When the system is shipped from the factory, the management interface has a preconfigured IP address. When using the Setup Wizard, you connect a laptop to that interface in order to perform the initial setup tasks. For information on the Setup Wizard, see Chapter 2, “Initial System Setup”.

The management interface is always configured as an individual network interface and cannot be part of a bonded interface.

You can modify the management interface by selecting eth0 from the following screen:

For information on the network configuration parameters you can modify, see “Ethernet Network Interfaces”.

---

Caution: If you configure an incorrect IP address for the management interface, you can make Appliance Manager inaccessible.

---

To see the available Ethernet network interfaces and change their parameters, select the following:

To modify an interface, select it. You can change the following fields:

Enabled?		Enables the interface. You cannot disable the management interface.
Speed		Displays the port speed of the Ethernet card, which is usually Autonegotiate .
Duplex		Displays the duplex of the Ethernet connection, which is usually Autonegotiate.
Automatic discovery by DHCP		Specifies that dynamic host configuration protocol (DHCP) will be used to configure the Ethernet interface. (Another system must be the DHCP server.)
Static		Specifies that a particular IP address is required for the network interface. If you select this, you must provide the IP address and subnet mask.
Dedicated		Specifies the local and remote IP address for a dedicated network connection between the storage server and another host, for example a dedicated VLAN network or single point-to-point network cable. A dedicated network interface is an interface, such as eth2, that has been configured to use a point-to-point connection with a single remote host. All network traffic to and from that server will go via the local dedicated network interface and no other traffic will appear on that interface. Dedicated network interfaces can be useful when there may be a large amount of network traffic to a specific host and you wish to prevent interference with other network traffic to other hosts. Note: Dedicated interfaces are an advanced option that may require configuration changes to the network infrastructure and on the remote host. You should only use dedicated interfaces if they are specifically required.

To see the available InfiniBand network interfaces and change their parameters, select the following:

To modify an interface, select it. You can change the following fields:

Enabled		Enables the interface.
Automatic discovery by DHCP		Specifies that dynamic host configuration protocol (DHCP) will be used to configure the Ethernet interface. (Another system must be the DHCP server.)
Static		Specifies that a particular IP address is required for the network interface. If you select this, you must provide the IP address and subnet mask.
Dedicated		Specifies the local and remote IP address for a dedicated network connection between the storage server and another host, for example a dedicated VLAN network or single point-to-point network cable. A dedicated network interface is an interface that has been configured to use a point-to-point connection with a single remote host. All network traffic to and from that server will go via the local dedicated network interface and no other traffic will appear on that interface. Dedicated network interfaces can be useful when there may be a large amount of network traffic to a specific host and you wish to prevent interference with other network traffic to other hosts. Note: Dedicated interfaces are an advanced option that may require configuration changes to the network infrastructure and on the remote host. You should only use dedicated interfaces if they are specifically required.

A bonded interface is a virtual network interface that consists of real interfaces working in tandem. You use bonded interfaces on NAS systems to increase bandwidth to NFS and CIFS clients. (It does not apply to CXFS clients because they are connected via Fibre Channel.)

A virtual interface can provide the aggregated bandwidth of all of the interfaces that you used to create it.

---

Note: Any single client can achieve the bandwidth of only a single interface at a time. A bonded interface increases the aggregate bandwidth for multiple clients.

---

For example, if you have three interfaces each with a bandwidth of 10, the aggregate bandwidth is 30. For an individual client, however, the maximum bandwidth remains 10. When additional clients access the bonded interface, the clients are assigned to the subinterfaces, and up to three clients can use a bandwidth of 10 at the same time. Thus multiple clients accessing the system increase the aggregate bandwidth, improving the performance to a maximum bandwidth of 30.

For example, Figure 3-2 shows a configuration in which all clients connect to a single IP address (192.168.0.3 ). The switch is responsible for sharing the load across 4 bonded interfaces (eth1-eth4). Therefore, 4 times as many clients can communicate with the same server without a loss in overall performance.

Figure 3-2. Bonded Network Interfaces

Output load balancing controls how the server chooses which subinterface to send replies. Input load balancing controls how clients are assigned to subinterfaces, and how and when clients are moved from one subinterface to another. Load balancing happens on a per-packet basis. When a client sends a packet, it traverses a switch, which determines at which subinterface the packet arrives. Input load balancing ensures that each client arrives at a different subinterface. The clients see only one interface because the balancing is done by the system.

