Automated Kickstart install with modified ISO (add specific rpm's)

Latest response

What we are trying to do is to setup a DVD that we can send to remote locations to upgrade the OS and an application. The intent is to have minimal requirements for personnel on-site. Put in dvd and boot to dvd. That really is all we want them doing.

So far i have a RHEL 6.5 iso that i've added a kickstart file to the root level and it works fine. Now i'm trying to get the extra RPM's that are needed to support the application to install as part of the process. I was hoping to add a software group that was made up of the required rpm's and call for that to be installed from the kickstart. Mostly i was trying to avoid doing things through post install scripts but if i have to do it that way i will.

Can anyone help me get this setup.

RO
Log in to join the conversation

Responses

james.radtke@siriusxm.com's picture Guru 6863 points
24 June 2014 6:51 PM

I bet Remmele will have a good answer for this as I know he does a lot of "disconnected Satellite" type stuff. (I.. do not ;-)

Initially I am thinking that you should create a separate repo on the ISO itself, create a *.repo file with the repo disabled. That way if you wanted to ship out a DVD with ONLY this application stack, you could. Then they would simply use yum --enablerepo=DVD-APP.

Great question!

RO Community Member 59 points
24 June 2014 7:02 PM

Thanks. I actually have a repo in the root dir of the ISO. I just counldn't figure/find/see a way to incorporate it into the installs. I figured it should be possible with the way they are now doing the Install software groups. With them having a directory off of root and the repo info under it. I really don't want onsite personnel to do ANYTHING if at all possible. Our admins do a lot of work remotely so the on-sites do minimal.

R. Hinton's picture Guru 6827 points
24 June 2014 9:16 PM

i've been thinking about this one, and we also wanted to do the same... and had some thoughts that I wanted to interrogate/test... my initial idea without testing yet, create a custom repo (adding the rpms to the location where there is an install iso dump and create repo, mkisofs, and then verify sanity of deps etc, and test, and try... with a virtual system... I have yet another customer where I have yet to build my 7th satellite server (awaiting some things like storage), but thinking also of such an iso file...

thomas.jones2@dodiis.mil's picture Guru 6435 points
24 June 2014 9:59 PM

Having created a repo structure from your RPMs and pointed your DVD's ks file to that RPM (via repo directive), are you not able to at least individually enumerate the RPMs in your %packages section?

R. Hinton's picture Guru 6827 points
25 June 2014 1:31 AM

updated, Richard, see links in my subsequent post

Tom, Yes, we do that (we always put the desired rpms in %packages area) already, but we do this on the satellite server, we present our 3rd party channels that have the benefit of a repomd.xml in a repository channel that was generated via the satellite server. But I am curious if a DVD whose original repomd.xml doesn't know of the non-Red-Hat rpms might take exception to the additional non-redhat packages.

For an DVD and no satellite server, the cited rpms in the %packages area, I am curious (see link below) if it would benefit to be able to have the additional rpms within the 'createrepo' with perhaps "createrepo --update" on the custom DVD since (it seems from what Richard mentions), these will be stand-alone with no available satellite server? Here is another example from 2009, but again, untested, initial thoughts

Would be interested in your thoughts... I haven't had time to try this at home, and with what James mentioned I only put out some untested initial thoughts. I plan on trying something sometime soon, but haven't had a chance yet.

Kind Regards,

R. Hinton's picture Guru 6827 points
25 June 2014 2:01 AM

Richard,

Haven't had time to test, but examine these links:
- this bit, along with the link in my previous reply to Tom Jones might be an initial starting place to test.
- This one from 2012, rhel 6.x seems detailed well Part 1 of 4
- This is Part 2 of 4
- This is Part 3 of 4
- This is Part 4 of 4

Examine those links in context. It seems those examples likely have what you are seeking

Kind Regards,

RO Community Member 59 points
26 June 2014 12:35 PM

I'm working through these now. Sorry for taking so long to get back to you.

R. Hinton's picture Guru 6827 points
26 June 2014 1:19 PM

Let us know how it goes Richard...

