From kanarip at kanarip.com Wed Aug 1 01:07:55 2007 From: kanarip at kanarip.com (Jeroen Van Meeuwen) Date: Tue, 31 Jul 2007 18:07:55 -0700 Subject: how to exclude package from repo Message-ID: <0JM100MNZWSV7N00@jes-fe2.zx.nl> That doesn't really answer the question because one would want 'kernel' to be installed, but from repo 'centos' or 'fedora', but not 'centosplus' or 'repoX', while the package lives in both repos. Being able to specify package NEVRA would solve that, but for the time being, one needs to create his own repository or use smarter composing software that does allow package NEVRA in kickstart or allows you to select specific packages from, say, a dialog. /me, from his cellphone -----Oorspronkelijk bericht ----- Van: "Thomas Callahan" Aan: "Discussion list about Kickstart" Verzonden: 07-07-31 08:29 Onderwerp: Re: how to exclude package from repo Put a minus in front of the package name under %packages in your ks file. That excludes that package from being installed. Ie -kde -firefox X11 wireshark Thanks, Tom Callahan On 7/31/07 11:13 AM, "Farkas Levente" wrote: > hi, > is there any way to exclude package or assign priority to repository in > kickstart file? with yum one can exclude packages from a repo or with > yum-protect or yum-priority i can assign priority among repos (in conf > file). who can i do it in a kickstart file? eg. i'd like to install > kmod-xfs from centosplus repo during kickstart, but unfortunately > centosplus repo also contains a modified kernel while i wanna use the > base kernel. so with yum i can protect or give higher priority to base > and updates and in this case kernel not updated from centos plus, but > kmod-xfs can be used. of if i can exclude=kernel* from centosplus that's > another solution. but none of the above work with kickstart. > is there any way to solve my problem? > thanks in advance. > > -- > Levente "Si vis pacem para bellum!" > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > From dxh at yahoo.com Wed Aug 1 19:59:30 2007 From: dxh at yahoo.com (Don Hoover) Date: Wed, 1 Aug 2007 12:59:30 -0700 (PDT) Subject: Specifing 32bit/i386 packages in ks file for 64bit install Message-ID: <351404.10530.qm@web60725.mail.yahoo.com> Ok...Could someone show me an example of specifying kickstart to install the 32 bit version of something in addition to the 64? For example, I would like to have the 32bit version of ncurses installed on my 64bit kickstarts. With yum is a simple matter of "yum install ncurses.i386", but I am not sure that ncurses.i386 will work in the %packages section. Thanks! From jkeating at j2solutions.net Wed Aug 1 20:03:13 2007 From: jkeating at j2solutions.net (Jesse Keating) Date: Wed, 1 Aug 2007 16:03:13 -0400 Subject: Specifing 32bit/i386 packages in ks file for 64bit install In-Reply-To: <351404.10530.qm@web60725.mail.yahoo.com> References: <351404.10530.qm@web60725.mail.yahoo.com> Message-ID: <20070801160313.6b304d52@localhost.localdomain> On Wed, 1 Aug 2007 12:59:30 -0700 (PDT) Don Hoover wrote: > With yum is a simple matter of "yum install > ncurses.i386", but I am not sure that ncurses.i386 > will work in the %packages section. Have you tried it? It should just work. -- Jesse Keating RHCE (jkeating.livejournal.com) Fedora Project (fedoraproject.org/wiki/JesseKeating) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From clintonlee.taylor at gmail.com Sun Aug 5 19:29:02 2007 From: clintonlee.taylor at gmail.com (Clinton Lee Taylor) Date: Sun, 5 Aug 2007 21:29:02 +0200 Subject: RAID option with start as failed ... Message-ID: Greetings ... This is my first post to this list, so please bear with me, thanks. I use kickstart to do server install, which I'm sure quite a few other do the same. One thing I have not been able to find yet, is a way to do a kickstart install with RAID started with a failed device. Basically, I'm trying to start a server install with one system drive, which is half of a mirror RAID-1 install. I do this, so that I can remove one hard drive of the