Skip to content

Storage Add - Fujitsu Eternus

Warning

This process will require restarting/reloading several core storage services. Whilst it seems to work online, ALWAYS do this in a change window.

Step 1 - Add multipath conf entry

vim /etc/multipath.conf In the devices section, after the pure storage (new line, no coma required) add

        device {
            vendor "FUJITSU"
            product "ETERNUS_DX.*"
            path_grouping_policy group_by_prio
            path_selector "round-robin 0"
            hardware_handler "1 alua"
            prio alua
            failback immediate
            no_path_retry 12
            rr_weight uniform
            rr_min_io 100
            path_checker tur
        }

Step 2 - Validate multipath file BEFORE RESTARTING

It will throw out the file if normal. Just check if any errors appear. 

multipath -t

Step 3 - Restart multipath

Information

Initial testing did not appear to have any material impact on existing iscsi connections

systemctl restart multipathd
# Check status to verify
systemctl status multipathd

Warning

This guide was built around the DX600 that was/is used in the OCS JNB1 prox platform. You WILL need to update specific settings for a different platform (e.g IP Addresses, IQNs )

Step 4 - Blacklist autodiscovered nodes which are invalid for this host

Since the host may find all possible nodes from the storage, it may think that it can use all paths. In this case, it cannot use paths on the 10.30.0.0/16 subnet, so we can blacklist them beforehand Edit /etc/iscsi/iscsid.conf and add the following at the end of the file (if it does not already exist). 

# Blacklist undesirable nodes being provided by the Eternus
node.blacklist = 10.30.*

Step 5 - Restart the iscsi services

To apply the blacklisting we need to restart the iscsi services. For some reason, when restarting iscsi.service it hung for a while. After leaving it for a period of time we ctrl+c exited the attempt and re-ran it, and it was effectively instant. 

 systemctl restart iscsid
 systemctl status iscsid
 systemctl restart iscsi
 systemctl status iscsi

Step 6 - Check/login to iscsi ports on Fujitsu

We need to inform the Fujitsu that we are available 

iscsiadm -m discovery -t sendtargets -p 10.13.1.110
iscsiadm -m discovery -t sendtargets -p 10.13.1.111
iscsiadm -m discovery -t sendtargets -p 10.14.1.111
iscsiadm -m discovery -t sendtargets -p 10.14.1.110

Step 7 - Remove autodiscovered junk iscsi nodes

The fujitsu provides a bunch of other nodes (from an old config), so we remove those 

iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.3.110 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.3.111 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.4.110 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.4.111 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.5.110 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.5.111 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.6.110 --op delete
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.30.6.111 --op delete

Step 8 - Login in to the desired nodes

Below will add the node first, then set the node to auto login on startup 

iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.13.1.110:3260 --login
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.13.1.110:3260 --op update -n node.startup -v automatic
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.14.1.110:3260 --login
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.14.1.110:3260 --op update -n node.startup -v automatic
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.13.1.111:3260 --login
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.13.1.111:3260 --op update -n node.startup -v automatic
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.14.1.111:3260 --login
iscsiadm -m node -T iqn.2000-09.com.fujitsu:storage-system.eternus-dxm:002a3060 -p 10.14.1.111:3260 --op update -n node.startup -v automatic

Step 9 - Now we can rescan the iscsi bus

  iscsiadm -m node rescan
  iscsiadm -m session rescan
We can confirm the devices are now visible with lsblk

Step 10 - Check if paths are there and multipath has consumed them

# Check if new sdX devices are at end of list and if they have mpathX in them with
lsblk
# Display all newly discovered devices (should be 2x20TB luns with Fujitsu in the name) 
multipath -ll

You should see 2 new 20TB fujitsu devices. If you see the sd devices but NOT the multipath devices run multipath -r and then check again with multipath -ll

Step 11 - Reboot the node to confirm that the node recovers correctly

THIS IS AN OPTIONAL STEP 

init 6

You will now need to register the data store in proxmox