5 Migrating Your ULTRIX System and Network Environment

This chapter describes how to set up a DIGITAL UNIX system for maximum compatibility with ULTRIX systems, and how to migrate file systems from an ULTRIX system to a DIGITAL UNIX system. This chapter also discusses the following topics:

Using the tar and pxtar commands

Configuring Small Computer System Interconnect (SCSI) devices

Setting up internationalization databases

Configuring the inetd daemon for ULTRIX compatibility

Configuring the mountd daemon for ULTRIX compatibility

Note

For information on migrating shadowed data from an ULTRIX system to a DIGITAL UNIX Version 3.0 or later system, see the Logical Storage Manager manual.

5.1 Mounting an ULTRIX File System on a DIGITAL UNIX System

You can mount an ULTRIX File System (UFS) on a DIGITAL UNIX system, provided the file system is created with an 8 kB block size and there are partition tables on the disk. The DIGITAL UNIX system can read the partition table created by the ULTRIX chpt command. Once you mount the ULTRIX file system, you can use it as you normally would. Using an ULTRIX file system on a DIGITAL UNIX system does not affect its usability on an ULTRIX system.

To move an ULTRIX file system to a DIGITAL UNIX system, follow these steps:

If the file system was created with a 4 kB block size, you must dump the disk to tape or to a disk that has a file system created with an 8 kB block size.

Install the disk containing the ULTRIX file system onto the DIGITAL UNIX system.

Check the ULTRIX file system by using the fsck command:
```
# /usr/sbin/fsck /dev/rrz0h
** /dev/rz0h
** Last Mounted On
IMPOSSIBLE INTERLEAVE = 0 IN SUPERBLOCK
SET TO DEFAULT ?
```
The IMPOSSIBLE INTERLEAVE message indicates that the DIGITAL UNIX system cannot use certain information on the ULTRIX disk. Answer the SET TO DEFAULT prompt by typing yes, as shown:
```
SET TO DEFAULT ? yes
IMPOSSIBLE NPSECT = 0 IN SUPERBLOCK
SET TO DEFAULT ?
```
The IMPOSSIBLE NPSECT message indicates that the DIGITAL UNIX system cannot use certain information on the ULTRIX disk. Answer the SET TO DEFAULT prompt by typing yes, as shown:
```
SET TO DEFAULT ? yes
** Phase 1 -- Check Blocks and Sizes
** Phase 2 -- Check Pathnames

.
.
.
#
```
The fsck command continues.

Note

You receive these messages from the DIGITAL UNIX fsck command the first time you use the command on an ULTRIX disk. If you use the fsck command to check the disk later, these messages do not appear.

Create a directory on which to mount the ULTRIX data. The following command creates a directory named ultrixdata:
```
# mkdir /ultrixdata
```

Mount the file system:
```
# mount /dev/rz0h /ultrixdata
```

Each time you move an ULTRIX disk from an ULTRIX system to a DIGITAL UNIX system or from a DIGITAL UNIX system to an ULTRIX system, run the fsck command. Then, mount the disk. For mounting UFS CD-ROM discs, use the -d option to the mount command. See mount(8) for more information.

5.2 Migrating Shadowed Data

This section describes migration from the ULTRIX Disk Shadowing product to the DIGITAL UNIX Logical Volume Manager (LVM) software.

Note

This section does not discuss migration to the Logical Storage Manager (LSM) software on DIGITAL UNIX systems. For migration information about LSM, see the Logical Storage Manager manual.

Before migrating ULTRIX shadowed data to a DIGITAL UNIX system, review the following guidelines:

The LVM subsystem has a broader management scope than the ULTRIX Disk Shadowing product. Nevertheless, the migration strategy presented in this section only focuses on the disk mirroring aspects of the LVM subsystem. For a complete description of the LVM subsystem, see the System Administration manual.

You must have root privilege on the DIGITAL UNIX system to mirror data using the LVM subsystem.

Creating physical volumes, which is an LVM concept, on a raw partition overwrites the existing data on that partition.

An ULTRIX shadow device can only consist of corresponding partitions on physical disks of the same type. Logical volumes do not have this restriction.

You cannot migrate shadowed root and swap partitions to the LVM subsystem.

Consider the user data size and the metadata size when allocating partitions for LVM physical volumes:
- User data
  If the existing ULTRIX shadowed partition is nearly full, migrate the data to a larger partition.
- Metadata
  A DIGITAL UNIX system requires more physical space to replicate data than an ULTRIX system requires because the LVM metadata uses more disk space. Use the default LVM parameters for maximum logical volumes, maximum physical volumes, and maximum physical extents in a volume group, which requires approximately 4 MB of additional disk space.

