Right then. So far (if you have been following along) we have done the following;
- Created a local yum repository based on the installation media on the initial server in our sand boxed lab
- Installed and setup DNS with Forward and Reverse zones
- Installed and configured dhcpd for the lab network
- Tested the DNS and dhcp services using a client machine
- And network enabled our yum repository by exposing the directory using Apache
The final steps to enable an ISO free installation of CentOS 7 into a KVM virtual machine are;
- Installing, configuring and testing Trivial FTP, adding additional configuration to DHCPd to enable PXE booting and testing that setup (this post)
- The final piece will be to create our kickstart file from which will define the standard installation on CentOS 7 (maybe splitting out into server and client)
So, lets not waste any time and get our hands dirty and flex our fingers with a bit of typing…
Trivial FTP (tftp)
First things first, lets log on to the server and get the packages installed. Thankfully they are part of the installation media and therefore part of the yum repository that was set up in the last post.
[toby@rhc-server ~]$ sudo yum install tftp*
Loaded plugins: fastestmirror
baselocal | 3.6 kB 00:00:00
Loading mirror speeds from cached hostfile
Resolving Dependencies
-- Running transaction check
--- Package tftp.x86_64 0:5.2-11.el7 will be installed
--- Package tftp-server.x86_64 0:5.2-11.el7 will be installed
-- Finished Dependency Resolution
Dependencies Resolved
============================================================================================================================================
Package Arch Version Repository Size
============================================================================================================================================
Installing:
tftp x86_64 5.2-11.el7 baselocal 35 k
tftp-server x86_64 5.2-11.el7 baselocal 44 k
Transaction Summary
============================================================================================================================================
Install 2 Packages
Total download size: 79 k
Installed size: 112 k
Is this ok [y/d/N]: y
Downloading packages:
--------------------------------------------------------------------------------------------------------------------------------------------
Total 654 kB/s | 79 kB 00:00:00
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
Installing : tftp-server-5.2-11.el7.x86_64 1/2
Installing : tftp-5.2-11.el7.x86_64 2/2
Verifying : tftp-5.2-11.el7.x86_64 1/2
Verifying : tftp-server-5.2-11.el7.x86_64 2/2
Installed:
tftp.x86_64 0:5.2-11.el7 tftp-server.x86_64 0:5.2-11.el7
Complete!
Note, I’ve installed both client and server RPMs on the server. For my tests from a client I will only install the client tftp package.
Next step is to prepare the folder structure where the files will be served from. Don’t forget to make sure SELinux contexts, etc. are defined correctly otherwise things will not work as expected.
[root@rhc-server lib]# ll -aZ /var/lib/tftpboot
drwxr-xr-x. root root system_u:object_r:tftpdir_rw_t:s0 .
drwxr-xr-x. root root system_u:object_r:var_lib_t:s0 ..
Note that the SELinux type is “tftpdir_rw_t”, we will need to apply this same type to the folder created next.
[root@rhc-server ~]# mkdir /tftpboot
[root@rhc-server ~]# chcon --reference=/var/lib/tftpboot/ /tftpboot
[root@rhc-server ~]# ll -Z /
[.. snip ..]
drwxr-xr-x. root root system_u:object_r:tftpdir_rw_t:s0 tftpboot
I am slightly cheating with the above command as I am using the standard tftp directory SELinux security types and contexts as a reference. Why make things more difficult than they need to be.
And now lets configure the tftp daemon to use this new location by default.
So lets just check where we are;
- tftp server and client installed? – check!
- folder structure created, permissions set and SELinux attributes defined correctly – check!