In addition to configuring a bonded interface in Appliance Manager, you must configure the ports on the switch so that they use either static trunking or 802.3ad dynamic trunking. For more information, refer to the user manual for your switch.

To create a bonded interface, select the following:

The available interfaces are displayed for selection.

When you configure a bonded interface, you specify the following:

Click Apply Changes to create the bond.

This section describes the following:

For background information about how Appliance Manager works, see Appendix A, “How Appliance Manager Configures Filesystems”.

Listing Filesystems

To display a brief description of the RAID to which Appliance Manager is connected, use the List option:

Management -> Resources -> Storage -> Filesystems -> List

This includes the worldwide name (WWN) of the RAID device and an indication of the RAID status, which will be ONLINE unless a hardware or software failure mode has prevented communication between Appliance Manager and the array firmware (such as if the array is powered down or a cable has been pulled out).

Appliance Manager will list filesystems under the following categories, depending on their current state:

Configured Filesystems		Filesystems created by Appliance Manager and filesystems that are able to be managed by Appliance Manager
Unconfigured Filesystems		Filesystems that are able to be managed by Appliance Manager but are not currently fully configured
Unmanaged Filesystems		Filesystems that are not manageable by Appliance Manager, such as manually created filesystems

The Type field on this screen indicates whether the listing is a filesystem, a snapshot repository, iSCSI storage, or available space.

---

Note: Unconfigured filesystems and unmanaged filesystems will show an approximate capacity (indicated by the ~ character) if they are not currently mounted.

---

If you have created a snapshot repository but have not scheduled any snapshots to be taken and stored on that repository, its size will appear as 0 on this display.

To discover unconfigured filesystems, click the Reconfigure Unconfigured Filesystems link on this page. See “Discovering Filesystems”.

Creating Filesystems

---

Note: To create a filesystem, all the storage arrays chosen to contain the filesystem must be supported by Appliance Manager. For best results, SGI recommends that the arrays are symmetrical with respect to the number of drives and trays installed as well as the type of drives installed -- such as Serial Attached SCSI (SAS), Serial ATA (SATA) or Fibre Channel (FC) -- and the speed/size of the drive.

---

The Create option steps you through a filesystem creation wizard. The steps that the wizard will take are listed in a box to the left of the screen, with the current step highlighted.

The filesystem creation procedure is mostly automatic. You provide the name, size, and general characteristics of the filesystem to create and Appliance Manager determines the underlying layout of the filesystem on the disk devices. For information on how Appliance Manager calculates the allocation of disk resources, see Appendix A, “How Appliance Manager Configures Filesystems”.

There is a limit to the number of filesystems on a particular array. This limit is fewer than 30 filesystems for a 4-tray array, but it can be smaller on large arrays (because each filesystem will use 2 or 3 of the total 254 LUN numbers per tray of disks in the array). The number of filesystems and repositories that you can create depends on the make and model of the storage arrays that are connected. Some arrays are capable of supporting up to 254 LUN numbers, but others support only 31 or fewer. The number of LUN numbers consumed by a filesystem/repository depends upon the number of disks and the size of the disks and trays that are connected to the storage array. SGI recommends that you create as few filesystems as possible in order to save LUN numbers (which can later be utilized to grow the filesystem) and because the storage subsystem performs better with fewer filesystems configured.

---

Note: When you create the filesystem, the system detects whether the disk configuration is supported and issues a warning if it is not. You can continue to create the filesystem under these circumstances, but the filesystem will not be an efficient one.

---

You can grow an XFS filesystem after you have created it, by whatever size you choose. It is most efficient, however, if you create a filesystem that fills the disk array and add additional disks if you need to grow the filesystem, filling those disks when you do.

Perform the following steps to create a filesystem:

1. Select the Create option:

   Management -> Resources -> Storage -> Filesystems -> Create

2. Appliance Manager searches for the RAID arrays on the system and displays them on the Arrays screen. If you have more than one storage array, a list of arrays will be presented and you can chose on which arrays the filesystem should be created. Selecting more than one array will result in a filesystem that spans the selected arrays. Spanning filesystems across multiple arrays is possible only for external storage arrays (the SGI InfiniteStorage series). Click Next.

