Presentation
In the iSCSI world, you’ve got two types of agents:
- an iSCSI target provides some storage (here called server),
- an iSCSI initiator uses this available storage (here called client).
As you already guessed, we are going to use two virtual machines, respectively called server and client. If necessary, the server and client virtual machines can be one and only one machine.
iSCSI Target Configuration
Most of the target configuration is done interactively through the targetcli command. This command uses a directory tree to access the different objects.
To create an iSCSI target, you need to follow several steps on the server virtual machine.
Install the following packages:
# yum install -y targetcli
Activate the target service at boot:
# systemctl enable target
Note: This is mandatory, otherwise your configuration won’t be read after a reboot!
Execute the targetcli command:
# targetcli Warning: Could not load preferences file /root/.targetcli/prefs.bin. targetcli shell version 2.1.fb34 Copyright 2011-2013 by Datera, Inc and others. For help on commands, type 'help'. />
You’ve got two options:
- You can create a fileio backstore called shareddata of 100MB in the /opt directory (don’t hesitate to use tab completion):
/> backstores/fileio/ create shareddata /opt/shareddata.img 100M Created fileio shareddata with size 104857600
Note: If you don’t specify write_back=false at the end of the previous command, it is assumed write_back=true. The write_back option set to true enables the local file system cache. This improves performance but increases the risk of data loss. In production environments, it is recommended to use write_back=false.
- You can create a block backstore that usually provides the best performance. You can use a block device like /dev/sdb or a logical volume previously created (# lvcreate –name lv_iscsi –size 100M vg):
/> backstores/block/ create block1 /dev/vg/lv_iscsi Created block storage object block1 using /dev/vg/lv_iscsi.
Then, create an IQN (Iscsi Qualified Name) called iqn.2014-08.com.example with a target named t1 and get an associated TPG (Target Portal Group):
/> iscsi/ create iqn.2014-08.com.example:t1 Created target iqn.2014-08.com.example:t1. Created TPG 1. Global pref auto_add_default_portal=true Created default portal listening on all IPs (0.0.0.0), port 3260.
Note: The IQN follows the convention of the RFC 3270 (see http://en.wikipedia.org/wiki/ISCSI to get more details).
Now, we can go to the newly created directory:
/> cd iscsi/iqn.2014-08.com.example:t1/tpg1 /iscsi/iqn.20...ample:t1/tpg1> ls o- tpg1 ................................................. [no-gen-acls, no-auth] o- acls ............................................................ [ACLs: 0] o- luns ............................................................ [LUNs: 0] o- portals ...................................................... [Portals: 1] o- 0.0.0.0:3260 ....................................................... [OK]
Below tpg1, three objects have been defined:
- acls (access control lists: restrict access to resources),
- luns (logical unit number: define exported resources),
- portals (define ways to reach the exported resources; consist in pairs of IP addresses and ports).
If you use a version pre-RHEL 7.1 (this step is now automatically done by the iscsi/ createcommand), you need to create a portal (a pair of IP address and port through which the target can be contacted by initiators):
/iscsi/iqn.20...ple:t1/tpg1> portals/ create Using default IP port 3260 Binding to INADDR_ANY (0.0.0.0) Created network portal 0.0.0.0:3260.
Whatever version, create a lun depending on the kind of backstore you previously chose:
- Fileio backstore:
/iscsi/iqn.20...ample:t1/tpg1> luns/ create /backstores/fileio/shareddata Created LUN 0.
- Block backstore:
/iscsi/iqn.20...ample:t1/tpg1> luns/ create /backstores/block/block1 Created LUN 0.
Create an acl with the previously created IQN (here iqn.2014-08.com.example) and an identifier you choose (here client), together creating the future initiator name:
/iscsi/iqn.20...ample:t1/tpg1> acls/ create iqn.2014-08.com.example:client Created Node ACL for iqn.2014-08.com.example:client Created mapped LUN 0
Optionally, set a userid and a password:
/iscsi/iqn.20...ample:t1/tpg1> cd acls/iqn.2014-08.com.example:client/ /iscsi/iqn.20...xample:client> set auth userid=usr Parameter userid is now 'usr'. /iscsi/iqn.20...xample:client> set auth password=pwd Parameter password is now 'pwd'.