last installed system, install Fedora, backup, copy or what ever is needed on the new install until I'm happy with it, and then put the second device back in and re-sync the two mirror drives. Is there a way to do this with a kickstart script option or something? Thanks Mailed LeeT -------------- next part -------------- An HTML attachment was scrubbed... URL: From dcutler at intelimedix.com Tue Aug 7 17:08:12 2007 From: dcutler at intelimedix.com (Dan Cutler) Date: Tue, 7 Aug 2007 13:08:12 -0400 Subject: Anaconda Exception: parserError: xmlParseMemory() failed Message-ID: <01CBA87D920178499B0DC51A4A735FD7FA0A36@MAIL.dsm.net> Fellow systems builders, I have been happily PXE booting Linux boxes with custom package lists via kickstart and a little cfengine. I was doing what I thought was a typical cycle: Added a few custom groups to ks.cfg, create rpm?s copy rpm?s to my netboot RPMS directory edit comps.xml - add rpms to my custom groups run genhdlist ( /usr/lib/anaconda-runtime/genhdlist --withnumbers --hdlist /i386/linux4.4/RedHat/base/hdlist /i386/linux4.4) I have been doing this awhile without incident (and enjoying the benefits ;-) until my build system suddenly broke while parsing comps.xml (although there doesn?t appear to be anything wrong with my comps.xml). >From anacdump.txt? Traceback (most recent call last): File "/usr/bin/anaconda", line 1181, in ? intf.run(id, dispatch, configFileData) ?snip? File "/usr/lib/python2.3/site-packages/libxml2.py", line 1175, in parseMemory if ret is None:raise parserError('xmlParseMemory() failed') parserError: xmlParseMemory() failed I?m seeing that folks are saying the ?genhdlist? is outdated. Maybe I should be generating hdlist in another way? What is the best way to generate hdlist*? How can I tell if my comps.xml is not matching up with my RPM repo? Thanks! --Dan Cutler No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.476 / Virus Database: 269.11.8/940 - Release Date: 8/6/2007 4:53 PM -------------- next part -------------- An HTML attachment was scrubbed... URL: From clumens at redhat.com Wed Aug 8 16:51:41 2007 From: clumens at redhat.com (Chris Lumens) Date: Wed, 8 Aug 2007 12:51:41 -0400 Subject: progress from kickstart scripts Message-ID: <20070808165141.GO6172@exeter.boston.redhat.com> You may be interested to know that the long-standing bug of being able to indicate progress in a kickstart %post script is now in MODIFIED. This bug is: http://bugzilla.redhat.com/147109 What I've done is add the zenity program to the stage2 image. This is sort of like dialog - it's a small program that you provide options to and it displays a GTK widget corresponding to those options. The return value or output of the program contains whatever the user did with the widget. This should give you the capability to display progress bars for especially large or slow %pre and %post scripts. The man page lists all the options you've got to work with. Problems: - It's kind of ugly. I don't think I care all that much. - I've only tested with %pre scripts, because rawhide is having problems today. - It only works for graphical installs, because it's a graphical program. - I don't think this will work in chrooted %post scripts. You may want to experiment with the $DISPLAY variable and all that to see if you can trick it into working. - This gives you guys all sorts of other widgets to use in your scripts, even allowing for interactive forms and the like. This should lead to even more complicated and bizarre kickstart files. Please try this stuff out in the next build of anaconda and let me know if there are any problems. Enjoy. - Chris From dcutler at intelimedix.com Wed Aug 8 18:26:09 2007 From: dcutler at intelimedix.com (Dan Cutler) Date: Wed, 8 Aug 2007 14:26:09 -0400 Subject: Anaconda Exception: parserError: xmlParseMemory() failed In-Reply-To: <01CBA87D920178499B0DC51A4A735FD7FA0A36@MAIL.dsm.net> Message-ID: <01CBA87D920178499B0DC51A4A735FD7FA0BAC@MAIL.dsm.net> Glory be! I?ve found my own answer? HYPERLINK "http://rhlinux.redhat.com/anaconda/comps.html"http://rhlinux.redhat.com/anaconda/comps.html This handy python script showed where my typo was in my comps.xml file. genhd is working once again! The world has been saved ;-D #!