Uptime Monitoring
Know it is down before your customers do.
Your domains are checked at regular intervals and your team is notified when the status changes, before a customer tells you the site will not open.
Why it exists
Learning about an outage from a customer
Most teams find out their site is down when a customer calls. The time until then is not only lost revenue, it is lost trust.
What it does
5 capabilities, one screen.
Regular checks
Availability and response time at minute intervals.
Outage history
When it went down, how long it lasted, what caused it.
Multi-channel alerts
Email, SMS and push, until the on-call person is reached.
Shareable report
Show your uptime to your own customers.
Expected signal (dead man's switch)
If a job that should run is NOT calling you, something is wrong. Your cron, backup or queue worker reports with one line when it finishes and you say "every hour, five minutes of tolerance". A signal arriving at 13:58 does NOT satisfy the 14:00 slot, the schedule is anchored to midnight, not to the minute you created the check.
How it works
In 3 steps.
Add the domain
Prove ownership with a TXT record.
Choose the interval
Check frequency and alert rules.
Connect alerts
Who is notified, on which channel.
Specifications
Does it really do that?
- Checks
- HTTP(S) status and response time
- Reporting
- 24-hour · 7-day · 30-day uptime
- Sharing
- A domain can be shared with another team (watch, not manage)
Questions
Things you might wonder.
How many domains can I monitor?
As many as your package allows. When you need more, raise the module limit.
How do I set up a "backup finished" signal?
Pick "expected signal" as the check type and you get a URL that needs no login. Add a single curl line to the end of your backup script. If the Server Monitoring agent runs on that machine you do not even need the URL: signalbird agent signal db-backup ok.
The job ran but failed. Must I wait for the alarm?
No. If your script sends a fail signal the window is skipped and the alarm fires at once. Silence only catches "the server is gone"; failure the server reports itself.
Works well with