Erasure Coding Scrub

Trigger an immediate cluster-wide bit-rot scrub of erasure-coded shards.
EndpointMethodAdmin Required
/api/v1/admin/cluster/erasure-coding/scrubPOSTYes

Triggers an immediate bit-rot scrub on every node. Each node re-verifies the per-stripe checksums of the shards it physically holds and rebuilds any it finds corrupt from healthy survivors. The EC counterpart of resync, and cluster-wide for the same reason: a scrub only ever sees a node's own local shards, so verifying the whole cluster requires every node to run its own pass.

Returns immediately with 200 OK - the scrub runs in the background. Watch its progress through erasure coding status's repair counts (a reconstruct reason means a scrub found a corrupt shard and queued its rebuild). Scrubbing also happens automatically on a schedule (CLUSTER_EC_SCRUB_INTERVAL_HOURS, default weekly); this endpoint is for on-demand verification, e.g. after a suspected disk problem.

Response

200 OK with an empty body once the scrub has been broadcast to the cluster.