- tftp server configured? – Nope
Best fix the TFTP configuration side of things before going further;
[toby@rhc-server lib]$ cat /etc/xinetd.d/tftp
# default: off
# description: The tftp server serves files using the trivial file transfer \
# protocol. The tftp protocol is often used to boot diskless \
# workstations, download configuration files to network-aware printers, \
# and to start the installation process for some operating systems.
service tftp
{
socket_type = dgram
protocol = udp
wait = yes
user = root
server = /usr/sbin/in.tftpd
server_args = -s /tftpboot
disable = no
per_source = 11
cps = 100 2
flags = IPv4
}
OK, so the remaining task at this point before we start the service is to make sure the firewall is configured to allow connectivity to the tftp service via its LAN interface and also to make sure the hosts.allow file has an entry for the tftp service. hosts.allow is used by the xinetd processes, and is required in addition to the firewall changes.
Lets get the hosts.allow file out of the way first;
[root@rhc-server ~]# cat /etc/hosts.allow
#
# hosts.allow This file contains access rules which are used to
# allow or deny connections to network services that
# either use the tcp_wrappers library or that have been
# started through a tcp_wrappers-enabled xinetd.
#
# See 'man 5 hosts_options' and 'man 5 hosts_access'
# for information on rule syntax.
# See 'man tcpd' for information on tcp_wrappers
#
in.tftp: ALL
And now for the firewall;
[root@rhc-server ~]# firewall-cmd --zone public --add-service tftp
[root@rhc-server ~]# firewall-cmd --permanent --zone public --add-service tftp
success
It is worth mentioning that if you use the “–permanent” parameter on the command line, it will not be applied immediately. Now we should be good to start the service and do some tests. We will create a test file to try and copy via tftp before performing the tests.
[root@rhc-server ~]# systemctl enable tftp.socket
ln -s '/usr/lib/systemd/system/tftp.socket' '/etc/systemd/system/sockets.target.wants/tftp.socket'
[root@rhc-server ~]# systemctl reload xinetd
Based on the above the service has started successfully and nothing appears to be out of the ordinary in the journal, so lets proceed with the testing, and here is my test file;
[root@rhc-server ~]# echo "Hello? Is it me you're looking for?" > /tftpboot/this_is_a_test
[root@rhc-server ~]# ll -Z /tftpboot/
-rw-r--r--. root root unconfined_u:object_r:tftpdir_rw_t:s0 this_is_a_test
Call me paranoid, but I wanted to make sure the file had inherited the SELinux type of “tftpdir_rw_t”. Which it did 🙂
Testing locally on server
[toby@rhc-server ~]$ tftp -4 localhost
tftp> get this_is_a_test
tftp> quit
[toby@rhc-server ~]$ ls
this_is_a_test
[toby@rhc-server ~]$ cat this_is_a_test
Hello? Is it me you're looking for?
Looking good so far! 🙂
Testing remotely from client
Before we can test lets determine if the tftp package is installed, in my case it wasn’t, so I installed it;
[toby@rhc-client ~]$ yum search tftp
Loaded plugins: fastestmirror, langpacks
Determining fastest mirrors
============================================================ N/S matched: tftp =============================================================
syslinux-tftpboot.x86_64 : SYSLINUX modules in /tftpboot, available for network booting
tftp.x86_64 : The client for the Trivial File Transfer Protocol (TFTP)
tftp-server.x86_64 : The server for the Trivial File Transfer Protocol (TFTP)
Name and summary matches only, use "search all" for everything.
