Skip to content

Est. 2011ยทMicrosoft Partner ID 7033487ยทDelivery under 3 minยทSupport 7 days a week

Your vault is empty.

Free Fix 15404

Error 15404: Could Not Obtain Information About Windows Group or User

11 min read Updated October 5, 2026 SQL Server

Fix it now

SQL Server asked Windows to translate an account identifier into a name and got nothing back. You normally meet it as an Agent job that fails before running a single step, because Agent cannot resolve the job’s owner. It is not a password failure and not a permissions problem inside SQL Server.

Run against the instance whose Agent jobs are failing

SELECT name, SUSER_SNAME(owner_sid) AS owner
FROM msdb.dbo.sysjobs
ORDER BY owner;

EXEC msdb.dbo.sp_update_job @job_name = N'YourJob', @owner_login_name = N'sa';
  1. Read the result of the first query. Any job whose owner comes back NULL is one whose stored identifier no longer resolves, and that is your list.
  2. Reassign those jobs to an account that will always resolve, normally sa, using the second statement. Do them one at a time so you can see which ones change behaviour.
  3. If many jobs failed at once rather than one, check that the server can still reach a domain controller before touching ownership at all.
  4. Read the job step history rather than the job outcome. The inner message names the account that could not be resolved, which is usually enough to explain how it went missing.

Changing the owner can change what the job is allowed to do. A job that relied on a domain owner’s rights needs a proxy configured for the relevant subsystem instead.

If the failing jobs now run, you are done. If the message is about decryption rather than about a name, the next section separates the two.

Why it happens

Every Agent job stores its owner as a security identifier. Before the job runs, Agent translates that identifier back into an account name so it can decide what the job may do: jobs owned by a sysadmin run under the Agent service account, while jobs owned by anyone else are checked against proxies and permissions. That translation is a Windows call, made by the SQL Server service account against Active Directory.

When it returns nothing you get error 15404, published as “Could not obtain information about Windows NT group/user” with an error code attached, and the job never starts. The usual reasons are that the account was deleted or renamed in the directory, that no domain controller can be reached, that the machine’s secure channel to the domain is broken, or that the job was created by an account local to a different machine.

Two neighbouring codes are the same lookup failing elsewhere. Error 15401 is published as “Windows NT user or group not found. Check the name again”, which is what you get creating a login for a name that is not there. Error 15407 is published as the name not being a valid Windows NT name, with the instruction to give the complete name as domain and user, which is what you get from an email-style name or a bare user name.

The decryption codes are a different fault that surfaces in the same place, and v1 of this article ran them together. Error 15466 is published simply as “An error occurred during decryption”. Error 15121 is not a decryption message at all: its published text is that an error occurred during the execution of a named module and that a call failed with an error code. It is a wrapper that tells you which call failed, and the code inside it is the thing to chase.

The owner account no longer exists or was renamed

You have this one if One job fails while others run, and SUSER_SNAME(owner_sid) returns NULL for it.

  1. Reassign the job with sp_update_job and an owner that will not disappear, normally sa.
  2. Check whether the same account owns other objects: credentials, proxies, maintenance plans.
  3. Adopt a convention that jobs are owned by sa and that a step needing a specific identity uses a proxy.

The instance cannot query the directory

You have this one if Many jobs fail at once, Windows logins behave oddly too, and it started after a network, site or domain controller change.

  1. Confirm a domain controller is reachable from the SQL Server itself, and check DNS on that machine.
  2. Test the machine’s domain trust from PowerShell with Test-ComputerSecureChannel.
  3. Fix the network path or repair the secure channel, then run one of the failed jobs by hand.

The service account cannot resolve domain accounts

You have this one if The instance runs under a local account, and every job owned by a domain user fails while local ones succeed.

  1. Set a domain service account through SQL Server Configuration Manager, so the lookups are made by a principal the directory recognises.
  2. If the server must stay on a local account, own the jobs with sa and keep domain principals out of Agent entirely.
  3. Restart SQL Server and then Agent after any account change.

The name was supplied in a form Windows does not accept

You have this one if Error 15407 or 15401 while creating a login or a proxy, with a name typed as a user principal name or with the domain missing.

  1. Use the down-level form, DOMAIN\username, which is exactly what the 15407 message asks for.
  2. For a group, confirm it exists and is a security group rather than a distribution group.
  3. Recreate the login with the corrected name and map it to the databases the job needs.

A credential or proxy cannot be decrypted

You have this one if Error 15466, or a 15121 whose inner call is about decryption, shortly after the service account was changed outside Configuration Manager.

  1. Set the service account again through SQL Server Configuration Manager. Microsoft documents this as the supported way to change the account, because it also updates the Windows local security store that protects the service master key.
  2. Recreate any credential whose password can no longer be decrypted, then retest the proxy from the job step.
  3. Restore the service master key from a backup if you have one, rather than regenerating it.

Do not reach for ALTER SERVICE MASTER KEY WITH OLD_ACCOUNT and OLD_PASSWORD. Microsoft marks those clauses, and their NEW_ACCOUNT and NEW_PASSWORD counterparts, as obsolete and says to use SQL Server Configuration Manager instead.