Now, to check the configuration, type:
/iscsi/iqn.20...om.example:d1> cd ../.. /iscsi/iqn.20...ple:tgt1/tpg1> ls o- tpg1 ................................................. [no-gen-acls, no-auth] o- acls ............................................................ [ACLs: 1] | o- iqn.2014-08.com.example:client ......................... [Mapped LUNs: 1] | o- mapped_lun0 ............................. [lun0 fileio/shareddata (rw)] o- luns ............................................................ [LUNs: 1] | o- lun0 .......................... [fileio/shareddata (/opt/shareddata.img)] o- portals ...................................................... [Portals: 1] o- 0.0.0.0:3260 ....................................................... [OK]
Finally, you can quit the targetcli command:
/iscsi/iqn.20...ple:tgt1/tpg1> exit Global pref auto_save_on_exit=true Last 10 configs saved in /etc/target/backup. Configuration saved to /etc/target/saveconfig.json
Note: The configuration is automatically saved to the /etc/target/saveconfig.json file.
Also, it can be useful to check the ports currently used:
# netstat -ant Active Internet connections (servers and established) Proto Recv-Q Send-Q Local Address Foreign Address State tcp 0 0 0.0.0.0:22 0.0.0.0:* LISTEN tcp 0 0 127.0.0.1:25 0.0.0.0:* LISTEN tcp 0 0 0.0.0.0:3260 0.0.0.0:* LISTEN tcp 0 0 192.168.1.81:22 192.168.1.81:33584 ESTABLISHED tcp6 0 0 :::22 :::* LISTEN tcp6 0 0 ::1:25 :::* LISTEN
Finally, open the 3260 tcp port in the firewall configuration:
# firewall-cmd --permanent --add-port=3260/tcp Success
Reload the firewall configuration:
# firewall-cmd --reload Success
iSCSI Initiator Configuration
To create an iSCSI initiator, you need to follow several steps on the client virtual machine.
Install the following package:
# yum install -y iscsi-initiator-utils
Edit the /etc/iscsi/initiatorname.iscsi and replace the content with the initiator name that you previously configured as acl on the target side:
InitiatorName=iqn.2014-08.com.example:client
If you previously set up a userid and a password on the server, edit the /etc/iscsi/iscsid.conffile and paste the following lines:
node.session.auth.authmethod = CHAP node.session.auth.username = usr node.session.auth.password = pwd
Start the iscsi service:
# systemctl start iscsi
Caution: This action is mandatory to be able to unmount the remote resource when rebooting. Don’t confuse iscsid and iscsi services!
Execute the iscsiadm command in discovery mode with the server ip address (here 192.168.1.81):
# iscsiadm --mode discovery --type sendtargets --portal 192.168.1.81 192.168.1.81:3260,1 iqn.2014-08.com.example:t1
Note1: If you don’t specify any port, the default port is 3260.
Note2: Don’t mention a DNS entry as your portal address (here 192.168.1.81), this would be a bad idea causing you a lot of trouble.
Execute the iscsiadm command in node mode with the server ip address (here 192.168.1.81):
# iscsiadm --mode node --targetname iqn.2014-08.com.example:t1 --portal 192.168.1.81 --login Logging in to [iface: default, target: iqn.2014-08.com.example:t1, portal: 192.168.1.81,3260] (multiple) Login to [iface: default, target: iqn.2014-08.com.example:t1, portal: 192.168.1.81,3260] successful.
Note: As before, if you don’t specify any port, the default port is 3260. Use of DNS entry as portal address only brings problems.
