Skip to main content
Antideploy watches whether your app is up and how it is doing: whether it is running, how many requests it served, how many failed and how fast it answered. You and your coding agent read the same things.

What is checked

About every five minutes, Antideploy sweeps every application that has a live deployment and records one of four states. The state appears on your application in the dashboard. Without this check the dashboard would report whatever was true at deploy time: an app that died an hour ago would still show a green badge, which is worse than showing nothing.
A single bad reading is not an outage. An app must fail two consecutive checks before it is reported as down, so a momentary blip in the underlying platform does not page you. An app that is asleep is healthy: the next visit starts it.
The check looks at whether your app is running, crashing or out of memory. It does not load your pages, so a running app that serves errors is something to find in the metrics and the logs. Health is also the last reading from a sweep, not a probe taken at the moment you look: compare when it was last checked with when you deployed before you conclude anything.

Email notifications

You are emailed when an application changes state: when it goes down, and again when it recovers. You are not emailed on every check, and there is no digest. Deploy failures are emailed separately, with the reason, at the moment the deploy fails.

Traffic metrics

Antideploy keeps the traffic your app served: requests, server errors and the error rate, the median and 95th percentile response times in milliseconds, and requests per day. They cover up to 30 days, 7 by default. Your agent reads them with GET /api/v1/metrics when you say the site is slow or failing: logs show what happened, metrics show how often. Requests count everything that reached the app, bots and each asset on a page included, so they are not a count of visitors.

What is not collected

Antideploy does not collect CPU, memory or database connection counts, and there are no alert rules to configure. The only alerts are the down, recovered and deploy-failed emails.If you need more, instrument your app and send the data to a service you control; the credentials go in your environment variables like any other.

Logs

Reading what your application actually printed.