[lustre-discuss] Problems setting up lustre 2.8

[Phill Harvey-Smith]

Hi all,

As we will soon be replacing our lustre servers, I am currently trying 
to setup a test filesystem to practice the process of setting it all up.

I've installed Centos 7.2 on a couple of old server nodes to use as my 
test mds/mdt and oss.

I have been following the setup guide from :
http://doc.lustre.org/lustre_manual.pdf

I have the server modules installed with yum and have started the lnet 
by loading the lnet module with modprobe.

yum --nogpgcheck install kernel-3.10.0-327.3.1.el7_lustre.x86_64.rpm \
  lustre-2.8.0-3.10.0_327.3.1.el7_lustre.x86_64.x86_64.rpm \
  lustre-modules-2.8.0-3.10.0_327.3.1.el7_lustre.x86_64.x86_64.rpm \
  lustre-osd-zfs-mount-2.8.0-3.10.0_327.3.1.el7_lustre.x86_64.x86_64.rpm \
  lustre-osd-zfs-2.8.0-3.10.0_327.3.1.el7_lustre.x86_64.x86_64.rpm \
  lustre-iokit-2.8.0-3.10.0_327.3.1.el7_lustre.x86_64.x86_64.rpm

modprobe lnet

The documentation tells me to use lnetctl to configure then network, 
this command does not exist, though oddly it's manual page does! This 
probably doesn't matter (in the test setup) as I only have one active 
network interface on the two machines.

I want to use zfs as my backing filesystem so have setup a zfs pool on 
two of my disks :

zpool status
   pool: testpool
  state: ONLINE
   scan: scrub repaired 0 in 0h0m with 0 errors on Mon Sep 12 10:43:07 2016
config:

         NAME        STATE     READ WRITE CKSUM
         testpool    ONLINE       0     0     0
           mirror-0  ONLINE       0     0     0
             driveb  ONLINE       0     0     0
             drivec  ONLINE       0     0     0

errors: No known data errors

However I'm not sure what to use as the block device to be fed to 
mkfs.lustre ? as the documentation seems a little sketchy regarding zfs.

Can anyone assist please?

Cheers.

Phill.