5.2.1 Migration Summary

The following steps summarize the procedure for migrating shadowed data from an ULTRIX system to a DIGITAL UNIX system:

Dump the ULTRIX shadowed file system to tape. (This is the only step performed on an ULTRIX system.)

Label the disks that you intend to use for disk mirroring. If you plan to migrate the shadowed disks, install the disks on the DIGITAL UNIX system before labeling.

Create and extend a nonmirrored logical volume.

Mirror the logical volume.

Create a DIGITAL UNIX file system on the mirrored logical device.

Mount the DIGITAL UNIX file system and restore the ULTRIX file system from tape.

Repeat this procedure for each ULTRIX shadowed file system.

5.2.2 Migration Example

The following example demonstrates how to migrate an ULTRIX shadowed file system to corresponding partitions on a DIGITAL UNIX system. The resulting migration automatically mirrors data on the DIGITAL UNIX system in the same manner that data was shadowed on the ULTRIX system. The elements of this example include:

ULTRIX Disk Shadowing example elements before migration:: File system: /fsShadow device: /dev/shd14gDisk partitions: /dev/rz1g and /dev/rz2gShadow set: two-memberDisk type: rz56
LVM example elements after migration:: File system: /fsLogical volume: logvolmirVolume group: /dev/vg01Physical volumes: /dev/rz1g and /dev/rz2gMirror capacity: single mirroredDisk type: rz56

Use the following example as a guide for migrating your ULTRIX shadowed data:

Dump the ULTRIX shadowed file system to tape by entering the following command on your ULTRIX system:
```
# dump 0uf /dev/rmt0h /fs
```
This command copies the entire contents of the /fs file system to the /dev/rmt0h tape. The command also records the date of the dump in the file /etc/dumpdates when the dump is successful.

On the DIGITAL UNIX system, create a label on the disks you will use for mirroring:
```
# disklabel -r -w rz1 rz56
# disklabel -r -w rz2 rz56
```
These commands install the standard label on the designated drive. (For more information about initializing disks, see the System Administration manual.)
You can omit this step if you have already installed a label on your disks.

On the DIGITAL UNIX system, create and extend a nonmirrored logical volume using the following steps:
1. Create the physical volumes you will use for disk mirroring by entering the LVM pvcreate command:
```
# pvcreate /dev/rrz1g
Physical volume /dev/rrz1g has been successfully created.
# pvcreate /dev/rrz2g
Physical volume /dev/rrz2g has been successfully created.
```
  This command initializes your direct access storage device for use as a physical volume in a volume group.
2. Create a volume group directory in the /dev directory:
```
# mkdir /dev/vg01
```
  Volumes that are mirrored must be in the same volume group. This command creates the directory that identifies the volume group vg01 for the LVM subsystem.
3. Create the volume group device file:
```
# mknod /dev/vg01/group c 16 0
```
  This command creates the volume group special device file, which is a direct connection between the volume group and the LVM driver code. The volume group special device file must be a character (c) device; it must use one of three predefined major device numbers, in this case 16; and it must have a minor device number of 0.
4. Create the volume group and populate it with the physical volumes you created with the pvcreate commands:
```
# vgcreate /dev/vg01 /dev/rz1g /dev/rz2g
Creating /etc/lvmtab.
Volume group /dev/vg01 has been successfully created.
```
  This command creates the /dev/vg01 volume group that has the members /dev/rz1g and /dev/rz2g. The /etc/lvmtab file contains information that allows the LVM software to access the physical volumes that compose its volume groups after a system reboot.
5. Create the logical volume:
```
# lvcreate -s y -n logvolmir /dev/vg01
A logical volume with name "logvolmir" will be created.
Logical volume "/dev/vg01/logvolmir" has been successfully
created with minor number 1.
```
  The lvcreate command creates a logical volume name, logvolmir.
6. Extend the logical volume to encompass all the physical extents of one physical volume. In this example, 63 is the total number of physical extents in the physical volume /dev/rz1g. The vgdisplay command lists the number of physical extents available on each volume.
  Specify a logical extent for the logical volume by using the lvextend command:
```
# lvextend -l 63 /dev/vg01/logvolmir /dev/rz1g
Logical volume "/dev/vg01/logvolmir" has been
successfully extended.
```
  The -l option extends the logical volume so that it encompasses 63 physical extents. The first argument to the command, /dev/vg01/logvolmir, names the logical volume. The second argument, /dev/rz1g, specifies that the logical extents are assigned to the physical extents on the /dev/rz1g physical device.

