Erasure Coding Status

See the cluster's erasure-coding parameters and shard-repair backlog.
EndpointMethodAdmin Required
/api/v1/admin/cluster/erasure-coding/statusGETYes

Returns the deployment's erasure-coding parameters (k/m, total shards, write quorum) alongside the shard-repair backlog: how many shard tasks are pending or dead-lettered, broken down by reason. The EC counterpart of rebalance status.

enabled is false on a cluster running plain replication (no CLUSTER_EC_* configuration), in which case every other field is 0. For the full detail behind a count - which shard of which object is stuck, and why - see erasure coding tasks.

Response

{
    "enabled": true,
    "dataShards": 4,
    "parityShards": 2,
    "totalShards": 6,
    "quorumThreshold": 5,
    "pendingCount": 2,
    "failedCount": 0,
    "pendingReconstructCount": 1,
    "pendingByReason": {
        "reconstruct": 1,
        "rebalance": 1
    }
}

Response Fields

enabled
boolean
Whether this cluster erasure-codes object data. false for a plain-replication cluster, in which case all remaining fields are 0.
dataShards
integer
The number of data shards, k (CLUSTER_EC_DATA_SHARDS).
parityShards
integer
The number of parity shards, m (CLUSTER_EC_PARITY_SHARDS) - how many simultaneous node losses an object survives.
totalShards
integer
dataShards + parityShards - the number of nodes each object's shards are spread across, and the minimum active-node count required to accept writes.
quorumThreshold
integer
The number of shards that must be durably acknowledged before a write returns success.
pendingCount
integer
Number of shard-repair tasks currently pending (queued or retrying).
failedCount
integer
Number of shard-repair tasks that exhausted their retry budget (8 attempts) and were dead-lettered. Retained for 7 days before automatic purge.
pendingReconstructCount
integer
Of the pending tasks, how many are reconstruct - rebuilding a permanently-lost shard from survivors. Broken out because it signals a genuine durability event, distinct from an ordinary shard move or catch-up.
pendingByReason
object
Pending shard-task count grouped by why it exists: write (a quorum write's straggler catch-up), rebalance (moving a shard to a node that gained its rank), reclaim (removing a stale shard from a node that lost its rank), or reconstruct (rebuilding a lost shard from survivors).