// uptime · servers · containers

Know it’s down before your customers do.

relay19 watches your endpoints and your machines. HTTP, TLS, SMTP, DNS, TCP and heartbeat checks, plus a one-line Linux agent for CPU, memory, disk and Docker. Alerts where you want them, an audit trail for the whole team.

Start monitoring → See how it works

$ curl -fsSL https://p.relay19.com/v1/install.sh | sh

✓ relay19-agent installed: reporting cpu · mem · disk · docker

Every layer of your stack, centralized.

From a public endpoint to a port on a private box: pick the check, set the rule, get the alert.

http(s)

Endpoint checks

Request any URL and assert the response: status code, redirects followed, latency recorded.

200 in 38ms · 24h 100%

keyword

Keyword monitoring

Assert a phrase is on the page, or that one never appears. Catches the broken checkout and the defaced page that still answer 200.

✓ "Add to cart" present · "maintenance" absent

tls

Certificate watch

Valid chain and days to expiry, with a warning well ahead of the renewal date.

✓ TLSv1.3 · chain valid · expires in 74d

dns

Record answers

Resolve A, AAAA, CNAME, MX or TXT and compare against what you expect.

A 76.76.21.21 · match

smtp

Mail server checks

Talk to the mail server, expect the banner, optionally over TLS.

220 mx1 ready · STARTTLS

tcp

Port reachability

Open a socket on any host and port. Up means reachable, with latency.

connect :5432 · open in 2ms

heartbeat

Dead man's switch

Your cron jobs and pipelines ping us. Miss the grace window and you hear about it.

last ping 42s ago · grace 300s

relay19-agent

Server monitoring

One line installs the agent: CPU, memory, load, every real disk, and your Docker containers.

cpu

mem

disk

docker

node-01 · agent v0.4.0 · load 0.42 · 12 containers running

From first miss to resolved, on the record.

One blip never wakes anyone. relay19 confirms a failure before it pages, alerts the channels you chose, and closes the loop the moment the check passes again.

14:02:11 fail 1/3

First miss

A check comes back wrong. One miss could be a blip, so nobody gets paged.

14:04:11 fail 3/3

Incident opens

Consecutive misses hit the threshold you set, three by default. Now it's real.

14:04:12 alert

Alerts fan out

Slack, email and every other channel fire with the monitor, the cause and the timestamp.

14:11:11 200 ok

Recovery seen

The first passing check resolves the incident and notifies again. No manual ack.

+9m 00s resolved

On the record

Downtime, cause and resolution stay in the incident log and the team audit trail.

✓ threshold per monitor · resolve on first pass · every transition notified · full history in the docs

Alerts land where you already are.

Eight channels ship today. Each one gets the incident formatted for that service rather than a wall of raw JSON, and fires again the moment the check recovers. Scope any channel to a monitor tag and only that team hears about it.

Email Whoever is on call, up to ten addresses.

Slack A formatted message in the channel you pick.

Discord A coloured embed in your server.

Microsoft Teams An adaptive card in the channel.

Telegram A message from a bot you control.

ntfy A push straight to your phone.

PagerDuty Opens an alert on your rota, closes it on recovery.

Webhook Signed JSON to anything you run.

✓ opened and resolved both notified · test any channel before you trust it · where to find each URL in the setup guide

Checked from around the world

Checks run from probes on three continents and regional results are compared, so an alert means your service is down, not one network having a bad minute. Picking the locations a monitor checks from, and rotating between them, starts on Starter, with the whole network on Team and up.

los1 Los Angeles, US

man1 Manassas, US

lon1 London, GB

nbg1 Nuremberg, DE

sof1 Sofia, BG

hel1 Helsinki, FI

sin1 Singapore, SG

There are four ways to write your address. They should all end up in one place.

When they do not, visitors hit errors and search engines index two copies of the same site. Our free site check follows all four, then reads the certificate behind them, the headers they return, and the records that decide whether anyone can send mail as your domain. No signup, and the report is never stored.

✓ redirects · certificate · security headers · SPF, DMARC and DKIM · what each check means