Mirror the logical volume on the /dev/rz2g device:
```
# lvextend -m 1 /dev/vg01/logvolmir /dev/rz2g
The newly allocated mirror is now being synchronized.
This operation will take some time.
Please wait...
Logical volume "/dev/vg01/logvolmir" has been
successfully extended.
```
The -m option specifies that the system maintains one mirror of the data in logical volume /dev/vg01/logvolmir. The /dev/rz2g argument specifies that the system maintain the mirror using physical extents on the /dev/rz2g physical device.

Create a file system on the logvolmir volume by using the newfs command:
```
# newfs /dev/vg01/logvolmir rz56
```

Mount the ULTRIX file system on the LVM mirrored logical device and restore the file system from tape:
```
# mount /dev/vg01/logvolmir /fs
# cd /fs
# restore -r 
```
The mount command mounts the /dev/vg01/logvolmir logical volume on the /fs directory. The cd command changes the current directory to /fs, and the restore -r command restores the ULTRIX data from tape to the current working directory.

The /fs file system is now converted to DIGITAL UNIX LVM disk mirroring.

5.3 Using the tar and pxtar Commands

The ULTRIX system supports two commands for maintaining tape archives: pxtar and tar. The pxtar command is POSIX-compliant; the tar command is not.

The DIGITAL UNIX system has one tape archive command, tar. The DIGITAL UNIX tar command is POSIX-compliant.

If you use the ULTRIX pxtar command to create a tape archive, you can read that tape archive by using the DIGITAL UNIX tar command. In addition, if you use the ULTRIX tar command to create archives that fit on a single volume, you can read those single-volume archives with the DIGITAL UNIX tar command.

However, the ULTRIX tar command allows you to create and read an archive that can span multiple tapes. The ULTRIX tar command writes a file header at the start of each continuation tape. By default, the DIGITAL UNIX tar command does not expect the ULTRIX header information. The header information is treated as data, resulting in an incorrectly extracted file and the DIGITAL UNIX tar command reporting a checksum error. To read an ULTRIX tar archive spanning multiple tapes using the DIGITAL UNIX tar command, use the -U option on the DIGITAL UNIX system. This option allows the DIGITAL UNIX tar command to read tapes and to ignore the header information specific to ULTRIX.

5.4 Configuring Small Computer System Interconnect Devices

During the doconfig portion of the installation, the sizer program determines what hardware (such as disks and tapes) is attached to your system and reports its findings in the system configuration file.

On ULTRIX systems, sizer automatically places 16 Small Computer System Interface (SCSI) device entries (rz0-rz7 for disks and tz0-tz7 for tapes) in the system configuration file. This behavior enables you to attach additional SCSI devices at any time without having to rebuild your kernel.

On DIGITAL UNIX systems, sizer finds only the SCSI devices physically attached to your system at the time of installation and specifies those devices in the system configuration file. For example, if you have an RZ56 as unit 0, a TLZ04 as unit 1, and an RZ24 as unit 2 on your system, sizer places only these three devices in your configuration file, as rz0, tz1, and rz2, respectively. If you later add new devices to your system, you must edit the configuration file to include the new devices and rebuild the kernel.

You can save yourself the need to repeat this process by using the /sys/conf/GENERIC file as a guide to edit the configuration file to add all possible rzn and tzn devices the first time you rebuild the kernel. If you are performing an advanced installation, you can edit the configuration file before the first kernel is built. For information about editing the DIGITAL UNIX configuration file and rebuilding the kernel, see the System Administration manual.

The RZ57 SCSI disk and TZK10 SCSI tape units are not supported on a DIGITAL UNIX system.

5.5 Configuring DIGITAL UNIX Shared Memory

Some applications can require you to configure shared memory. Configuring shared memory on a DIGITAL UNIX system is done in the same way as on an ULTRIX system, by editing the configuration file and rebuilding the kernel. However, the configuration parameters are slightly different, as shown in the following table:

Parameter on ULTRIX	Parameter on DIGITAL UNIX	Remarks
`smmax`	`shmmax`	Defines the maximum number of bytes of virtual memory at which a shared memory segment can be sized. The default value is 4 MB on DIGITAL UNIX systems. This value is expressed in pages on ULTRIX systems, and expressed in bytes on DIGITAL UNIX systems.
`smmin`	`shmmin`	Defines the minimum number of bytes of virtual memory at which a shared memory segment might be sized. The default value is 1 MB on DIGITAL UNIX systems. This value is expressed in pages on ULTRIX systems, and expressed in bytes on DIGITAL UNIX systems.
`smseg`	`shmseg`	Defines the maximum number of shared memory segments per process. The default value is 32 on DIGITAL UNIX systems.

