- clone with RAID1 poweroff all nodes and the chassis insert new blade server in the chassis and remove both disks insert one disk from an older node (e.g. hltv0102) and insert it in the new node (keep the same slot position (e.g. right disk goes to right slot in new machine) power up both nodes (old and new, both with only one disk in the raid) the old node should boot normally the new node should boot too. Login with ilo terminal

- on the cloned machine

set hostname in sysconfig/network HOSTNAME=hltv0103.lbdaq.cern.ch

- set IP address for data network (eth1) /etc/sysconfig/network-scripts/ifcfg-eth1

- on the bgvctrl machine add the new hlt node in the dhcp config (in /etc/dhcp/dhcpd.conf)

- edit the following files (add new node name) /etc/pcSrv.conf /etc/tmSrv.allow /etc/cmSrv.allow

- Notes It is important that raw_cap_hack is loaded such that the mep receiver of an HLT node can bind to a network socket while not being started as root. The module raw_cap_hack mus be loaded in each HLT node. The module is normally loaded automatically at boot with the service raw_cap_hack.

check status with # service raw_cap_hack status

if not loaded, then load it with # service raw_cap_hack start

if it fails because the module can’t be found, make sure the module is present in the kernel modules in # cd /lib/modules/$(uname -r)/kernel/net/

if it is not present, then copy it there. if it is present, but will not load because “not found”, then use depmod:

# cd /lib/modules/$(uname -r)/ # depmod

then load it with

# service raw_cap_hack start