RO Community Member 59 points
30 June 2014 3:26 PM

Ok, i've run throught the building a custom kickstart pages linked above. I still haven't gotten the correct packages for all dependencies but I'll get them worked out. THE PROBLEM i have is that it comes back with a bootloader error and states the system may not boot. It doesn't.

ks.cfg:
text
install
cdrom

key --skip
autostep --autoscreenshot
lang en_US.UTF-8
keyboard us

(eth0 config data removed)

bootloader --location=mbr --driveorder=sda --append="crashkernel=auto rhgb quiet"
firewall --service=ssh
authconfig --enableshadow --passalgo=sha512 --enablefingerprint
selinux --disabled
timezone --utc America/New_York

zerombr
clearpart --all --drives=sda

part /boot --fstype=ext4 --size=500
part pv.008002 --grow --size=1

volgroup vg_uemslavedev --pesize=4096 pv.008002
logvol /home --fstype=ext4 --name=lv_home --vgname=vg_uemslavedev --size=14336
logvol / --fstype=ext4 --name=lv_root --vgname=vg_uemslavedev --size=20168
logvol swap --name=lv_swap --vgname=vg_uemslavedev --size=5000
logvol /var --fstype=ext4 --name=lv_var --vgname=vg_uemslavedev --size=10000
logvol /var/audit --fstype=ext4 --name=lv_varaudit --vgname=vg_uemslavedev --size=10000
logvol /var/log --fstype=ext4 --name=lv_varlog --vgname=vg_uemslavedev --size=10000

james.radtke@siriusxm.com's picture Guru 6863 points
30 June 2014 3:33 PM

Have you tried adding --initlabel?
I have the following:

zerombr
bootloader --location=mbr --driveorder=sda,sdb --append="rhgb quiet crashkernel=auto"
clearpart --all --initlabel
RO Community Member 59 points
30 June 2014 5:09 PM

I have now, no difference. after the error message and reboot i see:

Booting from Hard Disk (then nothing, it freezes)

RO Community Member 59 points
30 June 2014 6:30 PM

I do appreciate the help. Though it looks like i have the dependencies all lined up I'm just beating my head agains the wall with the bootloader error. I'm not even sure how to find the problem with the bootloader. I've tried to build the Dell 1950 (versus a vm) so i can chroot /mnt/sysimage to look around. Also none of the %post items have been done. I have it set to create a couple repo's based on our local config but they aren't being made.

james.radtke@siriusxm.com's picture Guru 6863 points
30 June 2014 6:40 PM

Hey - I have run in to a somewhat similiar issue in the past, albeit on COMPLETELY different type of hardware... When I was testing a few different configs (VMware ESX, RHEV and RHEL) on some PCs I had built as a Lab - I ran in to this same issue.

My resoution - booted to Knoppix (or some liveDVD) and completely wiped the drive. All zero's ;-) Wiped the MBR, ran dd on the first few hundred megs and then ran shred on the entire drive. Things magically worked after that ;-)

I assume the 1950 is a bit older and does not have UEFI - so, that likely is not messing with anything. I wonder if there is a BIOS setting to direct the BIOS where to look for the boot block? I would try to completely wipe the drive (like I mentioned) and try again.

R. Hinton's picture Guru 6827 points
30 June 2014 8:25 PM

Richard, I've had to do as James described in his last post, and it helped. Did you create the raid array with the RAID BIOS? Dell has a RAID bios and until you create it (or even wipe it on occasion), the disk will not present itself.

RO Community Member 59 points
1 July 2014 11:05 AM

I understand what you are saying for the actual hardware version, but i'm getting this error even on my VM where i blow away and recreate the virtual drive.

I'm using the command line from the suggested reading. could this be part of the problem:
mkisofs -o custom.iso -b isolinux.bin -c boot.cat -no-emul-boot -boot-load-size 4 -boot-info-table -R -J -v -T TARGETDIR

R. Hinton's picture Guru 6827 points
1 July 2014 11:56 AM

I think I've used something along these lines: (remember the ending "." character)

mkisofs -o ../bootcd.iso -b isolinux/isolinux.bin -c isolinux/boot.cat -no-emul-boot -boot-load-size 4 -boot-info-table -J -r .

Hope this helps...

RO Community Member 59 points
1 August 2014 6:55 PM

Well it's been a while and i had it all working. So i moved to a different part of the project and another admin rebuilt the system i was building on. Of course he did not back it up. So i'm going through again and working on it with the copies i had backed up but i'm missing something. It is poping out the CDROM and saying no install media. I know i've seen this before but can't find it now.

The way i got it to add the packages i wanted to add was them to the core group as required.

Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.