Setting the iSCSI qualified name for the Linux host

To correctly specify the iSCSI qualified name (IQN) for the Linux® host, the initiator name that you choose must match the iSCSI name that is given to the host object for this system.

The iSCSI initiator IQN that you configure on the host side should never be the same as the target IQN on the system node.

Edit the /etc/iscsi/initiatorname.iscsi file to specify the IQN name. The following example shows a sample name:
```
InitiatorName= iqn.1986-03.com.ibm:iscsihost.0
```
After you change the initiator name, you must start the iSCSI-initiator service again.

Configuring more than one Linux host Ethernet port for iSCSI

If you want to configure more than one Linux host Ethernet port for iSCSI, methods are available to improve routing and maintain optimal redundancy in cases of dynamic discovery.

About this task

You can use one of the following two methods:

Configure different subnet IP addresses for the host ports and the system ports. Each target IP of a system node is accessible only from one host IP of the corresponding subnet. As a result, pairs of host and system ports are in separate subnets.
In the /etc/iscsi/ifaces subdirectory, create a separate iface configuration file for each network object that you want to bind a session to. Each iface config file must have a unique name that is less than or equal to 64 characters in length.

Procedure

The following iface example shows a simple case with two network devices (eth0 and eth1):

Edit the /etc/iscsi/ifaces/eth0 file to specify the iface.net_ifacename:
```
iface.net_ifacename = eth0
```
Similarly, edit the /etc/iscsi/ifaces/eth1 file:
```
iface.net_ifacename = eth1
```

iSCSI target discovery for Linux hosts

Two possible methods can be used to discover iSCSI target portals. The following list includes the methods:

Send Targets
Internet storage name service (iSNS)

After discovery, you use the iscsiadm command to log in to or log out of a specific portal on a discovered target.

Discovering iSCSI targets by using send targets

You can discover iSCSI target portals by using the send targets method. The iscsiadm command returns the IQN and target portals that are associated with the system node.

Run the discovery command:

iscsiadm --mode discovery --type sendtargets --portal x.x.x.x

where x.x.x.x is the IP address of a node Ethernet port on the system.

Tip: You can substitute -m for --mode,

-t

for --type, and -p for --portal in the discovery command.

Command output similar to the following example is displayed:

x.x.x.x:3260, 1 iqn.1986-03.com.ibm:2145.china5.hlnc111874

Discovering iSCSI targets on your Linux host by using iSNS

You can discover iSCSI target portals by using the internet storage name service (iSNS) method. Before you use the iSNS discovery, you must do the following tasks:

Configure the system to use iSCSI.
Configure the system to use your iSNS server.
Verify that the system nodes registered their target portals with the iSNS server. If the node iSCSI addresses are visible to the iSNS server, this registration occurs automatically.

To discover iSCSI targets by using iSNS, complete the following steps:

Modify the iSNS server settings in the /etc/iscsi/iscsid.conf file by entering the IP address and the port number that is used by your iSNS server:
```
isns.address = <iSNS server IP address>
isns.port = <iSNS server port>
```
Make the iSNS configuration changes effective by restarting the iSCSI initiator service.
Run the iSNS discovery command to generate a list of all iSCSI target portals that are registered with the iSNS server:
```
iscsiadm --mode discovery --type isns
```

Logging in to discovered targets

You can connect to a specific portal on a discovered target by using the iscsiadm command. The system supports only one iSCSI session between an initiator and a target. Ensure that you do not attempt to connect to the same target node more than once.

For SLES:

iscsiadm --mode node --target target_IQN --portal x.x.x.x --login

For example:

iscsiadm --mode node --target iqn.1986-03.com.ibm:2145.cluster10g1.node1 --portal
192.168.64.191 --login

For RHEL:

iscsiadm --mode node --targetname target_IQN --portal x.x.x.x --login

For example:

iscsiadm --mode node --targetname iqn.1986-03.com.ibm:2145.cluster10g1.node1
--portal 192.168.64.191 --login

When the login command completes, the host can access the volumes that are mapped to the host from the target after the SCSI devices are rescanned.

Logging out from targets

You can log out of a specific system target or you can log out of all established sessions.

Note: After you log out of a system target, the iSCSI volumes that it serves are no longer available to the host.

To log out of a specific system target, enter the following command:
```
iscsiadm --mode node --target <IQN> --portal x.x.x.x --logout
```
To log out of all your established sessions, enter the following command:
```
iscsiadm --mode node --logoutall=all
```

iSCSI timeout configuration settings for the system

You can set the timeout value for disks. The disk timeout is set in the /sys/block/<device>/device/timeout file. Depending on operating system and OS version, increase the default value of 30 or 60 seconds to 120 in most situations.

For example, if the device is /dev/sdc, you would use:

[root@R64SRV6 ~]#echo -n 60 > /sys/block/sdc/device/timeout

To increase the SCSI command timeout for the system, create the following udev rule so that it is not necessary to set the timeout manually for each mapped disk:

udev rules file
cat /etc/udev/rules.d/99-ibm-2145.rules

# Set SCSI command timeout to 120s (default == 30 or 60) for IBM 2145 devices
SUBSYSTEM=="block", ACTION=="add", ENV{ID_VENDOR}=="IBM",
ENV{ID_MODEL}=="2145", RUN+="/bin/sh -c 'echo 120 > /sys/block/%k/device/timeout'"

Reconfirm the settings after your restart the system. After you set up your volumes, use multipath -ll | grep sd to locate the block device paths. Also, check with cat /sys/block/sd<x>/device/timeout for 120 seconds, where <x> is a 2145 path. For example:

edam:~ # multipath -ll | grep sd | tail -1
  `- 10:0:0:4 sdl 8:176 active ready running
edam:~ # cat /sys/class/block/sdl/device/timeout
120
edam:~ #