Guide to clone_sles_instances

SUSE Linux Enterprise Server - Rapid Deployment of Instances
1
Satheesh Thomas
Guide to Clone SLES
Instances for Quick
Deployments
Satheesh Thomas – satheeshthomas@in.ibm.com

2
Satheesh Thomas
Contents
Introduction......................................................................................................................... 3
Deployment Scenarios ........................................................................................................ 3
Gold Image Preparation ...................................................................................................... 3
Deploy New Instance – Flash Copy Method ...................................................................... 5
Deploy New Instance – LVM & VSCSI Method ............................................................... 5
Reset RMC configuration & SSH Keys.............................................................................. 6
References........................................................................................................................... 6

3
Satheesh Thomas
Introduction
Different methods are available for deploying SLES instances. Cloud manager, auto-
mation tools, autoyast are some of them. Usually a gold copy of instance is created
with all required software packages and this image is replicated to new LPARs.
Autoyast or other tools require additional configurations including setting up install
servers or management servers etc. before images can be replicated to new LPARs.
This document describes a different approach to clone the gold copy without the need
for any install servers and by using minimal efforts for faster deployments
Deployment Scenarios
This method can be used in any of the following scenarios
 LPAR Boot disk is SAN based and Storage Copy Services are enabled (Flash
Copy)
 LPAR Boot disk is VSCSI and backing device is a Logical Volume in VIOS
Note: Unlike the other automation tools where post deployment configurations can be
automated or can be part of the deployment process itself, this method require manual
efforts for post implementation configurations like IP Address and other system
changes
Note: Ensure that the Gold image LPAR’s IP address is down during new de-
ployments unless gold image network config file is edited for new VM build.
The cloned image will have same IP address and can cause undesirable ef-
fects due to IP duplication
The methods are tested on the following cases
SLES 11 SP4
IBM Storvise V5010 storage used for SAN boot with NPIV virtualization
VSCSI based deployment using interval SAS disks on VIOS
Gold Image Preparation
Prepare an existing SLES instance as a gold image for deployment. This can be a new
install using traditional methods.
The following activities to be performed during initial setup only.
OS Preparation
 Tune the instance as per IBM best practices (multipath, I/O scheduler, LVM
filter, ulimit etc.)
 Install additional packages (IBM Tools, Additional bundles, Application related
files etc.)
 Configure required servers (NTP, DNS etc.)
 Delete or clean up log files

4
Satheesh Thomas
Boot Devices
 Configure boot device to use by-name and root devices to use UUID
o Root device change is not required if LVM is used for root
 Modify /etc/lilo.conf and change boot device from by-id to by-name represen-
tation
 Edit /etc/lilo.conf, making the root line correspond to the device UUID, and the
boot line correspond to the device name.
The following example shows what the lines may look like before you edit them:
boot = /dev/disk/by-id/scsi-360050768028180ee380000000000603c-part1
In this example, you would change those lines so they refer to the specific device
names, as follows:
boot = /dev/sda1
Sample /etc/lilo.conf
 Run lilo command
 Run mkinitrd command
Network Devices
 Change the network device rule file so that new image will create network
adapter as eth0. Otherwise cloned images will create a new network device like
eth1
 Edit /etc/udev/rules.d/70-persistent-net.rules and remove HW address binding
to eth0 as follows
SUBSYSTEM=="net",ACTION=="add",DRIVERS=="?*",ATTR{ad-
dress}=="ea:8e:37:f5:11:0c",ATTR{dev_id}=="0x0",ATTR{type}=="1",
KERNEL=="eth*",NAME="eth0"
SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{ad-
dress}=="?*", ATTR{dev_id}=="0x0", ATTR{type}=="1", KER-
NEL=="eth*", NAME="eth0"
IP Address Change (Optional) – Changes must be undone
This change must be made every time before deploying a new VM. If omitted cloned
LPAR will be activated with the same IP as of Gold Image LPAR
 Edit /etc/sysconfig/network file and change IP address to the desired IP address

