–5	2
`minFailureDurationMs`	0–300000 ms	30000
`notificationChannelIds`	up to 50 channels	none
`projectIds`	up to 20 projects	none

Pausing

Pausing a monitor stops its checks and its alerts, and excludes the paused time from uptime calculations. It is the right tool for planned work you did not schedule in advance; for planned work you did schedule, use a maintenance window instead so the status page explains the gap to your users.