Here's what you need when upgrading to Lab Management 2.5.
When upgrading to
Lab Management 2.5, please see
the following upgrade notes as well:
The following tasks explain what you need do to migrate the Lab Management manager from version 2.4 on RHEL 5 (32-bit) to
version 2.5 on RHEL 6.1 (64-bit). These instructions assume that the manager to be migrated
is using old VMware Server technology rather than ESXi. The instructions reference
mgr.f.sp.collab.net as an example.
Migrate the AD server
In the old setup, the AD server is set up as a guest node to the manager. For example,
ad.f.sp.collab.net is a guest of
mgr.f.sp.collab.net. So as a first step, you need to migrate this
to ESXi.
-
Install VMware OVF Tool.
-
Log into mgr.f.sp.collab.net.
-
Download the tool from
http://communities.vmware.com/community/vmtn/server/vsphere/automationtools/ovf
and install it in the /opt directory.
-
Copy the AD image to the ESXi server.
-
Stop VMware using the following command.
sudo vmctl stop
-
Edit
/var/lib/vmware/VirtualMachines/ad.f.sp.collab.net/conf.vmx.
Disable cdrom and floppy devices by setting the value present to FALSE in each
corresponding section.
-
Run the following command to copy the VMware image to the ESXi server.
In this example, we deploy VM as the guest "ad.f.sp.collab.net", to datastore
"datastore1", resource pool "F", and network group "f.sp.collab.net" on the ESXi
server devmgr2.sp.collab.net. If you are deploying to the default
resource pool, just leave out the resource pool name ("/F" in the command line
argument).
sudo /opt/vmware/ovftool/ovftool --powerOffSource -ds="datastore1" --name="ad.f.sp.collab.net"
--network="f.sp.collab.net"
/var/lib/vmware/VirtualMachines/ad.f.sp.collab.net/conf.vmx vi://devmgr2.sp.collab.net/F
-
Use vSphere client to start the AD server.
-
After the AD server is up, configure it to use static IP, something like
10.1.53.5.
-
Run the following commands and check the output to verify whether the new AD server is
up and has the necessary data.
cd /usr/local/cubit-mgr/bin
# To display users on the AD
sudo ./checkAccts -U
# To display groups on the AD
sudo ./checkAccts -G
Note: If the customer currently has a WDS server running on the manager as a
VMware-server guest, we will delete it, and rebuild a new WDS in the AD server box from
scratch.
Migrate the manager
-
Get a copy of the latest Lab Management 2.5 build.
-
Back up Lab Management-related data (assumes
/public/backup is a mounted backup volume).
sudo mkdir /public/backup/mgr.$THEDOMAIN
sudo ./cubit_backup.py -d /public/backup/mgr.$THEDOMAIN --skip_vm -c /var/ops/site.conf --skip-data
Important: During migration from a pre-2.5 32-bit system to a Lab Management 2.5 64-bit system, you need to run
/usr/local/cubit-mgr/bin/backup_cricket_data.py after the normal
cubit_backup.py script to dump cricket data.
-
Prepare the physical or virtual host that will house the Lab Management manager.
Note: The OS type should be 64-bit-capable -- "RedHat Linux 6 (64-bit)". It should
have two NICs, one for standard network traffic, and another for the ILO network. Use
the latest SCSI Controller, LSI Logic SAS.
-
Install RHEL 6.1 64-bit on the host.
Once the virtual machine is created, install RedHat Linux 6.1 on the machine. For this,
you need two images: floppy and DVD.
-
The floppy image is located at
/public/os/ks-6.1-x86_64.flp.
-
The DVD image is located at
/public/os/rhel-server-6.1-x86_64-dvd.iso.
-
Connect the floppy image as VM's floppy drive. Connect the DVD image as VM's
"CD/DVD drive".
Both should be connected at boot.
-
Use the above images to boot the newly built machine.
The system should boot from DVD.
-
Press the ESC key from the boot menu to get a prompt.
-
At the prompt, type linux ks=floppy"'.
This will install RedHat 6.1 OS and necessary packages.
The default root password is changeme.
-
Set up the network.
-
Edit /etc/sysconfig/network-scripts/ifcfg-eth0 and ensure that
it has correct information.
-
Edit /etc/sysconfig/network-scripts/ifcg-eth1 and ensure that
it has correct information.
-
Edit /etc/sysconfig/network and ensure that it has correct
information.
-
Edit /etc/hosts and add an entry for "mgr.$THEDOMAIN", mapped
to the primary IP of the host.
-
Stop the old manager by powering off the machine.
-
Reboot the new manager.
-
Fix /etc/fstab.
Mount the backup volume (see step 5) to /mnt/backup.
cd /mnt/backup/mgr.$THEDOMAIN/system_info
cat nfs_entries >> /etc/fstab
-
Reboot the system.
-
Modify /var/ops/site.conf file to add new tokens.