[toby@rhc-client ~]$ yum list tftp
Loaded plugins: fastestmirror, langpacks
Loading mirror speeds from cached hostfile
Available Packages
tftp.x86_64 5.2-11.el7 th_lab_server
[toby@rhc-client ~]$ sudo yum install tftp
[sudo] password for toby:
Loaded plugins: fastestmirror, langpacks
th_lab_server | 3.6 kB 00:00:00
Loading mirror speeds from cached hostfile
Resolving Dependencies
--> Running transaction check
---> Package tftp.x86_64 0:5.2-11.el7 will be installed
--> Finished Dependency Resolution
Dependencies Resolved
============================================================================================================================================
Package Arch Version Repository Size
============================================================================================================================================
Installing:
tftp x86_64 5.2-11.el7 th_lab_server 35 k
Transaction Summary
============================================================================================================================================
Install 1 Package
Total download size: 35 k
Installed size: 48 k
Is this ok [y/d/N]: y
Downloading packages:
warning: /var/cache/yum/x86_64/7/th_lab_server/packages/tftp-5.2-11.el7.x86_64.rpm: Header V3 RSA/SHA256 Signature, key ID f4a80eb5: NOKEY
Public key for tftp-5.2-11.el7.x86_64.rpm is not installed
tftp-5.2-11.el7.x86_64.rpm | 35 kB 00:00:00
Retrieving key from file:///etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7
Importing GPG key 0xF4A80EB5:
Userid : "CentOS-7 Key (CentOS 7 Official Signing Key) <security@centos.org>"
Fingerprint: 6341 ab27 53d7 8a78 a7c2 7bb1 24c6 a8a7 f4a8 0eb5
Package : centos-release-7-0.1406.el7.centos.2.3.x86_64 (@anaconda)
From : /etc/pki/rpm-gpg/RPM-GPG-KEY-CentOS-7
Is this ok [y/N]: y
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
Installing : tftp-5.2-11.el7.x86_64 1/1
Verifying : tftp-5.2-11.el7.x86_64 1/1
Installed:
tftp.x86_64 0:5.2-11.el7
Complete!
And now the test from the client.
[toby@rhc-client ~]$ sudo firewall-cmd --zone public --add-service tftp
[sudo] password for toby:
success
[toby@rhc-client ~]$ tftp rhc-server
tftp> get this_is_a_test
tftp> quit
[toby@rhc-client ~]$ ls
Desktop Documents Downloads Music Pictures Public Templates test this_is_a_test Videos
[toby@rhc-client ~]$ cat this_is_a_test
Hello? Is it me you're looking for?
[toby@rhc-client ~]$ sudo firewall-cmd --zone public --remove-service tftp
success
Note. If you don’t temporarily enable the tftp port on the client, the test will fail. I got the follow error via tcpdump which highlighted that the firewall was blocking the request;
23:25:40.388650 IP (tos 0x0, ttl 64, id 0, offset 0, flags [DF], proto UDP (17), length 54)
192.168.20.50.36588 > 192.168.20.1.69: [udp sum ok] 26 RRQ "this_is_a_test" netascii
23:25:40.390312 IP (tos 0x0, ttl 64, id 31747, offset 0, flags [none], proto UDP (17), length 70)
192.168.20.1.43788 > 192.168.20.50.36588: [bad udp cksum 0xa9c7 -> 0xddd1!] UDP, length 42
23:25:40.390721 IP (tos 0xc0, ttl 64, id 3881, offset 0, flags [none], proto ICMP (1), length 98)
192.168.20.50 > 192.168.20.1: ICMP host 192.168.20.50 unreachable - admin prohibited, length 78
IP (tos 0x0, ttl 64, id 31747, offset 0, flags [none], proto UDP (17), length 70)
192.168.20.1.43788 > 192.168.20.50.36588: [udp sum ok] UDP, length 42
At this point we have now completed the necessary steps to ensure we have a working tftp service.
Setting up PXE
The second part of today’s post covers the steps needed to enable booting from the network to facilitate building machines without the hassle of creating USB bootable images or burning ISO images to CD or DVD.
Out task list for this section is;
- Add the additional config to the DHCP scope
- Ensure we have the pxelinux/syslinux installed and copied to the required location
- Create a basic menu to provide end users with a installation options
- Test
Configuring DHCP
So, lets start by reminding ourselves what the current dhcpd.conf file looks like;
[toby@rhc-server ~]$ sudo cat /etc/dhcp/dhcpd.conf
[sudo] password for toby:
#
# lab.tobyheywood.com dhcp daemon configuration file
#
# 2016-02-22 - Initial creation
#
# Define which IP to listen on. NOTE. daemon can only listen to one
# IP at a time if defined.
local-address 192.168.20.1;
# option definitions common to all supported networks...
option domain-name "lab.tobyheywood.com";
option domain-name-servers ns.lab.tobyheywood.com;
default-lease-time 600;
max-lease-time 7200;
