The alert storm
Dozens of near-identical alerts wake the on-call again and again. TokayOps groups and deduplicates them into one Alert Group.
TokayOps is a lightweight incident-management layer between Alertmanager and your on-call engineer. It groups alerts, finds the right person by schedule, escalates until someone answers, and lets them Ack / Resolve right in Slack. One container, your Postgres, your data.
See how it worksThe problem
Four failure modes turn a quiet night into a bad one. TokayOps is built to remove each.
Dozens of near-identical alerts wake the on-call again and again. TokayOps groups and deduplicates them into one Alert Group.
Wrong team, stale rotation, nobody owns it. Routing by team + severity plus on-call schedules always find the right human.
Multi-step escalation with parallel fan-out - the whole shift is paged at once, with delays and timeouts. Nothing hangs on one unanswered ping.
Enterprise on-call SaaS bills by the head and owns your data. TokayOps is one container against your Postgres - your stack, your data, no per-seat pricing.
How it works
Five stages, no glue scripts. Everything below runs inside one process against your Postgres.
Alertmanager POSTs grouped alerts to the TokayOps webhook, authenticated by a stored secret.
Create or merge one Alert Group by groupKey (fingerprint fallback). The storm becomes a single incident.
Team comes from labels, severity selects the escalation policy, and the L1 schedule resolves the current on-call.
A staged job DMs the whole on-call group at once, plus the channel - with delays, timeouts and continue-on-failure.
Block Kit and Telegram buttons: Ack takes ownership and pauses escalation, Resolve closes it. The message updates in place.