For example, the following new tokens need to be set.
cubit_esx_user = cubit
cubit_esx_user_pwd = secret
cpu_archs = i386, x86_64
cpu_names = Xeon, Athlon, Opteron
models = HP DL320, HP DL380, HP z600, HP xw4600, HP xw6600, Generic, Fauxst
######################
# MANAGER
#
## This is the manager itself. This description is informational, but describes
## the default settings for the manager; it is *strongly* encouraged that you do
## not change these unless you need to:
## note: if mgr is deployed as a guest, for example, on an ESX server, you need
## to set "parent_hostname" attribute:
## parent_hostname = myparent.example.com
##
[infrastructure:mgr]
hostname = %(cubit_mgr_hostname)s
profile_name = rhel6.1_base_x86_64
install_mode = kickstart-net
cubit_manager = %(cubit_mgr)s
mgr_public_gateway = %(cubit_mgr_gateway)s
parent_hostname = pn002.f.sp.collab.net
memsize = 4096
memsize_unit = MB
disksize = 100
disksize_unit = GB
mac_addr = 00:0C:29:9D:0F:E8
[infrastructure:primary_ad_server]
ip_address = 10.1.53.5
parent_hostname = pn002.f.sp.collab.net
hostname = ad.%(cubit_domain)s
profile_name = Windows2k3stdR2_dc
mac_addr = 00:50:56:00:AD:F1
target_port_group = f.sp.collab.net
ncpu = 4
cpuarch = x86_64
cpuname = Xeon
cpumhz = 2000
cpucores = 1
memsize = 512
memsize_unit = MB
disksize = 20
disksize_unit = GB
wds_server = true
[infrastructure:pn002.f.sp.collab.net]
ip_address = 10.1.53.12
hostname = pn002.%(cubit_domain)s
profile_name = esx_4.0u1
unmanaged = true
datastore = datastore1
resource_pool = F
port_group = f.sp.collab.net
mac_addr = 00:50:56:70:92:20
memsize = 34000
memsize_unit = MB
disksize = 800
disksize_unit = GB
cpuarch = x86_64
cpucores = 2
ncpu = 4
#[infrastructure:miniroot]
#ip_address = 10.1.53.6
#parent_hostname = %(cubit_mgr_hostname)s
#hostname = miniroot.%(cubit_domain)s
-
Install Lab Management using the following commands:
cp /public/backup/mgr.$THEDOMAIN/site.conf /var/ops
cd /var/ops
tar xzvf /public/download/cubit-2.5-9653.dev.tgz
cd /var/ops/cubit-2.5-9653
./install.py -f /var/ops/site.conf
-
Restore Lab Management-related data from old manager.
mv /u1/cubit/data /u1/cubit.data.XXX
/usr/local/cubit-mgr/installation-support/scripts/stop-services.py
cd /public/backup/mgr.$THEDOMAIN/system_info
cat nfs_entries >> /etc/fstab
cd /public/download
./add_users
cd /var/ops/cubit-2.5-9653
./cubit_restore.py -d /public/backup/ mgr.$THEDOMAIN
-
Mount the Lab Management data folder to /u1/cubit/data.
-
Perform the migration -- use the following command to recreate runtime and run
migration scripts:
/usr/local/cubit-mgr/recreate-runtime.py -f /var/ops/site.conf --migrate
-
Update profiles.
Since you are migrating from an older version, the profiles in the migrated manager
might be older than the installer. It's always advisable to update the profiles to the
newer version. Use the following commands to do this:
-
Check out the local ai profile information:
svn co file:///u1/cubit/data/repos/ai/profiles
-
Copy updated/required profiles from
$CUBIT_MGR_DIR/installation-support/default-data/ai/profiles to
the profiles working copy.
-
Check in the required profiles to the manager repository using username
cubitrepo.
Install VMware-ovftool on the new manager
-
Download the tool from
http://communities.vmware.com/community/vmtn/server/vsphere/automationtools/ovf
and install it in the /opt directory.
Install VMware-vSphere-CLI on the new manager
-
Get VMware-vSphere-CLI from http://www.vmware.com/downloads/download.do?downloadGroup=VCLI41 and
run the following commands:
cd /tmp
sudo tar xzvf VMware-vSphere-CLI-4.1.0-254719.x86_64.tar.gz
cd vmware-vsphere-cli-distrib
sudo ./vmware-install.pl # Choose install location as /usr/local/vmware-vicli/bin
Rejoin the manager node into the AD domain
-
Delete the manager node from the AD server.
-
On the AD server, click .
-
Open CubitComputers OU.
-
Right-click the manager entry and select Delete.
-
On the MGR node, join the manager to the AD domain (assuming the AD domain
f.sp.collab.net is being used for the Lab Management domain).
$ /usr/bin/sudo kinit Administrator@F.SP.COLLAB.NET <<<---- domain name in all upper cases
Password: (this is your own password for sudo)
Password for Administrator@F.SP.COLLAB.NET: (type domain administrator password here)
$ /usr/bin/sudo net ads join -U Administrator
Administrator's password: (type domain administrator password here)
Using short domain name -- SERVERS
Joined 'MGR' to realm 'F.SP.COLLAB.NET'
$ /usr/bin/sudo net ads testjoin
Join is OK
-
Restart the samba service on the manager.
Set up WDS
-
Create WDS.
-
In the AD box, verify that WDS is installed.
-
Follow these steps to add Windows 7 and Windows 2008
server images to the WDS server.
-
Once the boot and install images are added to WDS, follow the steps under Configure the WDS server to support more
images to set the default boot program and images.
Note: If you have a separate NAS mounted volume as another drive (say E:), and will
use this as the WDS
RemoveInstall folder, make sure you say
"E:\RemoteInstall" here:
wdsutil /Initialize-Server /RemInst:"E:\RemoteInstall"
-
Add the CapturedImages folder to WDS.
-
On the AD server, click .
-
Click the WDS server node, right-click Install Images and
select Add Image Group.
-
Type "CapturedImages" and click OK.
For instructions on storing WDS images on a network share, see
Store WDS images on a network share.