To check the configuration, type:
# lsblk --scsi NAME HCTL TYPE VENDOR MODEL REV TRAN sda 2:0:0:0 disk LIO-ORG shareddata 4.0 iscsi
To be sure that your resource is not in read-only mode (1=read-only mode), type:
# lsblk | egrep "NAME|sda" NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT sda 8:0 0 100M 0 disk
Now, you can create a file system:
# mkfs.ext4 /dev/sda mke2fs 1.42.9 (28-Dec-2013) /dev/sda is entire device, not just one partition! Proceed anyway? (y,n) y Filesystem label= OS type: Linux Block size=1024 (log=0) Fragment size=1024 (log=0) Stride=0 blocks, Stripe width=4096 blocks 25688 inodes, 102400 blocks 5120 blocks (5.00%) reserved for the super user First data block=1 Maximum filesystem blocks=33685504 13 block groups 8192 blocks per group, 8192 fragments per group 1976 inodes per group Superblock backups stored on blocks: 8193, 24577, 40961, 57345, 73729 Allocating group tables: done Writing inode tables: done Creating journal (4096 blocks): done Writing superblocks and filesystem accounting information: done
Retrieve the UUID of this disk:
# blkid | grep "/dev/sda" /dev/sda: UUID="4a184c70-20ad-4d91-a0b1-c2cf0eb1986f" TYPE="ext4"
Add the disk UUID to the /etc/fstab file:
# echo "UUID=..." >> /etc/fstab
Note: Be very careful to type >> and not >, otherwise this will destroy all your configuration!
Make a copy of the /etc/fstab file before doing this operation if you don’t want to take any risk.
Edit the /etc/fstab file and add the mount point (here /mnt), the file system type (here ext4) and the mount options (_netdev):
UUID=... /mnt ext4 _netdev 0 0
Note: The _netdev mount option is mandatory to postpone the mount operation after the network initialization. If you don’t do it, the initiator boot process will be stopped after a timeout in maintenance mode.
To check your configuration, type:
# mount /mnt # touch /mnt/testFile
Optionally, you can dump all the initiator configuration (3=max output, 0=min output):
# iscsiadm -m session -P 3
iSCSI Transport Class version 2.0-870
version 6.2.0.873-28
Target: iqn.2014-08.com.example:t1 (non-flash)
Current Portal: 192.168.1.81:3260,1
Persistent Portal: 192.168.1.81:3260,1
**********
Interface:
**********
Iface Name: default
Iface Transport: tcp
Iface Initiatorname: iqn.2014-08.com.example:client
Iface IPaddress: 192.168.1.10
Iface HWaddress:
Iface Netdev:
SID: 1
iSCSI Connection State: LOGGED IN
iSCSI Session State: LOGGED_IN
Internal iscsid Session State: NO CHANGE
*********
Timeouts:
*********
Recovery Timeout: 120
Target Reset Timeout: 30
LUN Reset Timeout: 30
Abort Timeout: 15
*****
CHAP:
*****
username: usr
password: ********
username_in:
password_in: ********
************************
Negotiated iSCSI params:
************************
HeaderDigest: None
DataDigest: None
MaxRecvDataSegmentLength: 262144
MaxXmitDataSegmentLength: 262144
FirstBurstLength: 65536
MaxBurstLength: 262144
ImmediateData: Yes
InitialR2T: Yes
MaxOutstandingR2T: 1
************************
Attached SCSI devices:
************************
Host Number: 2 State: running
scsi2 Channel 00 Id 0 Lun: 0
Attached scsi disk sda State: running
Source: targetcli man page and Linux-iSCSI wiki.
Useful Tips
Before rebooting, set up a virtual console, this can be helpful!
If you need to shut down target and initiator, shut down the initiator first. If you shut down the target first, the initiator won’t be able to unmount the remote resource and will be stuck in the shutdown process.
During the exam, as an extra precaution, unmount the remote resource before rebooting the initiator, you will avoid any bad surprise.
Additional Resources
In addition, you can watch CalPOP’s video Creating iSCSI SAN Storage on Linux (CentOS 7.0) (10min/2015), Venkat Nagappan’s video Setting up iSCSI Target & Initiator (19min/2015) or follow this IBM iScsi tutorial.
There is also a wiki about Targetcli.
Dell offers some interesting information about iSCSI, MPIO and performance tips in its RHEL Configuration Guide for Dell Storage PS Series Arrays.