apache/dolphinscheduler · error · RegistryException

zookeeper get children error

Error message

zookeeper get children error

What it means

ZookeeperRegistry.children() lists the child znodes of a key via client.getChildren().forPath(key), sorts them in reverse order, and wraps any failure in a RegistryException 'zookeeper get children error'. A missing parent key (NoNode) is the most common cause; connection loss and ACL denial also surface here.

Source

Thrown at dolphinscheduler-registry/dolphinscheduler-registry-plugins/dolphinscheduler-registry-zookeeper/src/main/java/org/apache/dolphinscheduler/plugin/registry/zookeeper/ZookeeperRegistry.java:193

        try {
            client.create()
                    .orSetData()
                    .creatingParentsIfNeeded()
                    .withMode(mode)
                    .forPath(key, value.getBytes(StandardCharsets.UTF_8));
        } catch (Exception e) {
            throw new RegistryException("Failed to put registry key: " + key, e);
        }
    }

    @Override
    public List<String> children(String key) {
        try {
            List<String> result = client.getChildren().forPath(key);
            result.sort(Comparator.reverseOrder());
            return result;
        } catch (Exception e) {
            throw new RegistryException("zookeeper get children error", e);
        }
    }

    @Override
    public void delete(String nodePath) {
        try {
            client.delete()
                    .deletingChildrenIfNeeded()
                    .forPath(nodePath);
        } catch (KeeperException.NoNodeException ignored) {
            // Is already deleted or does not exist
        } catch (Exception e) {
            throw new RegistryException("Failed to delete registry key: " + nodePath, e);
        }
    }

    @Override
    public boolean acquireLock(String key) {

View on GitHub (pinned to 02eac45a1b)

Solutions

1. Guard with registry.exists(key) before listing children, and treat a missing parent as an empty result
2. Check the wrapped cause code: NoNode → create the parent or handle as empty list; CONNECTIONLOSS → retry after session recovery
3. Verify both writer and reader use the same registry root/prefix configuration
4. Confirm ACLs permit the client to list the parent znode

Example fix

// before
List<String> kids = registry.children("/tasks/" + groupId);
// after
List<String> kids = registry.exists("/tasks/" + groupId)
        ? registry.children("/tasks/" + groupId)
        : Collections.emptyList();

if (!registry.exists(key)) {
    return Collections.emptyList(); // parent never created or concurrently deleted
}
List<String> children = registry.children(key);

try {
    return registry.children(key);
} catch (RegistryException e) {
    if (e.getCause() instanceof KeeperException.NoNodeException) {
        return Collections.emptyList();
    }
    throw e;
}

When it happens

Trigger: Calling children(key) when the key has no znode (KeeperException.NoNode), when the session is expired/unreachable, when ACLs deny listing children, or when the key is a malformed path.

Common situations: Enumerating tasks/masters under a parent path that was concurrently deleted (failover, cleanup job); querying before the parent was ever created; ZooKeeper unreachable during startup; path prefix mismatch after changing the registry root config.

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

• Failed to subscribe listener for key:
• zookeeper get data error
• zookeeper check key is existed error
• Failed to put registry key:
• Worker registry client start up error

AI-assisted analysis of apache/dolphinscheduler@02eac45a1b (2026-09-06). Data as JSON: /api/errors/30142249e5721b39. Report an issue: GitHub.