UNSOLVED

abuzzi

updated

17 days ago

A

abuzzi

1 Rookie

•

73 Posts

•

8 Points

0

40

October 29th, 2026 10:18

VNX5300 Unified: what populates /nas/site/user_db?

I have two VNX5300 Unified systems.

 

One system has:

 

[nasadmin@emc1-cs0 bin]$ grep sysadmin /nas/site/user_db 
521:1:1:CLARIION_DOMAIN:sysadmin:storageDomain
522:1:1:CLARIION_DOMAIN:sysadmin2:storageDomain
[nasadmin@emc1-cs0 bin]$

 

The other system has:

 

[nasadmin@emc2-cs0 bin]$ grep sysadmin /nas/site/user_db 
509:1:1:CLARIION_DOMAIN:sysadmin3:storageDomain
[nasadmin@emc2-cs0 bin]$

 

How and when are CLARIION_DOMAIN entries created or removed from /nas/site/user_db?

What supported action synchronizes this file with the SP global-user directory?

 

Thank you!

(edited)

  • DELL-Josh Cr

    Community Manager

    •

    9790 Posts

    •

    45217 Points

    0

    0

    Posted October 29th, 2026 12:39

    Hello,

     

    Good question – this file is one of those places on a VNX Unified system that looks like a simple flat‑file database but is actually a cached, auto‑generated mirror of the Storage Processor’s global user directory, so it’s worth explaining how it is populated and what keeps it in sync.

     

    /nas/site/user_db is maintained by the Control Station’s NAS software, not by anything you edit directly. The CLARIION_DOMAIN entries you’re seeing correspond to Unisphere/Navisphere global users that exist on the block side (the SPs), specifically those accounts that have been granted the storageDomain role, which permits them to also operate within the file side admin context on a Unified array. Whenever a global user account with that role is added, removed, or has its role changed on the SP, the Control Station is supposed to pick up that change and reflect it in this local cache so that NAS side tools and CLI utilities don’t have to query the SP every single time they need to validate a storage domain user.

     

    The synchronization itself happens through the Control Station’s periodic domain user refresh process, which is part of the nas_mcd / nas_db synchronization framework. In practice, this refresh is triggered in a few ways: automatically on a scheduled interval as part of normal Control Station housekeeping tasks, whenever the Control Station reboots or the relevant NAS services restart, and also explicitly when certain administrative commands are run that force a resync between the SP's global user list and the Control Station's cached copy. The supported command for forcing this synchronization manually is typically /nas/sbin/nas_config -Rebuild or, in more targeted cases, running nas_server -list combined with a refresh of the security cache, though the most reliable supported method for forcing a full reconciliation is restarting the NAS services on the Control Station using /sbin/service nas restart, since that forces a fresh pull of domain user information from the SP during the NAS stack initialization sequence.

     

    The discrepancy you’re seeing between the two systems, one showing sysadmin and sysadmin2 while the other only shows sysadmin3, almost certainly reflects actual differences in what global users with the storageDomain role currently exist on each array’s SP, combined with whatever history of account creation, renaming, or deletion has occurred on each system over time. If an account is deleted from the SP's global user list but the Control Station was offline, or the sync process did not run cleanly at that moment, you can occasionally end up with slightly stale entries lingering in user_db until the next successful refresh cycle picks up the change and reconciles it.

     

    If you want to confirm the current source of truth, checking the global user list directly on each SP through Unisphere under Users and Groups, filtered to global scope, will show you exactly which accounts currently exist with the storageDomain role on the block side, and you can compare that against what user_db currently shows. If there is a mismatch between what Unisphere reports on the SP and what is cached in user_db, running the NAS service restart on the affected Control Station is the standard supported way to force reconciliation without needing to manually edit the file, which Dell explicitly does not support doing directly since it is treated as a generated cache rather than an editable configuration source.