/usr/bin/python import rhpl.comps import sys comps = rhpl.comps.Comps(sys.argv[1]) for group in comps.groups.values(): pkgs = [] for (type, pkg) in group.packages.values(): if type == u'mandatory': pkgs.append(pkg) print group.name, pkgs --Dan _____ From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Dan Cutler Sent: Tuesday, August 07, 2007 1:08 PM To: kickstart-list at redhat.com Subject: Anaconda Exception: parserError: xmlParseMemory() failed Fellow systems builders, I have been happily PXE booting Linux boxes with custom package lists via kickstart and a little cfengine. I was doing what I thought was a typical cycle: Added a few custom groups to ks.cfg, create rpm?s copy rpm?s to my netboot RPMS directory edit comps.xml - add rpms to my custom groups run genhdlist ( /usr/lib/anaconda-runtime/genhdlist --withnumbers --hdlist /i386/linux4.4/RedHat/base/hdlist /i386/linux4.4) I have been doing this awhile without incident (and enjoying the benefits ;-) until my build system suddenly broke while parsing comps.xml (although there doesn?t appear to be anything wrong with my comps.xml). >From anacdump.txt? Traceback (most recent call last): File "/usr/bin/anaconda", line 1181, in ? intf.run(id, dispatch, configFileData) ?snip? File "/usr/lib/python2.3/site-packages/libxml2.py", line 1175, in parseMemory if ret is None:raise parserError('xmlParseMemory() failed') parserError: xmlParseMemory() failed I?m seeing that folks are saying the ?genhdlist? is outdated. Maybe I should be generating hdlist in another way? What is the best way to generate hdlist*? How can I tell if my comps.xml is not matching up with my RPM repo? Thanks! --Dan Cutler No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.476 / Virus Database: 269.11.8/940 - Release Date: 8/6/2007 4:53 PM No virus found in this incoming message. Checked by AVG Free Edition. Version: 7.5.476 / Virus Database: 269.11.8/940 - Release Date: 8/6/2007 4:53 PM No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.476 / Virus Database: 269.11.8/940 - Release Date: 8/6/2007 4:53 PM -------------- next part -------------- An HTML attachment was scrubbed... URL: From herrold at owlriver.com Wed Aug 8 19:46:35 2007 From: herrold at owlriver.com (R P Herrold) Date: Wed, 8 Aug 2007 15:46:35 -0400 (EDT) Subject: progress from kickstart scripts In-Reply-To: <20070808165141.GO6172@exeter.boston.redhat.com> References: <20070808165141.GO6172@exeter.boston.redhat.com> Message-ID: On Wed, 8 Aug 2007, Chris Lumens wrote: > What I've done is add the zenity program to the stage2 image. I 'get it' from the balance of your message that you are not too keen with zenity; Out of curiousity why not gdialog, which has the dialog analogue, so ks/anaconda can handle both GUI and TUI progress stuff with similar code? Adding 'Yet Another Widget' set to use seems looks a loss. -- Russ Herrold From richzendy at gmail.com Sat Aug 11 05:29:26 2007 From: richzendy at gmail.com (Edwind Richzendy Contreras Soto) Date: Sat, 11 Aug 2007 01:29:26 -0400 Subject: Installation from kickstart don't eject cdrom Message-ID: <90ba020d0708102229i3fd247d5g949107b59de981a8@mail.gmail.com> Hi all. My name is Edwind Contreras and i wrote a kickstart file for install somethings fedora 7 clients ( IBM lenovo model 9645-G1S [1]), but after the install all packets and run %post scripts the cdrom tray don't eject, just reboot, somebody can see my kickstart file and suggest something about this? My kickstart file is this: #platform=AMD64 o Intel EM64T # System authorization information auth --useshadow --enablemd5 --enableldap --enableldapauth --ldapserver= 192.168.0.254 --ldapbasedn=ou=People,dc=domain,dc=com,dc=ve # Use text mode install text # Firewall configuration firewall --disabled firstboot --disable # System keyboard keyboard