redis/jedis · critical · JedisClusterOperationException

No cluster nodes available.

Error message

No cluster nodes available.

What it means

RoundRobinConnectionResolver selects the pool of node connections used for cluster-wide (broadcast or read-from-replica) commands. It throws JedisClusterOperationException("No cluster nodes available.") when the map of primary (or replica, for reads) node connections is empty. It means the client cannot find any eligible cluster node to route to.

Solutions

  1. If using read-from-replica, switch to reading from primaries (or configure at least one replica per master).
  2. Verify cluster topology with `redis-cli cluster nodes` — ensure masters exist and slots are assigned.
  3. Check that the startup node addresses in the client builder point to live cluster nodes and topology refresh is working.
  4. Wait for/repair failed nodes so the connection map repopulates.

Example fix

// before
client.setReadFrom(ReadFrom.REPLICA); // cluster has no replicas
// after
client.setReadFrom(ReadFrom.PRIMARY);
// or add a replica: redis-cli --cluster add-node <replica>:6379 <master>:6379 --cluster-slave
Defensive patterns

Strategy: validation

Validate before calling

Map<String, ConnectionPool> nodes = provider.getPrimaryNodesConnectionMap();
if (nodes == null || nodes.isEmpty()) {
  throw new IllegalStateException("no cluster nodes configured; check startup nodes and topology");
}

Try / catch

try {
  return broadcastCommand();
} catch (JedisClusterOperationException e) {
  if (e.getMessage().contains("No cluster nodes available")) {
    refreshTopology();
    return broadcastCommand();
  }
  throw e;
}

Prevention

When it happens

Trigger: A keyless/broadcast command requiring primary nodes while the cluster reports no primaries; a read command routed to replicas when the connection map for replicas is empty (no replicas configured or all replica nodes unreachable/unregistered).

Common situations: Cluster with zero replicas while reading from replicas (misconfigured fromReplica/readFrom settings); node list empty because the topology hasn't been refreshed or all nodes are down; wrong startup nodes pointing at a dead cluster.

Understand the failure class

Background: 'Could not be found', 'does not exist', 'not found in database': the resource-not-found family when an ID, slug, key, or URI lookup comes back empty — this error's family across 20 libraries.

Related errors

AI-assisted analysis of redis/jedis@6dac31d4c2 (2026-09-08). Data as JSON: /api/errors/28387bde456a96e5. Report an issue: GitHub.

Appendix: source

Thrown at src/main/java/redis/clients/jedis/executors/RoundRobinConnectionResolver.java:63

    ConnectionPool pool = selectedEntry.getValue();

    return pool.getResource();
  }

  private List<Map.Entry<String, ConnectionPool>> selectConnectionPool(
      ConnectionResolver.ConnectionIntent intent) {
    Map<String, ConnectionPool> connectionMap;

    if (intent == ConnectionResolver.ConnectionIntent.READ) {
      // For keyless READ commands, use all nodes for load distribution
      connectionMap = provider.getConnectionMap();
    } else {
      // Write operations always go to primary nodes
      connectionMap = provider.getPrimaryNodesConnectionMap();
    }

    if (connectionMap.isEmpty()) {
      throw new JedisClusterOperationException("No cluster nodes available.");
    }

    return new ArrayList<>(connectionMap.entrySet());
  }
}

View on GitHub (pinned to 6dac31d4c2)