What it is for
A watched URL is an address meulog opens every so often to see whether it responds. It is monitoring from the outside in: the error that never arrives because the whole service went down shows up on no error list.
What you record
- Name and URL.
- Method:
GETorHEAD. WhereHEADworks, it is cheaper for both sides. - Interval in minutes: how often to check.
- Timeout in seconds: how long to wait before calling it down.
- Expected status: not every healthy URL returns 200. A portal that always redirects returns 302, and expecting 200 from it would mean one alert a day, every day.
- Expected content (optional): a snippet that must appear in the response. It is what separates "the server answered" from "the application is up": a load balancer error page also returns 200.
- Failures before notifying: how many checks in a row must fail before firing. Networks blink; one isolated failure is not an outage.
What meulog decides for you
Redirects are followed to the destination, with a hop limit. And each check stores the response time, not just yes or no: rising latency is the warning that comes before the outage.
The two events
url_fora when it goes down, url_voltou when it comes
back. The second matters as much as the first: without it, whoever was woken at
three in the morning does not know they can go back to sleep.
Where to see the result
This is the record. The current state and the history are under Uptime.