Full reference

Telling a lookup failure from a decryption failure

What the history shows Which fault it is
The job fails instantly, naming a domain account Owner lookup: 15404
SUSER_SNAME(owner_sid) returns NULL The owner no longer resolves in the directory
Several jobs and other domain operations fail together The server cannot talk to a domain controller
“An error occurred during decryption” 15466: the key chain, usually after a service account change
A message naming a module and an inner error code 15121: read the inner code, not this one

Owner, run-as, and why they are not the same thing

Ownership decides whether Agent has to check anything. A job owned by a sysadmin runs under the Agent service account with no further checks. A job owned by anyone else is checked, and any step outside Transact-SQL needs a proxy that maps to a credential. That is why moving ownership to sa clears 15404 but can change what a step is able to reach: the job is now running as the Agent service account rather than as the person who wrote it.

  • Own jobs with sa so ownership never depends on a directory lookup.
  • Record who actually owns the work in the job description, since the owner field no longer says.
  • Where a step genuinely needs a specific identity, create a credential and a proxy for the subsystem that step uses, and point the step at the proxy.
  • Review proxies after any service account change, because their stored passwords are protected by the key chain that change touches.

The service master key, and the safe way to touch it

Credentials, proxies, linked server passwords and database master keys sit under the service master key. Changing the SQL Server service account outside Configuration Manager is the usual way to break that chain, and the symptom is a decryption error rather than anything that mentions accounts.

Statement Status
ALTER SERVICE MASTER KEY REGENERATE Supported. Regenerates the key and re-encrypts what it protects
ALTER SERVICE MASTER KEY FORCE REGENERATE Supported but destructive; see the warning below
... WITH OLD_ACCOUNT / OLD_PASSWORD Documented as obsolete. Use SQL Server Configuration Manager instead
... WITH NEW_ACCOUNT / NEW_PASSWORD Documented as obsolete. Use SQL Server Configuration Manager instead

Microsoft’s caution on FORCE REGENERATE is that the service master key is the root of the encryption hierarchy, and if a dependent key cannot be decrypted during a forced regeneration, the data that key secures is lost. Use it only when regeneration has failed and you cannot restore the key from backup, and back up the key and the affected databases first.

Checks worth running before you blame the directory

  • Confirm the failing lookup is happening on the SQL Server, not on your workstation. The call is made by the service account from that machine.
  • Confirm the account really is gone rather than moved: a rename keeps the identifier, so a renamed account still resolves.
  • Check whether the job owner is a local account from a machine that has been rebuilt, which is a common leftover after a migration.
  • Look at whether the same account owns credentials or maintenance plans, because those fail with different messages and are easy to miss.

Every code this article covers

Code What it points at Source
15404 Could not obtain information about the Windows group or user, with a Windows error code attached Microsoft Learn
15121 An error occurred during the execution of a named module; a call failed with the error code shown. It is a wrapper, not a decryption message Microsoft Learn
15466 An error occurred during decryption Microsoft Learn
15401 The Windows user or group was not found; check the name again Microsoft Learn
15407 The name given is not a valid Windows name; supply the complete name in the form DOMAIN\username Microsoft Learn

Confirm the fix worked

  1. The owner query returns no NULL owners.
  2. The previously failing job runs to completion when started by hand, and its step history is clean.
  3. Any proxy the job uses works when its step is run on demand.
  4. The job succeeds unattended on its own schedule at least once.
  5. No credential or proxy reports a decryption error after the change.

Questions people ask about this

Is it safe to own every job with sa?

It is common practice and it removes this whole class of failure. The trade-off is that ownership no longer records who created a job, so put that in the description, and use proxies where a step genuinely needs a particular identity.

Does this cost anything to fix?

No. Owner resolution is a directory and configuration matter. No licence, edition or key affects whether a name can be looked up.

The account exists in Active Directory. Why does the lookup still fail?

Because the lookup is made by the SQL Server service account from that server. If the server cannot reach a domain controller, or its computer account has lost its trust with the domain, an account that plainly exists still cannot be resolved.

Why does the job work when I run the same script by hand?

You are running it as yourself, in a session that already resolved. Agent has to resolve the job’s stored owner first, and that happens before any of your code runs.

Can I repair the service master key by supplying the old account and password?

Microsoft documents the OLD_ACCOUNT and OLD_PASSWORD clauses, and their NEW_ACCOUNT and NEW_PASSWORD counterparts, as obsolete and directs you to SQL Server Configuration Manager instead. Change the service account there and let it handle the key material.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

Free Fix Setup Error 0x851A0019: Could Not Find the Database Engine Startup Handle License Error Error 1412: The Remote Copy Has Not Been Rolled Forward Far Enough License Error Error 948 and 946: The Database Was Created by a Newer SQL Server Version License Error SSIS Error 0xC0047062: Packages Fail as Agent Jobs but Run in Visual Studio
โ† Back to Knowledge Base