# Use this to enble / disable dynamic dns updates globally.
#ddns-update-style interim;
# This is the authoritative DHCP server.
authoritative;
# Use this to send dhcp log messages to a different log file (you also
# have to hack syslog.conf to complete the redirection).
log-facility local7;
# Interface which can be accessed from outside the sandbox
# **** NOT IN USE ****
subnet 192.168.122.0 netmask 255.255.255.0 {
}
# The lab network
subnet 192.168.20.0 netmask 255.255.255.128 {
range 192.168.20.50 192.168.20.99;
option routers rtr.lab.tobyheywood.com;
}
So in order to get PXE (Preboot eXecution Environment) to function we will need to add a few more lines to the configuration, as highlighted below.
[toby@rhc-server ~]$ sudo cat /etc/dhcp/dhcpd.conf
[sudo] password for toby:
#
# lab.tobyheywood.com dhcp daemon configuration file
#
# 2016-02-22 - Initial creation
#
# Define which IP to listen on. NOTE. daemon can only listen to one
# IP at a time if defined.
local-address 192.168.20.1;
# option definitions common to all supported networks...
option domain-name "lab.tobyheywood.com";
option domain-name-servers ns.lab.tobyheywood.com;
default-lease-time 600;
max-lease-time 7200;
# Use this to enble / disable dynamic dns updates globally.
#ddns-update-style interim;
# This is the authoritative DHCP server.
authoritative;
# Use this to send dhcp log messages to a different log file (you also
# have to hack syslog.conf to complete the redirection).
log-facility local7;
# Interface which can be accessed from outside the sandbox
# **** NOT IN USE ****
subnet 192.168.122.0 netmask 255.255.255.0 {
}
# The lab network
subnet 192.168.20.0 netmask 255.255.255.128 {
range 192.168.20.50 192.168.20.99;
option routers rtr.lab.tobyheywood.com;
}
# Additional configuration for PXE booting
allow booting;
allow bootp;
option option-128 code 128 = string;
option option-129 code 129 = text;
next-server 192.168.20.1;
filename "/pxelinux.0";
Setting up the required syslinux files in /tftpboot
During the section above when testing the tftp service on the client, I saw that there may be a shortcut to getting things up and running. Everything I have read says you need to manually copy the syslinux files into the /tftpboot directory, however, if you search the rpm database with yum for anything tftp related, I see that there is potentially an easier way.
[toby@rhc-server ~]$ yum list tftp
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
Installed Packages
tftp.x86_64 5.2-11.el7 @baselocal
[toby@rhc-server ~]$ yum search tftp
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
=========================================== N/S matched: tftp ===========================================
syslinux-tftpboot.x86_64 : SYSLINUX modules in /tftpboot, available for network booting
tftp.x86_64 : The client for the Trivial File Transfer Protocol (TFTP)
tftp-server.x86_64 : The server for the Trivial File Transfer Protocol (TFTP)
As you can see there appears to be a syslinux-tftpboot rpm. So lets see what it gives us;
[toby@rhc-server ~]$ sudo yum install syslinux-tftpboot
[sudo] password for toby:
Loaded plugins: fastestmirror
baselocal | 3.6 kB 00:00:00
Loading mirror speeds from cached hostfile
Resolving Dependencies
--> Running transaction check
---> Package syslinux-tftpboot.x86_64 0:4.05-8.el7 will be installed
--> Finished Dependency Resolution
Dependencies Resolved
=========================================================================================================
Package Arch Version Repository Size
=========================================================================================================
Installing:
syslinux-tftpboot x86_64 4.05-8.el7 baselocal 425 k
Transaction Summary
=========================================================================================================
Install 1 Package
Total download size: 425 k
Installed size: 1.3 M
Is this ok [y/d/N]: y
Downloading packages:
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
Installing : syslinux-tftpboot-4.05-8.el7.x86_64 1/1
Verifying : syslinux-tftpboot-4.05-8.el7.x86_64 1/1
Installed:
syslinux-tftpboot.x86_64 0:4.05-8.el7
Complete!
