Files
james.maesandClaude Fable 5.2 f0c6c77354
maven / build (push) Successful in 2m52s
build / maven (push) Successful in 2m52s
docs: upgrade notes for the statement reduction
Co-Authored-By: Claude Fable 5.2 <noreply@anthropic.com>
2026-11-10 15:39:57 -06:00

2.6 KiB

Upgrading

Unreleased: fewer statements from the scheduled jobs

  1. Nothing to add for a host that includes alerts.changelog.sql. Changeset 55 (index ix_alerts_alert_flapping) runs with the host's next migration. A host that copied the DDL instead adds CREATE INDEX IF NOT EXISTS ix_alerts_alert_flapping ON alerts_alert (flapping);.
  2. No configuration. The reference cache (types, groups, escalation policies, steps, mutes) uses QQQ core's Memoization and ClusterCaches. A host with more than one node gets changes to those tables on every node at once when it has a ClusterBus transport (ClusterBusMetaData); without one, the other nodes see a change within 60 seconds.
  3. Behavior changes to know about.
    • timedNotificationCheckedAt on an alert is the last reminder or escalation attempt. It used to be rewritten (with the alert's modify date) on every open alert every minute, whether or not anything was due.
    • Updates of a type's openCount and lastRaisedAt are no longer record-audited.
    • Old notifications are purged hourly, not on every delivery tick.
    • Writes to alerts_type, alerts_group, alerts_escalation_policy, alerts_escalation_step and alerts_mute that do not go through QQQ (direct SQL) are seen by the engine within 60 seconds, not at once.

Unreleased: starter defaults, setup card, table help

  1. Add the defaults changelog after alerts.changelog.sql in the host's master changelog:

    - include: { file: db/liquibase/qbit-alerts/alerts-defaults.changelog.sql }
    

    It is insert-only and guarded (row exists by name: no change; table missing: skipped), so it is safe on an installed host. It creates the in-app and email channels and the alerts and other groups only, the same rows the runtime creates on first use.

  2. Behavior changes to know about.

    • A group's Owner Team is now used: an alert at WARNING or above that no subscription covers (by scope and severity) is sent to that team's members; follow-ups (resolved, reminders) go only to members who were told about the raise. Hosts that never set an owner team see no change.
    • Default Channels on My Alert Settings is now used for subscriptions that name no channels. The bell stays whenever In-App is not turned off; the list replaces the built-in email choice with the channels you chose (unknown ids are ignored). Hosts that never set it see no change.
    • Alert types, groups and default monitors now also sync on the first request of the bell, not only on the home or the first signal.
  3. Nothing else is required. The help bundle and the Setup card need no configuration.