Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

SQL Server error 19471: 'The WSFC cluster could not bring the Network Name resource online' when creating an availability group listener

Technical article
Troubleshooting

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

Error 19471 during listener creation means the Windows failover cluster could not bring the listener's Network Name resource online. Microsoft's troubleshooting page attributes failures with this message to the cluster name object (CNO) lacking permission in Active Directory to create and read the listener's computer object. A DNS name clash or a stopped cluster service are the other causes Microsoft names, so read the cluster log before changing Active Directory.

What the error means

When you create a listener, the cluster creates a Network Name resource and a virtual computer object (VCO) in Active Directory, and registers the name in DNS. The CNO, which is the cluster's own computer object, does this work. If it cannot, the Network Name resource does not come online and SQL Server returns the error. Microsoft's page shows two messages:

Msg 19471, Level 16, State 0, Line 2: "The WSFC cluster could not bring the Network Name resource with DNS name '' online. The DNS name may have been taken or have a conflict with existing name services, or the WSFC cluster service may not be running or may be inaccessible. Use a different DNS name to resolve name conflicts, or check the WSFC cluster log for more information."

Msg 19476, Level 16, State 4, Line 2: "The attempt to create the network name and IP address for the listener failed. The WSFC service may not be running or may be inaccessible in its current state, or the values provided for the network name and IP address may be incorrect. Check the state of the WSFC cluster and validate the network name and IP address with the network administrator."

Microsoft states: "The majority of time, listener creation failure resulting in the previous messages are due to a lack of permissions for the Cluster Name Object (CNO) in Active Directory to create and read the listener computer object."

Microsoft's listener page adds a limit. A CNO does not by default have the Create Computer objects permission explicitly and can create 10 VCOs. After 10, creation of more fails. VCOs for availability groups that were deleted are not removed from Active Directory automatically and still count against the 10 until someone deletes them.

Likely causes

Cause What you see How to confirm
The CNO cannot create computer objects in its OU or container 19471 or 19476 on a first attempt in a locked-down OU The CNO's rights on the OU in Active Directory Users and Computers
The 10-VCO default is used up Earlier listeners worked; this one fails Count computer objects created by the CNO, including leftovers from deleted groups
A prestaged VCO exists but the CNO lacks rights on it Failure after someone created the account by hand The VCO's Security tab shows no Full control for the CNO
The DNS name is taken, or clashes with another cluster's name Name conflict in the cluster log Search AD and DNS for the name; NetBIOS keeps only the first 15 characters
A computer object exists and Kerberos is not enabled on the Network Name Resource will not come online Microsoft's page on this case
The cluster service is stopped or unreachable Messages 19471 and 19476 together Cluster state and the cluster log

Prove the cause before changing anything

Everything here only reads. Run the cluster commands on a cluster node.

List the cluster's resources and look at the state of the Network Name resource, using the command on Microsoft's Get-ClusterResource page:

Get-ClusterResource

Then collect the cluster log. Microsoft's Get-ClusterLog page documents -TimeSpan (minutes), -UseLocalTime and -Destination, where "." copies the logs for each node into the current folder:

Get-ClusterLog -TimeSpan 15 -UseLocalTime -Destination .

-TimeSpan covers only the last 15 minutes here, so run it straight after a failed attempt, or use a larger number of minutes.

Collect the Windows System event log as well; it can show whether Active Directory or DNS operations fail when the listener resource comes online. Search the cluster log for the listener's name and the words denied or conflict.

In Active Directory Users and Computers, switch on Advanced Features. Look for a computer object with the listener's name, and note whether it is disabled. Open the Security tab of the OU that holds the CNO and see whether the CNO has Create Computer objects on it. Count the VCOs, including stale ones.

For the SQL Server side, Microsoft gives this query for listener state after creation:

SELECT dns_name AS AG_listener_name, port, ip_configuration_string_from_cluster
FROM sys.availability_group_listeners;

A good result at this stage is a Network Name resource that is online and a listener row. A bad result is a Network Name resource in a failed state with an access-denied or name-conflict line in the cluster log.

Fix

Name conflict. Microsoft's text says to use a different DNS name. Names must be unique in the domain and in NetBIOS, contain only alphanumeric characters, dashes and underscores, and be at most 63 characters, or 15 in SSMS. Microsoft warns that two clusters in one Active Directory using names longer than 15 characters with an identical 15-character prefix produce an error that the Virtual Network Name resource could not be brought online.

CNO rights. Microsoft's prestage page grants the CNO permission on the OU so that the cluster can create VCOs itself.