5
Satheesh Thomas
Hostname Change (Optional) - Changes must be undone
Update hostname for new VM
 Update /etc/hosts file and /etc/HOSTNAME file
Deploy New Instance – Flash Copy Method
Complete the following pre-requisites for Flash Copy base deployment
 Create a new LPAR
 Complete NPIV mappings
 Complete SAN Zoning for new LPAR
Deploy the new LPARs by
 Take flash copy of the gold image disk using Storage GUI or CLI methods with
auto delete option
 Assign flash copy disk to the new LPAR
 Boot the LPAR
 Select boot device and set bootlist in SMS in case LPAR is not detecting the
boot disk automatically
Perform post deployment validation
 Set IP address and hostname (if required)
 Reboot and validate
Note: Flash Copy creation and attaching the Flash Copy is out of scope of this
document
Deploy New Instance – LVM & VSCSI Method
This method is used for environments where Logical volumes are used as backing
device for client rootvg disks. They use VSCSI Mappings on VIOS. In this scenario the
Gold Image LPAR is also booted on a LVM backed device and the new LPAR boot device
will be created on the same volume group as of the gold image rootvg disk
This method is most suited for new deployments since this require downtime for the
Gold Image LPAR and the LVM copy process is relatively slow
Complete the following pre-requisites for new deployment
 Create a new LPAR
Prepare the Gold Image LPAR for Copy
 Shutdown the Gold Image LPAR
 Detach / Remove vhost mapping rootvg disk of Gold Image LPAR

6
Satheesh Thomas
 Ensure volume group is having sufficient space to accommodate the new disk
 Initiate LVM copy using cplv
In this example a new LV name dr-ecpr_root will be created by copying from
LV dr-bwep_root
 Map the new LV to the new LPAR created
 Boot the LPAR
 Select boot device and set bootlist in SMS in case LPAR is not detecting the
boot disk automatically
 Re-Map rootvg disk of Gold Image LPAR and boot Gold Image LPAR (After all
deployments completed)
Perform post deployment validation
 Set IP address and hostname (if required)
 Reboot and validate
Reset RMC configuration & SSH Keys
Reconfigure RMC and recreate SSH Keys after the new VM is created
 Reset RMC by running /usr/sbin/rsct/install/recfgct
Cloud-init for automating post implementation
Activation engine and cloud-init are technologies that take user input and configure
the operating system and software on deployed virtual machines. Cloud-init is a script
that enables virtual machine activation and initialization, and is widely used for Open-
Stack.
Follow the below steps for cloud-init based deployment
 Complete pre-capture preparation on the gold image LPAR such as
o deleting or cleaning up log files
o change the devices so that they are mounted by UUID (Refer: Boot
Devices section above)
 Install the appropriate cloud-init RPM on the Gold Image VM

7
Satheesh Thomas
SUSE Linux 11: cloud-init-0.7.4-2.4.ppc64.rpm
SUSE Linux 12: cloud-init-0.7.5-8.10.ppc64le.rpm
Download cloud-init and its dependencies from
ftp://ftp.unicamp.br/pub/linuxpatch/cloud-init-ppc64/sles11/
Note: These are PowerVC version of the RPM.
 Modify the cloud.cfg file available at /etc/cloud/cloud.cfg with the following val-
ues:
Remove the following field:
users: -root
Add the following fields:
ssh_pwauth: true
ssh_deletekeys: true
disable_ec2_metadata: True
datasource_list: ['ConfigDrive']
Run the following commands.
chkconfig -s cloud-init-local on
chkconfig -s cloud-init on
chkconfig -s cloud-config on
chkconfig -s cloud-final on
Cloud-init is a script that enables virtual machine activation and initialization, and is widely
used for OpenStack Cloud-init is a script that enables virtual machine activation and ini-
tialization, and is widely used for OpenStackCloud-init is widely used in OpenStack.
Cloud-init is the Reconfigure RMC and recreate SSH Keys after the new VM is created
References
References used in this document given below
Changing devices to be mounted by UUID (SUSE Linux only) -
https://www.ibm.com/support/knowledgecenter/en/SSXK2N_1.2.3/com.ibm.powerv
c.standard.help.doc/powervc_changing_devices_hmc.html
Cloned machine does not boot up properly -

8
Satheesh Thomas
https://www.novell.com/support/kb/doc.php?id=3048119
How to rename Ethernet devices through udev
https://www.novell.com/support/kb/doc.php?id=3012993

Guide to clone_sles_instances

More Related Content

What's hot (19)

Similar to Guide to clone_sles_instances (20)

Recently uploaded (20)

Guide to clone_sles_instances