These DIGITAL UNIX defaults are set to values that are common to most layered products. See the System Administration manual for information about modifying the configuration file and rebuilding the kernel.

5.6 Setting Up Internationalization Databases

The DIGITAL UNIX internationalization features allow you to receive messages and give input in your native language, even when you are in single-user mode. For this feature to operate correctly, you must store message catalogs and locale databases for the /sbin commands in the /etc directory. You must also be sure that the LANG environment variable is defined correctly.

To store message catalogs and locale databases for the /sbin commands in the /etc directory, follow these steps:

Translate the message catalogs to the appropriate language, if necessary.
The message catalogs are stored in the /usr/lib/nls/msg/en_US.88591 directory. Other message catalogs might also be available in subdirectories of the /usr/lib/nls/msg directory if someone has, for example, translated the system catalogs.

Create subdirectories in the /etc/nls directory.
Programs search for the message catalogs in the /etc/nls/msg/%L directory, where %L represents the currently defined locale. You must create the msg/%L subdirectories. For example, suppose you want to use message catalogs for French as it is spoken in Canada. Enter the following commands to create subdirectories:
```
% cd /etc/nls
% mkdir -p msg/fr_CA.88591
```

Copy to the /etc directory the message catalogs and locale databases for the language and commands you want to use.
For example, suppose you want to use French as it is spoken in Canada when you are in single-user mode. Suppose that someone has translated the system-supplied message catalogs and has stored them in the /usr/lib/nls/msg/fr_CA.88591 directory. In this case, you would enter the following cp commands:
```
% cp /usr/lib/nls/loc/fr_CA.88591 /etc/nls/loc/fr_CA.88591
% cp /usr/lib/nls/loc/fr_CA.88591.en \
/etc/nls/loc/fr_CA.88591.en
% cp /usr/lib/nls/msg/fr_CA.8859/* /etc/nls/msg/fr_CA.8859/.
```
The first cp command copies the French-Canadian character database, the second command copies the environment database, and the third command copies the message catalogs. Delete any message catalogs from the /etc/nls/msg/fr_CA.8859 directory that do not correspond to an /sbin command. This frees up space in the root partition.

Announce to the system that you want to use the French-Canadian locale when you are in single-user mode. To do this, define the LANG environment variable as follows:
```
% setenv LANG fr_CA.88591
```
You can also set the LANG variable in root's .profile file or shell resource file.

5.7 Configuring the inetd Daemon for ULTRIX Compatibility

Both DIGITAL UNIX and ULTRIX systems include the /etc/inetd.conf file, which contains information for the inetd daemon. The inetd daemon is the Internet service daemon.

The DIGITAL UNIX inetd.conf file contains a new field. The following list describes the fields in the DIGITAL UNIX inetd.conf file:

ServiceName, which names one of the services in the /etc/services file.

SocketType, which is either a stream value or a datagram value.

ProtocolName, which is one of the protocols in the /etc/protocols file.

Wait/NoWait, which determines whether the inetd daemon waits for a datagram server to release the socket. (Stream sockets are always NoWait.)

UserName, which specifies the user name that the inetd daemon should use to start the server.

ServerPath, which specifies the full pathname of the server the inetd daemon should execute.

ServerArguments, which are the command-line arguments passed to the server.

The new UserName field allows you to specify what user name inetd should assign to a server when it starts. On ULTRIX systems, servers were automatically started with the root user name. For compatibility, specify root in this field for each service. However, if your server does not need root privileges, consider specifying another user name in this field. As long as your server does not need root privileges, you should not notice a difference between the operation of an ULTRIX server and the operation of a DIGITAL UNIX server that is started under a user name other than root.

5.8 Configuring the mountd Daemon for ULTRIX Compatibility

The mountd daemon works with other daemons to provide the NFS service. This daemon checks the access permission of the client and returns a pointer to the file system or directory that is to be mounted by the NFS service.

By default, the mountd daemon on DIGITAL UNIX systems accepts requests only from the superuser of a remote system. By contrast, the ULTRIX daemon accepts mount requests from any user.

You can configure the mountd daemon on a DIGITAL UNIX system to accept requests from users other than the superuser. To do so, start the daemon with the -n option, as shown:

# /usr/sbin/mountd -n

This command starts the daemon so that it operates the same as the ULTRIX mountd daemon.