Disruptive step: this is an Active Directory permission change. Impact: the CNO can create computer objects anywhere in that OU, and in some organisations the security policy prohibits it, as Microsoft notes. Back-out: remove the permission entry you added.

In Active Directory Users and Computers, with Advanced Features on, right-click the OU, choose Properties, then Security, then Advanced, then Add. Select the CNO as the principal, setting Object Types to Computers. Set Type to Allow and Applies to to "This object and all descendant objects", then select Create Computer objects. Wait for replication: Microsoft's cluster page notes that with several domain controllers the change can take a replication cycle, 15 minutes by default.

Prestaged VCO. If policy requires a precreated account, create a computer object with the listener's name in the CNO's OU, then give the CNO Full control on it, as Microsoft's steps show. Microsoft's listener page says it is generally simplest not to prestage the account for a listener, and that prestaging needs membership of Account Operators or help from a domain administrator.

Disruptive step: creating the computer object and editing its permissions changes Active Directory. Impact: a wrongly named or wrongly permissioned object can block the listener again or clash with another name, and the object replicates to every domain controller. Back-out: remove the Full control entry for the CNO, and delete the computer object only if you created it and nothing else uses it. Deleting it cannot be undone except through your own Active Directory recovery process.

Stale VCOs and the 10 limit. Either grant the CNO Create Computer objects explicitly, or delete the leftover VCOs of availability groups that no longer exist. Disruptive step: deleting a computer object removes it from Active Directory. Impact: if it belongs to a live group or role, that name goes offline. Back-out: none documented here beyond your own Active Directory recovery process, so confirm the owner of each object before deleting.

Existing computer object, Kerberos off. Microsoft's options: delete the matching computer object, or select Enable Kerberos Authentication on the Network Name resource.

Disruptive step: deleting the matching computer object removes it from Active Directory, and any live role using that name goes offline. There is no way back except your own Active Directory recovery process, so confirm the object's owner first. Selecting Enable Kerberos Authentication on the Network Name resource changes how that resource authenticates. Back-out: clear the option again and bring the resource online again, and note the setting before you change it.

Cluster service not running. Get the cluster healthy first. Do not force a cluster start to create a listener.

Before trying again, run Get-ClusterResource to see whether the failed attempt left a Network Name or IP address resource behind. Microsoft's pages do not say how to clear such leftovers, so treat that as a limit and involve the cluster owner. Then try again from SQL Server, which Microsoft prefers to creating the first listener directly in the cluster:

Risk: ADD LISTENER creates a Network Name resource, an IP address resource and a computer object in Active Directory. Microsoft's listener page gives no steps for removing a listener, so there is no documented way back from a created listener here. Record the cluster resources and the Active Directory objects first, and involve the cluster owner if the listener must be undone.

ALTER AVAILABILITY GROUP [AG1]
ADD LISTENER 'ag1-listener' (WITH IP (('10.0.20.50', '255.255.255.0')), PORT = 1433);

Confirm the fix

Run Get-ClusterResource and check that the Network Name and IP address resources are online. Run the sys.availability_group_listeners query and see the listener name with its port. Microsoft also gives SELECT * FROM sys.dm_tcp_listener_states; to check the listener is online. Test the connection to the listener name and port with Test-NetConnection from a client, as Microsoft's page does, and check TcpTestSucceeded.

Example (illustrative): in the CONTOSO domain, the CNO for sqlclu01 sits in the default Computers container. Seven VCOs are live and three belong to availability groups deleted last year, so the CNO has created 7 plus 3, which is 10. Creating ag1-listener at 10.0.20.50 in 10.0.20.0/24 fails with 19471 because an eleventh VCO is refused. After confirming the three stale objects belong to no live role, the administrator deletes them, leaving 7, so 3 more can be created, and ADD LISTENER succeeds.

What to watch afterwards

  • Delete VCOs when an availability group is dropped, or the limit comes back.
  • Record the CNO's rights on its OU in the build notes for each cluster.
  • Watch for Group Policy or hardening changes to Active Directory rights or to the Add workstations to a domain right, because Microsoft's cluster service account page ties computer-object creation to both.
  • Test the listener after each patch of the cluster nodes, since the listener's Network Name resource depends on Active Directory and DNS.

Risks and limits

These commands have not been run or lab-tested. The 19471 text comes from a page that lists SQL Server 2012 to 2016 as its original product versions, and later versions can differ. The cluster service account page describes an older account model, so apply its 10-object quota statement with care. Changing Active Directory permissions and deleting computer objects can affect systems other than SQL Server, so involve the directory owner.

Sources