How confirmation and recovery periods work

Not every failed check means your site is down. Network blips, temporary server issues, and other transient problems can cause false positives. OnlineOrNot uses confirmation and recovery periods to distinguish real outages from noise.

The problem with instant alerts

Imagine your site has a brief hiccup - maybe a 5-second database connection timeout. Without any verification:

1. OnlineOrNot checks your site
2. The check fails
3. You get paged at 3 AM
4. By the time you check, everything's fine

This is a false positive, and it erodes trust in your monitoring.

Confirmation period

When a check first fails, OnlineOrNot doesn't immediately alert you. Instead, it enters a "confirming" state:

1. First failure detected - check enters "confirming" state
2. OnlineOrNot double-checks from multiple regions
3. If still failing after the confirmation period (default: 60 seconds), an incident is declared
4. You receive an alert

This eliminates most transient failures while still catching real outages quickly.

Recovery period

Recovery works similarly:

1. A down check starts passing - enters "recovering" state
2. OnlineOrNot verifies from multiple regions
3. If still up after the recovery period (default: 3 minutes), the incident ends
4. You receive a recovery alert

The longer recovery period (vs. confirmation) prevents "flapping" - rapid up/down/up cycles that flood your alerts.

When to adjust these settings

Shorter confirmation period:

• Critical payment or auth endpoints
• When you need to know about issues immediately
• Sites with excellent reliability (false positives are rare)

Longer confirmation period:

• Landing pages and marketing sites
• Services with known brief maintenance windows
• Sites behind flaky network infrastructure

Longer recovery period:

• Services that take time to fully warm up
• Sites with caching that may show stale "up" responses
• When you want to be certain the fix is stable

Configuring in OnlineOrNot

When adding or editing an uptime check, find these settings under advanced options:

You can set:

• Immediate start - Alert on first failure (no confirmation)
• Custom confirmation period - Wait N seconds before alerting
• Immediate recovery - End incident on first success
• Custom recovery period - Wait N minutes before declaring recovery