3. The Options screen displays the filesystem configuration options. These are based on the devices that are available to the system and include the following categories:

   Drive type		Specifies the drive type: Serial Attached SCSI (SAS), Serial ATA (SATA) or Fibre Channel (FC). You cannot create a filesystem that spans multiple types of disks.
   Goal		Specifies the goal of the filesystem optimization. You can select a filesystem optimized for performance or capacity (if appropriate for your system). If you select for capacity, Appliance Manager will use all the available disk space to create the filesystem, although this will usually come at the cost of slower performance.
   Workload		Selects the workload type. You can select a filesystem optimized for bandwidth or for I/O per second (IOPS). Select Bandwidth when you will have a small set of files and you must perform streaming reads and streaming writes as fast as possible. Select IOPS when you will be performing random reads and writes to different sets of files. Normally, IOPS will be the better choice. If you are optimizing for IOPS, it is best to build one large filesystem. In general, there is a cost to having multiple filesystems.
   Available Space		Displays the available space in gigabytes ( GiB, 1024 megabytes).

   Click Next.

4. On the Purpose screen, select whether the filesystem will be a clustered CXFS filesystem or an XFS filesystem. The Purpose screen will appear if Appliance Manager is managing a SAN (CXFS) system or if DMF is installed. Depending on the existence of CXFS and DMF, you will be asked if you want to create a clustered CXFS filesystem or a local XFS filesystem, and with or without DMF support. The DMF filesystem option will create the filesystem with 512-byte inodes, and the dmapi and mtpt mount options as required for DMF support. (It will not add the filesystem to the DMF configuration file; you must do this later manually.) For more information about DMF, see “DMF Configuration”.

5. On the Name & Size screen, enter the following:

   • Filesystem mount point (must be begin with /mnt/ as shown).

   • Filesystem size in gigabytes.^[1]The default filesystem size is the size of a filesystem that will completely fill the disk devices. If you choose less than this maximum size, the filesystem will be divided up among the disks. For example, if you create a filesystem that is 20% of the maximum size, it will be spread out among the first 20% of each disk. If you create a second filesystem that is also 20% of that maximum size, it will be spread out among the second 20% of each disk.

     ---

     Note: If you plan to use the XVM snapshot feature, you should ensure that the filesystem capacity entered will leave enough remaining free capacity to create a snapshot repository. For further information, see “XVM Snapshots” in Chapter 1. XVM snapshots are not available on DMF or CXFS filesystems.

     ---

     
     

   • Optional snapshot repository size. The size of the repository that you will need depends on several factors:

     • The size of the filesystem for which you are creating a snapshot. A repository that is approximately 10% of this size is a reasonable starting estimate.

     • The volatility of the data in the volume. The more of the data that changes, the more room you will need in the repository volume.

     • The snapshot frequency. (More frequent snapshots results in smaller individual snapshots.)

   Click Next.

6. The Confirmation screen summarizes the filesystem options you have selected. Click Next to confirm your choices and create the filesystem.

7. The Create filesystem screen displays a "please wait" message and transitional status during the filesystem creation process. Click Next after the operation is finished and the completion message displays.

8. The Create repository screen (if you have chosen to create a snapshot repository) displays a "please wait" message and transitional status during the filesystem creation process. Click Next after the operation is finished and the completion message displays.

9. The NFS and CIFS screen lets you configure the filesystem so that it can be exported with NFS or CIFS network protocols. (If you NFS export and/or CIFS share a CXFS filesystem, it will only be exported/shared from the CXFS metadata server, not from CXFS clients.) ^[2]For information, see “NFS Configuration” and “CIFS Configuration”. Click Next.

10. The Finished screen indicates that the filesystem has been created. Click Done .

Growing Filesystems

---

Note: You cannot use Appliance Manager to grow a CXFS filesystem.

---

You can use a filesystem normally as you grow it. (You do not need to disable access or unmount it, or take any other special actions before growing the filesystem.)

To increase the size of an existing XFS filesystem, do the following:

1. Select the Grow option:

   Management -> Resources -> Storage -> Filesystems -> Grow

2. The Filesystem screen lists the current filesystems along with their usage and size. Select the filesystem you want to grow and click Next.

3. The Size screen lets you enter the size in gigabytes^[3]by which the filesystem should be grown. Click Next.

4. The Confirmation screen displays the current size of the filesystem and the amount to grow the filesystem. Click Next.

5. The Growing screen displays a "please wait" message during the growing process. Click Next after the operation is finished and the completion message displays.