la-latin1 # System language lang es_VE # Root password rootpw --iscrypted $1$asdasreyt8IX/s//iouoiS$z1hVloiuoi7wj7Kyd11ytytutyiyui5869K.y/ # SELinux configuration selinux --disabled # Install OS instead of upgrade install # Use CDROM installation media cdrom #network network --bootproto static --ip 192.168.0.10 --netmask 255.255.255.0--gateway 192.168.0.254 --nameserver 192.168.0.254 # Reboot after installation reboot # System timezone timezone --isUtc America/Caracas # X Window System configuration information xconfig --driver=intel --defaultdesktop=GNOME --depth=24 --resolution=1024x768 --videoram=256 # System bootloader configuration bootloader --append="vga=791" --location=mbr # Partition clearing information clearpart --linux --drives=sda # Disk partitioning information part /boot --fstype="ext3" --ondisk=sda --size=100 --bytes-per-inode=4096 part pv.4 --grow --ondisk=sda --size=1 --bytes-per-inode=4096 volgroup VolGroup00 --pesize=32768 pv.4 logvol swap --fstype="swap" --grow --maxsize=992 --size=496 --name=LogVol01 --vgname=VolGroup00 --bytes-per-inode=4096 logvol / --fstype="ext3" --grow --size=9000 --name=LogVol00 --vgname=VolGroup00 --bytes-per-inode=4096 %post chkconfig messagebus off chkconfig avahi-daemon off chkconfig avahi-dnsconfd off chkconfig haldaemon off chkconfig ConsoleKit off chkconfig yum-updatesd off chkconfig fistboot off chkconfig pcscd off echo "session required pam_mkhomedir.so skel=/etc/skel/ umask=0022" >> /etc/pam.d/gdm cat << EOF > /etc/rc.local #!/bin/sh # # This script will be executed *after* all the other init scripts. # You can put your own initialization stuff in here if you don't # want to do the full Sys V style init stuff. /sbin/netconfig touch /var/lock/subsys/local EOF echo '#!/bin/sh' >> /sbin/netconfig echo 'IP=`/sbin/ifconfig eth0 | grep inet | cut -d : -f 2 | cut -d " " -f 1`' >> /sbin/netconfig echo 'if [ $IP == 192.168.0.10 ] ; then' >> /sbin/netconfig echo '/usr/bin/system-config-network' >> /sbin/netconfig echo 'sed -i "s/"id:3:initdefault:"/"id:5:initdefault:"/g" /etc/inittab' >> /sbin/netconfig echo 'for i in {1..6} ; do sed -i "s/"$i:2345"/"#$i:2345"/" /etc/inittab ; done' >> /sbin/netconfig echo 'reboot' >> /sbin/netconfig echo 'else' >> /sbin/netconfig echo 'echo "IP Configurada adecuadamente"' >> /sbin/netconfig echo 'fi' >> /sbin/netconfig chmod 775 /sbin/netconfig cat << EOF > /sbin/profilesync #!/bin/sh rsync -r rsync://192.168.0.254/profiles /etc/desktop-profiles/ EOF chmod 775 /sbin/profilesync echo "*/10 * * * * /sbin/profilesync" >> /var/spool/cron/root chmod 600 /var/spool/cron/root /etc/init.d/crond restart cat << EOF > /etc/modprobe.conf alias eth0 tg3 alias scsi_hostadapter ata_piix alias snd-card-0 snd-hda-intel options snd-hda-intel position_fix=1 model=3stack EOF eject wait eject /tmp/cdrom wait eject /dev/cdrom wait %packages .................... something packets .................... Thanks, any help will be appreciated [1] http://www-307.ibm.com/pc/support/site.wss/document.do?sitestyle=lenovo&lndocid=MIGR-67159 -------------- next part -------------- An HTML attachment was scrubbed... URL: From clumens at redhat.com Mon Aug 13 13:57:00 2007 From: clumens at redhat.com (Chris Lumens) Date: Mon, 13 Aug 2007 09:57:00 -0400 Subject: Installation from kickstart don't eject cdrom In-Reply-To: <90ba020d0708102229i3fd247d5g949107b59de981a8@mail.gmail.com> References: <90ba020d0708102229i3fd247d5g949107b59de981a8@mail.gmail.com> Message-ID: <20070813135654.GR6172@exeter.boston.redhat.com> > # Reboot after installation > reboot Change to read: reboot --eject > %post > eject > wait > eject /tmp/cdrom > wait > eject /dev/cdrom > wait This won't work because when the %post scripts are run, we still have the CD/DVD mounted. - Chris From B.King at surrey.ac.uk Tue Aug 14 16:43:50 2007 From: B.King at surrey.ac.uk (Bevis R W King) Date: Tue, 14 Aug 2007 17:43:50 +0100 Subject: Problems with DHCP in F7 kickstart Message-ID: <1187109830.5578.325.camel@scooter.eps.surrey.ac.uk> I'm currently