     

    If after a forced resync the file still does not match what you see on the SP, that would be worth raising with Dell support directly, since it could indicate a deeper issue with the domain user synchronization service on that particular Control Station rather than just a timing related staleness.

  • abuzzi

    1 Rookie

    •

    73 Posts

    •

    8 Points

    0

    0

    Posted October 29th, 2026 18:37

    @DELL-Josh Cr​ 

     

    @DELL-Josh Cr​ 

     

    Very interesting... thank you for the explanation.

     

    We checked the available options for `nas_config` as root.

    The usage output does not list `-Rebuild`:

     

    [root@emc2-cs0 bin]# /nas/sbin/nas_config -help
    usage: nas_config -IPalias { -list
                               | -list6
                               | -create [ -name device_name ] <numeric_id>
                               | -delete [ -name device_name ] <numeric_id>
                               | -create6 [ -name device_name ] <numeric_id>
                               | -delete6 [ -name device_name ] <numeric_id> }
                  | -ssl
                  | -sessiontimeout [ <number_in_minutes> | off ]
                  | -password { [-min <6..15>] [-retries <max_allowed>]
                                [-newchars <min_num>] [-digits <min_num>]
                                [-specchars <min_num>] [-lcase <min_num>]
                                [-ucase <min_num>] }
                  | -password -default
    [root@emc2-cs0 bin]#

     

    We also could not find a way to create an account with a `storageDomain` role.

    The available account role was Administrator.

     

    We rebooted the Control Station, but the result is unchanged:

     

    [nasadmin@emc2-cs0 bin]$ /nas/sbin/naviseccli -h spa -user sysadmin3 -password **** -scope 0 security -list -type
    Username: sysadmin2
    Role:      administrator
    Scope:     global
    Type:      user

    Username: sysadmin3
    Role:      administrator
    Scope:     global
    Type:      system

    [nasadmin@emc2-cs0 bin]$ grep sysadmin /etc/passwd
    sysadmin3:x:509:201::/home/sysadmin3:/bin/bash

    [nasadmin@emc2-cs0 bin]$ grep sysadmin /nas/site/user_db
    509:1:1:CLARIION_DOMAIN:sysadmin3:storageDomain

     

    Could you clarify what setting or procedure assigns the `storageDomain` role, and whether the reboot should have refreshed `/nas/site/user_db` in our case?

     

    Thank you,
    A.

     

  • DELL-Josh Cr

    Community Manager

    •

    9790 Posts

    •

    45217 Points

    0

    0

    Posted October 29th, 2026 19:04

    Try this:

    It is normal to see Global users like "sysadmin" to appear as "sysadmin5" (number at suffix) on the control station (/etc/passwd). There may be multiple users with status defunct (/sbin/nologin for the shell). To understand this, refer the following definitions and the explanation followed by.

    Local user for File : This users will be present only in Control Station password database. Can manage only VNX File/Control station and no access on Storage Processors (VNX Block) - they will not be made defunct
    Local user for Block : This users will be present only in Storage Processor password database. Can manage only VNX Block (storage process) for the connected VNX Block. They cannot manage VNX File/Control station and even another VNX Block system. This user will not exist in the CS, so no defunct is possible for this users.
    Global Users: On the VNX Unified system, global user can manage both Block and File. On a VNX Unisphere Domain with multiple members, they can manage all VNX Systems (with the permission allowed) - this user may be made defunct
    LDAP Users: LDAP users are same as Global Users. Only difference is, password will be authenticated against the LDAP server they are defined (Microsoft Active Directory is a popular LDAP Service) - this user also maybe made defunct
    Defunct Users: This user account is an account formerly mapped to a storage domain user account. On leaving the domain, the account is marked defunct and can no longer be used to login. Defunct accounts cannot be reactivated. Defunct accounts are maintained for auditing purposes. The administrator may delete this account if they choose.

