ISCSI

Information in configuring open-iscsi by clients (not to become a storage provider)

Setup and initial configuration

Install

apt-get install open-iscsi

This should create your unique initiator name which if you need to configure access control on the iscsi server you can get by

tail -n1 /etc/iscsi/initiatorname.iscsi

Once you know there is a volume available for use with the server being configured use

iscsiadm -m discovery -t sendtargets -p portal.ip.add.ress

NB portal IP address may not be the same address as used for managing the SAN

to get the name of the volume available to the server. This will be in the form of

storage.ip.add.ress:port,1 uniqueinitiatornameofstorage-volumename

Depending on the storage server, the initiator name can be quite long, but you just need to be able to cut and paste parts into the next command:

iscsiadm -m node -T uniqueinitiatornameofstorage-volumename -p storage.ip.add.ress:port -l

The iscsi volume should now be available as a disk device to be managed by the OS. Check by

ls /dev/disk/by-path/

The output will tell you what disk device to address e.g.

# ls -l /dev/disk/by-path/
total 0
lrwxrwxrwx 1 root root  9 Nov  1 16:38 ip-storage.ip.add.ress:3260-uniqueinitiatornameofstorage-volumename-lun-0 -> ../../sda

where sda is the key element. Using that information create a volume on the drive:

parted /dev/sda

Then create a filesystem as needed

mkfs.ext4 /dev/sda1

(what disk filesystem you use depends on need, but assuming backup etc. is taken care of, EXT4 probably best all-round performance option)

NB if using ext4, the defaults leave 5% of space reserved for root operations. If the volume is for use for data storage, then you should reduce this with

tune2fs -m 1 /dev/sda1

To ensure your iscsi volume is available on reboot

vi /etc/iscsi/iscsid.conf

and set

node.startup = automatic

Then edit the settings for your iscsi node

vi /etc/iscsi/nodes/uniqueinitiatornameofstorage-volumename/ip-storage.ip.add.ress,3260,1/default

and set

node.conn[0].startup = automatic

Note if you only have one iscsi volume on a server, there will only be one nodes subdirectory so Tab autocomplete is your friend.

Now mount the volume (note if this is to extend space in use you may need to mount the volume at a temporary location, move existing data and then remount at the desired permanent location. This best done with the UUID, which got created when the volume did.

mkdir /path/to/mymountpoint
ls -l /dev/disk/by-uuid
mount /dev/disk/by-uuid/youruuid /path/to/mymountpoint

Assuming success (best to test you can write to the mounted volume) next make the mount auto remount on reboot by

vi /etc/fstab

and appending a line of the form

UUID=youruuid /path/to/mymountpoint      ext4    defaults,relatime,_netdev        0       0

for normal file stores, and

UUID=youruuid /path/to/mymountpoint      ext4    defaults,noatime,_netdev        0       0

for backups (where noatime means access times are not recorded, which should preserve the accesstime from when the file was last accessed on the active store). Where a key option is _netdev, which tells the operating system not to try and mount this volume until the network is up.

Troubleshooting

Logging out of ISCSI sessions

If you need to logout of the iscsi storage to allow changes to the volume etc., unmount it and run

iscsiadm -m node -T uniqueinitiatornameofstorage-volumename -p storage.ip.add.ress:port -u

To logout of all just run

General information

To see all known nodes

iscsiadm -m node

To see current active sessions

iscsiadm -m session -P 3