On 08/26/2016 06:52 PM, [email protected] wrote:
You need to define static IP's in darcut format on the kernel boot command
line now since 7. Look at the kickstart instructions for fedora for details.
Right or wrong the idea behind it is that with IPv6 coming in the future every
one should be using DHCP every where.
Well, DHCPv6 is, ah, interesting. Regardless, there are network
segments and subnets here where I do not want automatic configuration.
Period. It shouldn't be easy for someone to plug something in to this
particular vlan/segment/subnet and just get a usable IP config. Those
are my requirements. In any case, the question was raised about the
requirement of DHCP, and, no DHCP is NOT required. If you pass the
static IP information on the installer command line you won't even have
to change that information in the installer.
In the example I present I am using the VNC installation method, mostly
in order to grab screenshots of the installer. If you're at the console
or using virtualization and have the true console going in a window, you
won't need the portions related to VNC below.
I first needed to know the interface name. For the server I am
installing this is 'eno1' which I found by booting the CentOS install
DVD and running the rescue mode, then issuing an 'ifconfig' from the
rescue shell.
Upon booting the NetInstall CD, I selected 'install' and hit the tab key
to get to the options. My command line (partially sanitized) was:
>vmlinuz initrd=initrd.img inst.stage2=hd:LABEL=entOS\x207\x20x86_64
quiet ip=10.250.130.55::10.250.130.254:24:test.pari.edu:eno1:none
nameserver=192.168.1.3 vnc vncpasswd=nottherealone
When I connected to 10.250.130.55:1 via VNC I was greeted with the
normal installer sequence. The very first thing I did was check the
network settings, and they were carried over from the command line. I
selected the install source as a network install and http:// selected in
the pulldown, and put the URL to the install tree there, following the
instructions at:
http://www.if-not-true-then-false.com/2014/centos-7-netinstall-guide/?PageSpeed=noscript
(see specifically section 3.10).
The install is now proceeding, using my http server I selected. You do
need to make sure the install source is set up properly; this is found
in Red Hat's upstream documentation at:
https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Installation_Guide/s1-steps-network-installs-x86.html#idp19192208
(You need to change a few obvious things for either a CentOS or SL install).
While I was writing that last paragraph, the install completed, and I
have clicked 'reboot' and we'll see if it comes up with the same
address.... yep, once the 'license' was accepted in text mode (manually,
of course, meaning I had to manually type things in on the keyboard
through the KVM switch) everything came up, with the correct IP address,
and I'm successfully logged-in with ssh.
While I know that I was installing CentOS 7, the SL 7 is basically
identical (with different branding, mostly) and should work the same way.
If you don't provide the static IP address information on the command
line you do have the opportunity to select static (the installer uses
the term 'Manual') addressing through the Network & Hostname spoke's
Configure button (located in the lower right quadrant of that spoke's
screen). Select the interface, click configure, click the 'IPv4
Settings' tab, and select 'Manual' from the drop-down next to Method.
Enter your static addressing information in the appropriate blanks of
the form, and make sure the interface is set to automatically start in
the General tab (I didn't screencap the General tab, so it might have a
different term in the actual dialog).
Hope that helps.