6. The Finished screen indicates that the larger filesystem is available. Select Done.

Internet Small Computer Systems Interface (iSCSI) is a protocol that is used to transport SCSI commands across a TCP/IP network. This allows a system to access storage across a network just as if the system were accessing a local physical disk. To a client accessing the iSCSI storage, the storage appears as a disk drive would appear if the storage were local.

In an iSCSI network, the client accessing the storage is called the initiator and runs iSCSI Initiator software. The remote storage that the client accesses is called the target, which is what appears to the initiator as a disk drive.

A common application of an iSCSI network is to configure an Exchange Server as an iSCSI initiator that uses an iSCSI target as its mail store.

Figure 3-3 illustrates iSCSI storage. Each client (initiator) is configured to connect to a specific iSCSI target (an area allocated in the RAID iSCSI storage pool), and views this target as if it were a local disk. The lines in Figure 3-3 indicate data flow.

Figure 3-3. iSCSI Storage

You can use Appliance Manager to create iSCSI targets on the RAID storage. An iSCSI initiator will be able to connect to the system and access those targets, format them, and use the targets as it would use a disk drive.

You cannot configure Appliance Manager itself as an initiator, and you cannot re-export iSCSI targets with NFS, CIFS, or CXFS. In addition, you cannot export existing filesystems that you have created with Appliance Manager as iSCSI targets; you can create filesystems and configure them to be exported by NFS, CIFS, or CXFS, but you must configure iSCSI targets separately on the RAID device.

---

Note: Due to the nature of iSCSI as a block-level protocol (as distinct from file-level protocols such as NFS and CIFS), particular care must be taken in the event of a system crash, power failure, or extended network outage. See “Power Outage and iSCSI” in Chapter 5.

---

This section discusses the following:

This section discusses the following:

Schedule Snapshots

To schedule how often the system will create a snapshot of a filesystem, do the following:

1. Select the Schedule Snapshots menu:

   Management -> Resources -> Storage -> Snapshots ->  Schedule Snapshots

2. Select the filesystem for which you want to schedule snapshots.

3. Specify the following options:

   Scheduled?
   	Specifies that a snapshot will take place for the filesystem.
   Scheduled Snapshot Times
   	Specifies the hours at which a snapshot should take place. You can select multiple boxes.
   Custom Time Specification
   	Specifies the times and frequency that a snapshot should take place (the minimum interval is 30 minutes). You can specify this value using one of the following forms: Spelled out using the following syntax: every XX minute/s|hour/s|day/s [from XX:XX to XX:XX] The specific times listed within brackets are optional. For example: every 1 hour every 20 minutes from 8:00 to 22:00 every 4 days Exact times. For example: 12:45 23:00 9:30 The word never, which disables snapshots.
   Maximum number of snapshots
   	Specifies the maximum number of snapshots that will be retained in the repository before the oldest snapshot is deleted when a new snapshot is taken. By default, the system will retain 32 snapshots. The maximum number is 256. SGI recommends that you use the default. Caution: If you set the maximum number of snapshots to a number lower than the current number of existing snapshots, old snapshots will be deleted automatically.

   Click Schedule snapshots to apply your settings.

4. Verify that you want to update the snapshot schedule by clicking Yes. (To return to the previous screen, click No.)

---

Note: The system will delete the oldest snapshot if it determines that repository space is running low.

---

Snapshots are made available in the /SNAPSHOTS directory of the base filesystem. They are named according to the date and time at which they are taken. For example, a snapshot might be named as follows:

/mnt/data/SNAPSHOTS/2006_07_30_113557_Sun

Windows clients can access snapshots using the Windows Shadow Copy Client. This feature allows a Windows client to right-click a file or directory, select Properties, and access previous snapshot version of the file. Windows 2000 and Windows XP users should download and install the ShadowCopyClient.msi installer, which is discussed at:

http://support.microsoft.com/kb/832217

Users with Windows 2003, Windows Vista, or later will already have this software installed on their systems.

Appliance Manager supports most common DMF configurations. There are some limitations to this support, however. Specifically, the following are assumed to be true:

The Empty Tape Volume screen uses the herr, hvfy, and hlock DMF database flags to record the progress of the emptying procedure. If you use the dmvoladm(8) command to inspect the database entry for a tape while it is being emptied, you may see unexpected settings of these flags. Appliance Manager's use of these flags does not interfere with DMF's.

Appliance Manager does not make any use of the VOL database flags reserved for site use, although the Import and Export screens do allow you to manipulate them.

