Summary
Sentry's only alert rule for this project is its own shipped default ("Send a notification for high priority issues"), and its only action is email to issue owners, falling through to active members. All 28 currently-open issues are scored high-priority by Sentry, and all 28 are unassigned — so notification has been going to a fallback nobody watches in real time. Wire a real-time action instead.
Requirements
- Configure Sentry's native Discord integration (org-level, via Sentry's Integrations page) and add "Send a Discord notification" as an action on the existing (or a new) issue alert rule — ideally the same Discord channel/webhook family Alertmanager already posts to, so an operator has one place to watch, not two.
- Once PagerDuty exists (see the sibling Alertmanager sub-issue under this epic), also configure Sentry's native PagerDuty integration and route
severity: critical-equivalent issues there. Sentry has its own priority/severity concept independent of Alertmanager's — map it deliberately, don't assume they line up 1:1.
- Decide and document: should every new high-priority issue notify, or only ones crossing some volume/recurrence threshold? The current default (any high-priority issue) is what's already configured and has produced 28 open issues with no differentiation between a 1-event one-off and a 4,000-event, 12-day-ongoing incident — consider whether Sentry's alert-rule filters (event frequency, user count) should be used to separate "notify immediately" from "visible on next triage pass."
Deliverables
Part of #5007.
Summary
Sentry's only alert rule for this project is its own shipped default ("Send a notification for high priority issues"), and its only action is email to issue owners, falling through to active members. All 28 currently-open issues are scored high-priority by Sentry, and all 28 are unassigned — so notification has been going to a fallback nobody watches in real time. Wire a real-time action instead.
Requirements
severity: critical-equivalent issues there. Sentry has its own priority/severity concept independent of Alertmanager's — map it deliberately, don't assume they line up 1:1.Deliverables
Part of #5007.