[toby@rhc-server ~]$ ls /tftpboot/
cat.c32 dmitest.c32 host.c32 ls.c32 pcitest.c32 rosh.c32 vesamenu.c32
chain.c32 elf.c32 ifcpu64.c32 lua.c32 pmload.c32 sanboot.c32 vpdtest.c32
cmd.c32 ethersel.c32 ifcpu.c32 mboot.c32 poweroff.com sdi.c32 whichsys.c32
config.c32 gfxboot.c32 ifplop.c32 memdisk pwd.c32 sysdump.c32 zzjson.c32
cpuid.c32 gpxecmd.c32 int18.com memdump.com pxechain.com this_is_a_test
cpuidtest.c32 gpxelinux.0 kbdmap.c32 meminfo.c32 pxelinux.0 ver.com
disk.c32 hdt.c32 linux.c32 menu.c32 reboot.c32 vesainfo.c32
It looks like we have a winner! Now we just need to make sure that the Preboot environment has a kernel or two to play with;
[root@rhc-server ~]# cp /software/centos7/images/pxeboot/* /tftpboot/centos7/
[root@rhc-server ~]# ls -Z /tftpboot/centos7/
-rw-r--r--. root root unconfined_u:object_r:tftpdir_t:s0 initrd.img
-r--r--r--. root root unconfined_u:object_r:tftpdir_t:s0 TRANS.TBL
-rw-r--r--. root root unconfined_u:object_r:tftpdir_t:s0 upgrade.img
-rwxr-xr-x. root root unconfined_u:object_r:tftpdir_t:s0 vmlinuz
OK, so lets now create the directory which will hold the PXE menus;
[toby@rhc-server ~]$ sudo mkdir /tftpboot/pxelinux.cfg
[sudo] password for toby:
[toby@rhc-server ~]$ ls -Zd /tftpboot/pxelinux.cfg
drwxr-xr-x. root root unconfined_u:object_r:tftpdir_t:s0 /tftpboot/pxelinux.cfg
And now lets look at what is required to create the menus from which we will be able to select installation options.
[root@rhc-server centos7]# cat /tftpboot/pxelinux.cfg/default
DEFAULT menu.c32
PROMPT 0
TIMEOUT 300
ONTIMEOUT localdisk
MENU TITLE PXE Network Boot
LABEL localdisk
MENU LABEL ^Local Hard Drive
MENU DEFAULT
LOCALBOOT 0
LABEL Install_CentOS_7_2
MENU LABEL CentOS 7.2
KERNEL centos7/vmlinuz
APPEND initrd=http://rhc-server.lab.tobyheywood.com/centos7/isolinux/initrd.img inst.repo=http://rhc-server.lab.tobyheywood.com/centos7
At this point, all I wanted to do was prove that the pxe settings in dhcp were correct and that maybe, just maybe, I could build a vm across the network, first time. No such luck! 🙁 Well, I guess 50% of the way there.
You can skip the next and jump to here if you don’t want to understand the pain I went through to get this operational.
I thought things were going well. I created a blank VM, made sure to select that I was going to boot from the network to install the OS in Virtual Manager and turned the VM on.

So far so good!

I’m now feel pretty happy with myself. I select CentOS 7.0 and hit the enter key… and then am presented with the expected Welcome to CentOS 7 screen, from where I can kick off a manual installation.

So all, in all, things look good. However there is more work to be done, as the next step is to create a kickstart file to define how the base install should look.
Reference Material
Trivial FTP (tftp)
PXE
Featured image taken by Ed Robinson who kindly uploaded it to Flickr.com. I have made no changes to this image and left it in its original form for your viewing pleasure. And to give the page a bit of colour 😉