The Empty Tape Volume screen's Empty Volume, Remove Volume, and Reuse Volume options cannot remove soft-deleted files from a tape volume, unlike the Merge Volume button. You must wait until they have been hard-deleted by the scheduled run_hard_deletes.sh task or by the dmhdelete(8) command.

Also, these three buttons may need access to the output file from the previous run of the scheduled run_filesystem_scan.sh task or the dmscanfs(8) command. If it cannot be found or is older than the files remaining on the tape, some files may be misreported in the Alerts screen as soft-deleted and remain on the tape as described above. Trying again after the next run of run_filesystem_scan.sh is likely to succeed in this case.

For more information, see the dmemptytape (8) man page for more information.^[5]

You can use the DMF Configuration screens to inspect and modify various DMF parameters.

For initial configuration of DMF, use the Edit link:

This link allows you to directly modify the configuration file or import another configuration file.

---

Caution: You must ensure that the changes you make are safe. For more information, see the dmf.conf(5) man page and the DMF 4 Administrator's Guide for SGI InfiniteStorage.)

---

The Check link allows you to perform syntax and sanity checks on the current configuration of DMF:

SGI recommends that you use the Check link after making any modification to ensure that the changes are safe.

The Global link displays parameters for all of DMF:

If you click Switch to Expert Mode on the Global page, Appliance Manager presents more parameters. You should use expert mode with care. To return to normal mode, click Switch to Normal Mode. Excluded from both modes are parameters that are:

To work around these restrictions, the Edit link allows you to edit the DMF configuration file directly.

The other links provide quick access to commonly altered parameters of already-configured features. You should make changes with care. Parameters that can be dangerous to change are displayed but may not be altered; this includes those parameters that control the search order of volume groups and media-specific processes (MSPs) when recalling files.

---

Note: On the DMF Configuration screens, disk sizes use multipliers that are powers of 1000, such as kB, MB, and GB. This is for consistency with the DMF documentation and log files. However, the rest of Appliance Manager, including the DMF Monitoring screens, use multipliers that are powers of 1024, such as kiB, MiB, and GiB.

---

Appliance Manager can create and add local user and group accounts to access the storage server locally. This is a local database only; these users and groups do not interact with the users and groups provided by the name server. If you search the site directory and do not find the user or group data you are looking for, the system searches this local database. The local user accounts will be used for authentication for CIFS shares if you are not using LDAP or Active Directory authentication.

---

Caution: If you create a local user and subsequently add that user in the sitewide directory, access problems may result. For example, if you create local user Fred with a UID of 26, Fred will be able to create local files. But if you subsequently add a user Fred on a sitewide name services directory with a different UID, user Fred will be unable to access those local files because the system will use the sitewide name and UID first.

---

If you are using LDAP or Active Directory as a name service client, a user must be present in LDAP or Active Directory and you will not be able to authenticate local users and groups. In this case, adding local users and groups may be useful for ID mapping, but authentication does not use the local password files.

When you select the Import option for either Local Users or Local Groups , you can choose among the following actions:

Accounts with a UID or GID of less than 1000 are considered restricted and are not imported or replaced.

If you use a shadow file, which is a file that is protected from all access by non-root users and stores the encrypted passwords, then you can use the Import Users screen to import this file as well as the password file itself.

Appliance Manager will create new filesystems with both user and group quotas enabled by default.

This section discusses the following:

User Quotas

You can use the following screen to specify the user for whom you want to modify quotas:

Management -> Resources -> Users & Groups -> User Quotas

Enter the name of the user and click Submit. (To modify the default for user quotas, leave the field blank.) The following screen displays the current amount of disk space that can be used (disk limits, in KiB) and the number of files that can be owned (file limits):

• The soft limit is the number of 1-KiB blocks or the number of files that the user is expected to remain below. If a user hits the soft limit, a grace period of 7 days will begin. If the user still exceeds the soft limit after the grace period expires, the user will not be able to write to that filesystem until he or she removes files in order to reduce usage.

• The hard limit is the number of 1-KiB blocks or the number of files that the user cannot exceed. If a user's usage reaches the hard limit, he or she will be immediately unable to write any more data.

---

Note: The administrator can set quotas for the root user. However, instead of enforcing these quotas against the root user specifically, they will apply to all users that do not have their own quotas set. In other words, setting quotas for the root user will set the default quotas for all normal users and groups. (The actual root user is exempt from quota limits.)

