A.1 VLDB

A.1.1 Managing the VLDB Service

Use the vldb command options in Table A-1 to manage the VLDB service and VLDB file for a DFS replica server. For Linux, make sure you are logged in as the root user or equivalent in the terminal console.

Table A-1 DFS Volume Location Database (vldb) Commands

Options

Description

vldb help

Display a list of available VLDB commands.

vldb start service

Resume operation of the VLDB.

vldb stop service

Suspend operation of the VLDB.

vldb exit

Terminate the Volume Location Service on this machine.

vldb status

Display the current status of the Volume Location Service.

vldb context

Display this VLDB server's management context.

For Linux cluster load scripts:

vldb -dir /vldbpath

Start the VLDB service for the Novell Cluster Services™ cluster of the VLDB services for a DFS management context. Replace vldbpath with the path to the VLDB file. The path much exactly match what you entered for the DFS management context.

Add this command to the cluster load script. Make sure the volume that contains the VLDB is mounted before issuing the command.

vldb refresh

Update the VLDB entries from the other VLDB server.

A.1.2 Managing VLDB Entries (Linux)

For VLDB replica sites on OES 2 Linux servers, the additional commands in Table A-2 are available to manually add volume entries to or delete volume entries from the VLDB database for a single volume at a time without having to run the vldb repair command. Using these commands to manually add and delete entries can be faster than running vldb repair, particularly when you have a large tree but only a few entries that need to be modified. If you have a second replica site, the changes you make in one replica site are automatically synchronized to the second VLDB. The second VLDB replica site can be NetWare or Linux.

To issue these commands:

  • The VLDB service must be running on the replica site.

  • You must be logged in as the root user or equivalent in the terminal console on the Linux replica site.

The action results and errors are displayed on the console from which the operation is done, and are written to the /var/log/messages file.

Table A-2 Additional DFS Volume Location Database (vldb) Commands for Linux

Options

Description

vldb list

Display a list of the in-memory VLDB entries for the currently running VLDB. Displays the server name, volume name, and volume GUID for each volume entry.

vldb add svrname volname

Add an entry for the specified volume to the in-memory entries of the VLDB database, then synchronize the change to the VLDB file on the disk.

The specified volume must have a Volume object in the eDirectory tree, and be in the management context. If the volume’s eDirectory Volume object already contains a DFS GUID attribute, this GUID is used for the entry in the VLDB. Otherwise, this command automatically generates a DFS GUID for the volume, then stores the GUID as an attribute of the Volume object and uses it for the entry in the VLDB.

This command requires eDirectory authentication. The command prompts for the valid administrator username and password of the user who has sufficient rights in eDirectory to update the attributes of Volume objects. Enter the username in typeless fully distinguished format (username.ou_context.o_context, such as admin.eur.company). After successful authentication, the operation is performed.

Replace svrname with the fully distinguished server name (such as .server151.example.com.). Replace volname with the name of the volume (such as DATA).

For example, enter

vldb add .svr151.example.com. DATA
vldb delete vol_dfsGUID

Delete an entry for the specified volume from the in-memory entries of the VLDB database, then synchronize the change to the VLDB file on disk. The delete operation only removes the entry from the database. It does modify or delete the DFS GUID attribute for the volume’s Volume object in eDirectory. It does not delete the Volume object from eDirectory.

This command requires eDirectory authentication. The command prompts for the valid administrator username and password of the user who has sufficient rights in eDirectory to update the attributes of Volume objects. Enter the username in typeless fully distinguished format (username.ou_context.o_context, such as admin.eur.company). After successful authentication, the operation is performed.

Replace vol_dfsGUID with the DFS GUID of the volume as it appears in the report results of the vldb list command. For example, enter (all on the same line, of course)

vldb delete 0x6affb60fdc56dc01800174685ff0d412

IMPORTANT:Deleting the volume entry from the VLDB disables any junction resolution for junctions that target this volume.

If you later run a VLDB repair in the DFS management context, the repair discovers all volumes with Volume objects in eDirectory that are in the management context. It is possible for deleted entries to be added back to the VLDB.

A.1.3 Repairing the VLDB

Use the vldb command options in Table A-3 to repair the VLDB service and VLDB file for a DFS replica server. For Linux, make sure you are logged in as the root user or equivalent in the terminal console.

Table A-3 DFS Volume Location Database (vldb) Repair Commands

Options

Description

vldb repair

Walk the eDirectory tree to rebuild the VLDB by adding the DFS GUID for all volumes with a Volume object in the DFS management context.

You must run the command from the server console to allow a username and password to be entered that is valid across multiple containers and management contexts.

Make sure you perform this task as an admin user with sufficient rights to access the necessary objects in the tree. Otherwise, VLDB Repair cannot scan the entire tree within the DFS management context, and the repair affects only those areas where you have sufficient rights. Problems that occur as a result of logging in with a username with insufficient rights (and any other errors such as crashed servers or eDirectory problems) are logged in the repair log. Administrators should review the repair log to look for errors.

Review the repair log to look for VLDB repair errors:

  • Linux: /var/opt/novell/log/dfs/vlrpr.log

  • NetWare: sys:\etc\vlrpr.log

vldb cancel repair

Stop a running repair operation.

A.1.4 Hidden VLDB Command

When you create a DFS management context, on the first time that the VLDB service is loaded, the vldb command will run with the -init flag (vldb -init) in order to create a new, empty VLDB. This is done automatically by the DFS configuration tools, not by the administrator.

The -init option is generally never used by the administrator unless instructed to do so by Novell Support. The -init option must be issued at the command line in the event that the VLDB file becomes corrupt and the VLDB service cannot be restarted. The VLDB service refuses to load if it finds a corrupt database, so there is no opportunity to repair the VLDB unless the -init flag is specified.