Intelligent Alerting That Doesn't Wake You Up at 3 AM

Related alerts from across monitors are automatically grouped into single incidents using temporal and semantic correlation - drastically reducing notification volume without hiding real problems. For Web3 infrastructure, a single RPC provider failure can generate dozens of correlated alerts across oracle monitors, block height checks, and DeFi health monitors simultaneously. Without deduplication, that becomes 20 pages for what is effectively one incident.

P0 goes to phone and Telegram in under 30 seconds. P1 goes to Slack. P3 goes to an email digest. Routing rules are configurable per team, per service, and per severity level. Routing configuration is per-team and per-service, so the DeFi team's oracle monitor routes differently from the validator team's sync monitor. Each team defines its own severity thresholds and channel preferences without affecting other teams.

Schedule maintenance windows to suppress expected alerts during planned downtime. Create silence rules based on monitor, chain, or alert type - all without touching config files. Maintenance windows support recurring schedules (weekly maintenance at 02:00–04:00 UTC every Tuesday), so you configure once and the silence applies automatically on every occurrence - no manual re-configuration each week.

P0 alerts automatically create incidents pre-populated with monitor context, affected chain, block height, and alert timeline - the triage step happens before your phone rings. Auto-created incidents include the monitor's alert timeline (all prior state changes), the triggering metric and threshold, the affected chain and block height, and a direct link to the monitor configuration - so your engineer has full context before opening the first runbook step.

Each team member configures their own notification preferences: which channels to use, which severities to receive, and quiet hours for non-critical alerts. Quiet hours allow engineers to mute non-critical alerts during off-hours without missing P0 escalations. P0 alerts always bypass quiet hours - ensuring that on-call engineers are never unreachable for true emergencies.