---

Group Quotas

You can use the following screen to specify the group for which you want to modify quotas:

Management -> Resources -> Users & Groups -> Group Quotas

Enter the name of the group and click Submit. (To modify the default for group quotas, leave the field blank.) The following screen displays the current amount of disk space that can be used (disk limits, in KiB) and the number of files that can be owned (file limits):

• The soft limit is the number of 1-KiB blocks or the number of files that the group is expected to remain below. If any user in that group hits the soft limit, a grace period of 7 days will begin. If the user still exceeds the soft limit after the grace period expires, the user will not be able to write to that filesystem until he or she removes files in order to reduce usage.

• The hard limit is the number of 1-KiB blocks or the number of files that the group cannot exceed. If the usage for a user in that group reaches the hard limit, he or she will be immediately unable to write any more data.

---

Note: The administrator can set quotas for the root group. However, instead of enforcing these quotas against the root group specifically, they will apply to all groups that do not have their own quotas set. In other words, setting quotas for the root group will set the default quotas for all normal groups. (The actual root user is exempt from quota limits.)

---

The Global Options screen lets you specify the following:

Enable NFSv4		Specifies whether NFSv4 is enabled (checked) or not. If enabled, an NFS exported filesystem will be accessible via both NFSv3 and NFSv4. The following fields are only relevant if you have enabled NFSv4.
NFS serving domain		Specifies the serving domain. If NFSv4 is enabled, the mapping of user/group IDs between the client and server requires both to belong to the same NFS serving domain.
Enable Kerberos		Specifies whether Kerberos is enabled (checked) or not. Enabling Kerberos forces encrypted authentication between the NFS client and server. Furthermore, the NFS exported filesystems will only be accessible to a Kerberos enabled client via NFSv4. The following fields are only relevant if you have enabled Kerberos. Note: Appliance Manager supports Kerberos 5. You must use a mechanism to synchronize the time between all systems.
Realm		Specifies the Kerberos realm in which the NFSv4 server operates.
Domain		Specifies the DNS domain name that corresponds to the realm.
KDC		Specifies the key distribution center (KDC). In most cases, the KDC will be the same system as the Kerberos admin server. However, if the admin server in your Kerberos environment is not used for granting tickets, then set the KDC to the system that grants tickets.
Admin Server		Specifies the server containing the master copy of the realm database.
Keep Existing Keytab		Select this radio button to keep the existing keytab without changes.
Update Keytab		Select this radio button to change the principal user and password for the existing keytab.
Principal		Specifies a user that belongs to the Kerberos server with sufficient privileges to generate a keytab for the NFS server.
Password		Specifies the principal's password.
Upload Keytab		Copies the selected file to /etc/krb5.keytab on the NFS server. Click Browse to see a list of available files.
Verify Keytab		Specifies that the keytab should be verified. This is not supported by Active Directory.

You can choose to export or not export a filesystem by clicking the check box. When you enable a filesystem for export, you can do one of the following:

After specifying the configuration parameters, click Apply changes.

Use Export Options

If you select Use export options, you must specify the following:

Read-only		Specifies that the client has access to the filesystem but cannot modify files or create new files.
Asynchronous writes		Specifies whether or not to use asynchronous writes. Data that is written by the client can be buffered on the server before it is written to disk. This allows the client to continue to do other work as the server continues to write the data to the disk. By default, writes are performed synchronously, which ensures that activity on the client is suspended when a write occurs until all outstanding data has been safely stored onto stable storage.
Allow access from unprivileged ports		Allows access for Mac OS X clients or other NFS clients that initiate mounts from port numbers greater than 1024. If there are no such clients on your network, leave this option unchecked.
All hosts		Allows connections from anywhere on a network.
Local subnet		Allows connections from the indicated subnet. You can select any subnet from those that have been defined for the network interfaces.
Kerberos aware clients (krb5)		Allows connections only from those systems that are Kerberos aware (if Kerberos is enabled in “Global Options”) over NFSv4.
Kerberos with Integrity support aware clients (krb5i).		Allows connections only from those systems that are Kerberos with Integrity support aware (if Kerberos is enabled in “Global Options”) over NFSv4
Restrict to hosts		Specifies the set of hosts that are permitted to access the NFS filesystem. You can specify the hosts by hostname or IP address; separate values with a space or tab. For example, you could restrict access to only the hosts on a Class C subnet by specifying something like the following: 150.203.5 To allow hosts of IP address 150.203.5.* and myhost.mynet.edu.au, specify the following: 150.203.5. myhost.mynet.edu.au You can also specify hosts by network/subnet mask pairs and by netgroup names if the system supports netgroups. To allow hosts that match the network/subnet mask of 150.203.15.0/255.255.255.0 , you would specify the following: 50.203.15.0/255.255.255.0 To allow two hosts, hostA and hostB, specify the following: hostA hostB Note: Access still requires suitable user-level passwords. The localhost address 127.0.0.1 will always be allowed.

