Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.novacula.io/llms.txt

Use this file to discover all available pages before exploring further.

By default, every node uses the org-wide alert rule settings. Per-node overrides let you raise or lower that policy for individual nodes — typical reasons: silencing a canary that’s allowed to flap, or tightening the disk threshold on a critical node where 85% is too late. Overrides live on NotificationNodeOverride rows keyed by (organizationId, nodeId). There is at most one override row per node.

​
What you can override

Each setting can be overridden independently. Leaving a field unset on the override means “inherit org default.”
FieldOrg-wide default (typical)Override semantics
nodeDownEnabledtruetrue / false to set; unset to inherit
diskUsageEnabledtruetrue / false to set; unset to inherit
diskUsageThresholdPercent851..100 integer; unset to inherit
The effective policy used for evaluation is the merge: org default for unset fields, override value for set fields. The notification settings page exposes each node’s effective values alongside the override (so you can see what’s currently in force at a glance).

​
Where to set them

From Settings → Notifications → Per-node policies. Each node in the org appears as a row showing:
  • Node name, chain, network, executor.
  • Effective nodeDownEnabled, diskUsageEnabled, diskUsageThresholdPercent.
  • An edit affordance to change the override.
Saving an override calls the same saveNotificationSettings mutation as the org-wide form, with a nodeOverrides array. Removing all override fields on a node deletes its override row.

​
Common patterns

  • Silence a canary node. Set nodeDownEnabled = false on the canary. Disk alerts still fire (or override that too).
  • Stricter disk threshold for a high-write chain. On a Sui mainnet node that grows fast, drop diskUsageThresholdPercent from 85 → 75 so you have lead time to expand the volume before it fills.
  • Don’t alert on a stopped-but-managed node. If you intentionally keep a node desiredState = stopped for testing, set nodeDownEnabled = false so it doesn’t open an incident.

​
Permissions

  • Read effective policy and override status — any org member.
  • Edit an override — owner, admin.
See Roles and permissions.

​
Audit trail

Override changes are written to the Events feed and the audit log. The actor and the diff (org default → override value, or override removed) are captured for compliance.