    During initial configuration, Control station will create the global and ldap users in the NAS Password DB (/nas/site/user_db). If any name conflict occurs, like, global user sysadmin for example, is also defined in the control station, it will create new user with a number at suffix, like, sysadmin1 in /etc/passwd and map to the NAS Password DB.

    Also, at every 30 minutes, the control station will re-check the Unisphere Domain for any changes to various domain data including updates made for Global Users and Ldap Users. The domain can also push the updates at any time when a change was made. If any changes made to the users, control station will mark existing global users as defunct with login shell set to nologin, then it will create new entries for the global and ldap users. Again it will increment the number suffix, like sysadmin may now be sysadmin2.

    It is important to monitor this as if any issue with domain, it may incorrectly keep creating new users and over time, the defunct users may be over hundreds which can affect the Unisphere operation because of such large user entries.

    Also note that, a CS failover and failback can also mark existing global users defunct and new global users will be created. Unfuse/Refuse of the domain also can mark existing users as defunct.

    To cleanup defunct users:

    via GUI :

    Unisphere -> Select the vnx array -> Settings -> Security -> User Management -> Local Users for File

    - Filter with "Show Users with State: Defunct"
    - Select all the users (If more users, select the first user, hold shift key and select the last user)
    - Delete

    via CLI :

    Also refer task 14

    1) Login as nasadmin to the Control station via putty (or other ssh terminal)
    2) "su" to become root
    3) Issue the following command as in the example to identify the defunct users:
    Example :
    # grep defunct /nas/site/user_db
    501:0:0:CLARIION_DOMAIN:sysadmin:defunctStorageDomain
    4) grep ":501:" /etc/passwd
    sysadmin1:x:501:201::/home/sysadmin1:/sbin/nologin
    5) Delete the user as in the following example (501 is the uid noted ):
    Example:
    # /nas/http/webui/bin/admin_management.pl -delUser -uid 501 
    6) Verify the home directory is removed, if not, delete the home directly manually
    # ls -al /home/sysadmin1
    # /bin/rm -r /home/sysadmin1

    If too many users, following quick one line can help, but verify the details first :
    # awk -F: '/defunct/ {print "/nas/http/webui/bin/admin_management.pl -delUser -uid "$1}' /nas/site/user_db
    Once verified, copy/paste to run the actual syntax, and delete the defunct users home directly from /home

    Note: Similar condition may occur with groups too. Very similar procedure can be done with defunct groups.

  • abuzzi

    1 Rookie

    •

    73 Posts

    •

    8 Points

    2

    0

    Posted October 29th, 2026 21:19

    @DELL-Josh Cr​ 

     


    Thanks for the explanation.

     

    We checked both systems with `grep -n 'defunct' /nas/site/user_db`; neither has defunct entries. There are some leftover `/home` directories, but they are not marked as defunct in `user_db`.

     

    To clarify, we are not concerned about the Control Station’s internal username having a numeric suffix such as `sysadmin9`. We want the global username used for login to be `sysadmin`:
    - emc1 uses `sysadmin`,
    - while emc2 currently uses `sysadmin3`.

     

    Ideally, we would move emc2 to `sysadmin` too, even if the Control Station maps it internally to a suffixed account such as `sysadminXX`.

     

    After the account changes and a Control Station reboot, emc2’s `/nas/site/user_db` still contains only `sysadmin3`. The SPs list `sysadmin2` as a global Administrator of Type user and `sysadmin3` as a global Administrator of Type system.

     

    How can we determine whether the Control Station is polling the SP or receiving pushed domain updates, but failing to reconcile them, versus not attempting the synchronization at all?

     

    Is there a specific log, trace, or diagnostic command that records those synchronization attempts and any errors?

     

    Thank you,
    A.

  • DELL-Josh Cr

    Community Manager

    •

    9790 Posts

    •

    45217 Points

    2

    0

    Posted October 30th, 2026 12:45

    You will have to call phone suppport, they have access to more tools.