Use a Custom Definition

If you select Use custom definition, you can enter any NFS export options that are supported in the Linux /etc/exports file.

For example, the following entry gives 192.168.10.1 read-write access, but read-only access to all other IP addresses:

192.168.10.1(rw) *(ro)

---

Note: There cannot be a space between the IP address and the export option.

---

For information on the /etc/exports file, see the exports(5) man page. ^[6]

Use the System Name screen to set the following system components:

System name		Specifies the fully qualified domain name (FQDN) for this storage server. The default hostname is sgiserver. (You cannot change the default hostname for a SAN Server.) Note: After changing the hostname of the NAS Server, the various Appliance Manager screens will still display the old hostname. SGI recommends that you reboot the system to complete the hostname change.
Workgroup		Specifies the NetBIOS workgroup to which the machine should belong. The default is WORKGROUP. If you are not using CIFS, you can ignore this setting.
Default network gateway		Specifies the IP address of the router that this system should use to communicate with machines that are outside of its subnet.
Management IP address		Specifies the IP address of the management interface
Subnet mask		Specifies the subnet mask of the management interface.
Use DHCP		Specifies whether or not to use dynamic host configuration protocol (DHCP).

You can also use the Network Interfaces screen for eth0 to configure or modify the management interface. For information on these options, see “Ethernet Network Interfaces”.

The Name Service Client screen lets you specify a name service (or directory service) for the system. A name service is the application that manages the information associated with the network users. For example, it maps user names with user IDs and group names with group IDs. It allows for centralized administration of these management tasks.

You can specify whether you are using local files (if you have no sitewide protocol and names and IDs are kept locally on server), Active Directory services, lightweight directory access protocol (LDAP), or the sitewide network information service (NIS).

---

Note: When specifying servers on the Name Service Client screen, you must use IP addresses rather than hostnames, because the system may require a name service client to determine the IP address from the hostname.

---

Active Directory

Active Directory is a directory service that implements LDAP in a Windows environment. It provides a hierarchical structure for organizing access to data. CIFS authentication will automatically use the Active Directory service.

---

Note: The Active Directory section is disabled if there are no Active Directory DNS servers specified. See “DNS and Hostnames”.

---

The following Active Directory components appear on the Name Service Client screen:

Active Directory domain		Specifies the full domain name of the Active Directory. Note: If you later change the server hostname on which Appliance Manager runs, you must rejoin the Active Directory domain because the Active Directory Security Identifier (SID) will be changed.
Domain Controller		Specifies a domain controller.
Administrative user		Specifies the user with administrator privileges.
Allow this user to remotely manage CIFS share permissions		Specifies whether or not the Administrative user specified will be able to use the Windows MMC Computer Management GUI to manipulate CIFS share permissions remotely when you join the Active Directory domain.
Password		Specifies the password for the administrator user. For security reasons, the Active Directory password cannot contain the following characters: ;*&|`<>?[]
Re-enter password		Verifies the password for the administrator user.
UID/GID Mapping		Lets you manage UNIX user ID (UID) and group ID (GID) mapping on the Active Directory server, using one of the following: RFC 2307 (Microsoft Windows Server 2003 R2) . In order for this to function correctly: The Active Directory domain controller must be running Microsoft Windows Server 2003 R2. The Identity Management for UNIX service must be installed on the domain controller. You must use the UNIX Attributes tab in Active Directory user management to set up UIDs and GIDs for all users requiring access to this system. Microsoft Windows Services For UNIX. In order for this to function correctly: Microsoft Windows Services for UNIX must be installed on the Active Directory domain controller. You must use the UNIX Attributes tab in Active Directory user management to set up UIDs and GIDs for all users requiring access to this system. Automatic assignment based on Windows SID. In this mode, UIDs and GIDs are automatically based on the Windows SID and are set to be in the range 16777216 through 33554431. Note: This method can only be used within a single Active Directory domain and is incompatible with trusted domains. Automatic assignment in range 10000-20000. In this mode, UIDs and GIDs in the range 10000 through 20000 will be automatically assigned to Active Directory users on a first-come, first-served basis. Note: The default is Automatic assignment based on Windows SID. For best interoperability, SGI recommends that you choose either RFC 2307 (Microsoft Windows Server 2003 R2) or Microsoft Windows Services For UNIX when applicable, as appropriate for your environment.

