Maintenance & actions

Active maintenance shows running or paused jobs reported by the Agent. Upcoming maintenance shows projected completions in the next 7 days; it is not a scheduler or a list of future jobs configured on the appliance.

Projected collisions compares active-job completion estimates with facility operating hours. Recent completed jobs require a reported completion time; an empty history may reflect unavailable telemetry. Nodes without telemetry have unknown maintenance state.

Set the facility timezone and daily windows in Settings → Facility and Operating Hours. Review projected collisions after changes. Estimates depend on the reported remaining work and rate, and can move as foreground I/O changes.

1. Open the affected system and review maintenance state, current telemetry, and the intended pool or drive.
2. An Operator, Facility Admin, or Organization Admin can request available actions. Scrub pause/resume and locator LED actions depend on the platform and advertised host capabilities; not every host exposes every action.
3. Review the confirmation and submit the action once. Pillar queues commands for delivery on a subsequent authenticated heartbeat.
4. Wait for the reported result and refreshed telemetry. Queued or dispatched does not mean the host completed the action. Offline nodes, unsupported actions, errors, or expiry can prevent completion.

Pausing a scrub delays integrity checking and may reduce competing I/O, but does not promise playback recovery. Resilvers and parity operations are distinct from scrubs: follow your appliance vendor’s procedure. LED identification supports physical verification; also confirm serial number, pool membership, and bay before any hardware change.

Use the runbooks for investigation steps. Pillar does not provide an unrestricted remote shell or automatically replace disks.