---

Caution: Depending on your environment, making changes to the UID/GID mapping may result in ownership changes of user files.

---

;*&|`<>?[]

http://directory.fedora.redhat.com/wiki/Howto:Samba

http://www.samba.org/samba/docs/man/Samba-HOWTO-Collection/passdb.html#id327194 

http://www.unav.es/cti/ldap-smb/ldap-smb-3-howto.html

NIS

Network information service (NIS) is a network lookup service that provides a centralized database of information about the network to systems participating in the service. The NIS database is fully replicated on selected systems and can be queried by participating systems on an as-needed basis. Maintenance of the database is performed on a central system.

---

Note: NIS cannot be used for CIFS authentication.

---

Specify the following:

Domain name		Specifies the NIS domain name for this system.
NIS server IP address		Specifies the IP address of the NIS server. If the NIS server is on the same subnet as Appliance Manager, Appliance Manager finds the NIS server IP address and provides it as a default. If you are not on the same subnet, you must enter the address in this field.

Click Apply changes.You will then be presented with a confirmation screen that allows you to verify whether or not you want to commit the changes.

You can use the DNS and Hostnames screen to specify how to map hostnames to IP addresses for the system. Click Edit local hosts table to access the Hosts screen, where you can edit the /etc/hosts file that contains local mappings or import the contents of a file you specify. For information on the /etc/hosts file, see the hosts(5) man page. ^[8]

You can also specify the DNS servers to map hostnames to IP addresses and to resolve hostnames that are incomplete.

---

Note: If you specify one or more servers for DNS, all name resolution will be provided by the specified DNS servers (plus the contents of /etc/hosts). If you do not specify a server, only .local names will be resolvable via multicast DNS (plus the contents of /etc/hosts). You cannot use both DNS to resolve names and multicast DNS to resolve .local domain names. If you specify one or more DNS servers, SGI InfiniteStorage Appliance Manager adds mdns off to the /etc/host.conf file in order to force resolution of .local names to go to the DNS server rather than using multicast DNS. If you later remove the DNS servers, the value of mdns off in /etc/host.conf remains the same. If you manually edit /etc/host.conf to force mdns on, Appliance Manager will not change this setting provided that you do not specify DNS servers via “DNS and Hostnames”.

---

Use the Time and Date screen to set the following:

AEST-10AEDT,M10.5.0,M4.1.0

Appliance Manager is shipped with temporary licenses. The Licenses screen provides information required to request licenses and a text box in which you can type in or paste permanent licenses obtained from SGI. Some licenses, such as the license for XVM snapshot, will not take affect until you reboot the system.

The Administrator Password screen changes the Appliance Manager administration password, which is the password required to perform server configuration and management. This password is not required to view the Appliance Manager monitoring screens.

The Save/Restore Configuration screen screen lets you save the current Appliance Manager configuration or restore a previously saved version. The configuration information saved includes how the interfaces are configured and what filesystems should be mounted. You may find this useful if you have made an error in the present configuration and you wish to return to a previously configured state.

---

Caution: This procedure does not provide a system backup and specifically does not save or restore user data; it provides a snapshot record of the configuration.

---

This screen lists previously saved configurations, labeled by date. After restoring a configuration, you should restart the system.

If there is a problem with the system, SGI Call Center Support may request support data in order to find and resolve the problem. The Gather Support Data screen lets you generate an archive containing copies of the storage server's software and hardware configuration and log files.

To collect the data, select Yes, gather information. This process can take more than 30 seconds on large RAID configurations and requires at least 200 MB of free space in /tmp.

This screen lets you capture and download archives of performance data from the server on which Appliance Manager is running. SGI may request such an archive for performance-analysis purposes, but please be aware that it may contain potentially sensitive information such as network traces.

---

Note: The Performance Data screen in Appliance Manager is only available if you have installed the oprofile and ethereal packages.

---

From the Shutdown screen, you can specify to reboot or shut down the system in a specified number of seconds.