trying to get Fedora 7 (F7) to kickstart a PC. I've tried various kickstart files, but the one I've settled on is an anaconda-ks.cfg created by a manual workstation install via HTTP on that very machine. If it contains the line: network --device eth0 --bootproto=static --ip=131.227.95.104 --netmask=255.255.255.0 --gateway=131.227.95.1 --nameserver=131.227.76.3 --hostname=willow.eps.surrey.ac.uk it works. The workstation then installs perfectly. If it contains the line: network --device eth0 --noipv6 --bootproto dhcp --gateway 131.227.95.1 --nameserver=131.227.76.3 --netmask=0xffffff00 --onboot=yes It doesn't. Silly thing is after getting the initial boot menu, hitting , adding " ks" to the boot string and hitting return, the machine DHCP's for it's address successfully, gets the ks file and goes after the correct HTTP-based install repos. It then calls getNetConfig, I assume as a result of the above directives, sends on the the DHCPDISCOVER calls and after about five attempts, fails to get the DHCPv4 address it was given before. The syslog on the DHCP server indicates that a DHCPOFFER is being sent each time. The console 3 log says: INFO : DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 17 INFO : DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 17 INFO : DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 7 INFO : DHCPv4 eth0 - TIMED OUT. INFO : doing kickstart... setting it up ERROR : no DNS servers, can't look up hostname INFO : starting to STEP_URL and then fails to get stage2. The DHCP server in question worked fine for dynamic assignment in FC6, and has all the options configured for that - dns-servers, etc, etc. Any thoughts on what might be causing this? Regards, Bevis. From kanarip at kanarip.com Tue Aug 14 16:51:44 2007 From: kanarip at kanarip.com (Jeroen van Meeuwen) Date: Tue, 14 Aug 2007 18:51:44 +0200 Subject: Problems with DHCP in F7 kickstart In-Reply-To: <1187109830.5578.325.camel@scooter.eps.surrey.ac.uk> References: <1187109830.5578.325.camel@scooter.eps.surrey.ac.uk> Message-ID: <46C1DDA0.2060004@kanarip.com> Didn't F7 anaconda have a bug which made it not really disable IPv6 all to well? I remember encountering it a while ago and having to specify local media rather then network media. Kind regards, Jeroen van Meeuwen -kanarip Bevis R W King wrote: > I'm currently trying to get Fedora 7 (F7) to kickstart a PC. I've tried > various kickstart files, but the one I've settled on is an > anaconda-ks.cfg created by a manual workstation install via HTTP on that > very machine. > From kernel at linuxace.com Tue Aug 14 17:26:52 2007 From: kernel at linuxace.com (Phil Oester) Date: Tue, 14 Aug 2007 10:26:52 -0700 Subject: Problems with DHCP in F7 kickstart In-Reply-To: <1187109830.5578.325.camel@scooter.eps.surrey.ac.uk> References: <1187109830.5578.325.camel@scooter.eps.surrey.ac.uk> Message-ID: <20070814172652.GA23600@linuxace.com> On Tue, Aug 14, 2007 at 05:43:50PM +0100, Bevis R W King wrote: > I'm currently trying to get Fedora 7 (F7) to kickstart a PC. I've tried > various kickstart files, but the one I've settled on is an > anaconda-ks.cfg created by a manual workstation install via HTTP on that > very machine. > > If it contains the line: > network --device eth0 --bootproto=static --ip=131.227.95.104 > --netmask=255.255.255.0 --gateway=131.227.95.1 --nameserver=131.227.76.3 > --hostname=willow.eps.surrey.ac.uk > > it works. The workstation then installs perfectly. > > If it contains the line: > network --device eth0 --noipv6 --bootproto dhcp --gateway 131.227.95.1 > --nameserver=131.227.76.3 --netmask=0xffffff00 --onboot=yes > > It doesn't. Did you check bugzilla? https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=241783 Phil From richzendy at gmail.com Wed Aug 15 04:06:31 2007 From: richzendy at gmail.com (Edwind Richzendy Contreras Soto) Date: Wed, 15 Aug 2007 00:06:31 -0400 Subject: Installation from kickstart don't eject cdrom In-Reply-To: <20070813135654.GR6172@exeter.boston.redhat.com> References: <90ba020d0708102229i3fd247d5g949107b59de981a8@mail.gmail.com> <20070813135654.GR6172@exeter.boston.redhat.com> Message-ID: <90ba020d0708142106h49f43893y7aba70fd008c2fb3@mail.gmail.com> 2007/8/13, Chris Lumens : > > > # Reboot after installation > > reboot > > Change to read: > > reboot --eject > > > %post > > eject > > wait > > eject /tmp/cdrom > > wait > > eject /dev/cdrom > > wait > > This won't work because when the %post scripts are run, we still have > the CD/DVD mounted. Thanks Chris, reboot --eject really work - Chris > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From richzendy at gmail.com Wed Aug 15 04:06:31 2007 From: richzendy at gmail.com (Edwind Richzendy Contreras Soto) Date: Wed, 15 Aug 2007 00:06:31 -0400 Subject: Installation from kickstart don't eject cdrom In-Reply-To: <20070813135654.GR6172@exeter.boston.redhat.com> References: <90ba020d0708102229i3fd247d5g949107b59de981a8@mail.gmail.com> <20070813135654.GR6172@exeter.boston.redhat.com> Message-ID: <90ba020d0708142106h49f43893y7aba70fd008c2fb3@mail.gmail.com> 2007/8/13, Chris Lumens : > > > # Reboot after installation > > reboot > > Change to read: > > reboot --eject > > > %post > > eject > > wait > > eject /tmp/cdrom > > wait > > eject /dev/cdrom > > wait > > This won't work because when the %post scripts are run, we still have > the CD/DVD mounted. Thanks Chris, reboot --eject really work - Chris > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From ggerard at ggerard.com Thu Aug 16 01:59:20 2007 From: ggerard at ggerard.com (Gregory Gerard) Date: Wed, 15 Aug 2007 18:59:20 -0700 Subject: Why does the anaconda-ks.cfg file *not* reflect the packages I chose during my manual install? Message-ID: <46C3AF78.8050704@ggerard.com> Hello, Longtime listener, first time caller... Just installed CentOS 5 and along the way I selected all optional packages for things. 1. It's incredibly annoying I cannot right-click on a top-level category and say "Install all optional packages" but must through each sub-category and tediously do it. 2. I was hoping the ks file generated would have a good flight recording I could then replay but it only reflects the default packages, not the "all optional packages" I'd selected during the manual install. 3. While CentOS 5 configurator doesn't crash like the FC6 did (wow, that sucked a lot of time), it doesn't allow me to select optional packages at all making it pretty much a starting point for a boilerplate KS file that I must then extensively hand edit. 4. In my naive attempt at getting this behavior to work, I did a "rpm -qa >> mymachine.cfg" in order to have anaconda install the RPMs that were installed when I did the original manual install (sorry, English grammar is elusive right now). This just resulted in the install just sitting there pegging the CPU with "Completed Completed" being the last message on the console. Ideas? thanks, greg From jos at xos.nl Thu Aug 16 07:40:15 2007 From: jos at xos.nl (Jos Vos) Date: Thu, 16 Aug 2007 09:40:15 +0200 Subject: Why does the anaconda-ks.cfg file *not* reflect the packages I chose during my manual install? In-Reply-To: <46C3AF78.8050704@ggerard.com>; from ggerard@ggerard.com on Wed, Aug 15, 2007 at 06:59:20PM -0700 References: <46C3AF78.8050704@ggerard.com> Message-ID: <20070816094015.A9871@xos037.xos.nl> On Wed, Aug 15, 2007 at 06:59:20PM -0700, Gregory Gerard wrote: > 4. In my naive attempt at getting this behavior to work, I did a "rpm > -qa >> mymachine.cfg" in order to have anaconda install the RPMs that > were installed when I did the original manual install (sorry, English > grammar is elusive right now). This just resulted in the install just > sitting there pegging the CPU with "Completed Completed" being the last > message on the console. I think kickstart expect package names, not n-v-r lines, so maybe you should try to put the output of "rpm -qa --qf '%{name}\n'" in the ks file. -- -- Jos Vos -- X/OS Experts in Open Systems BV | Phone: +31 20 6938364 -- Amsterdam, The Netherlands | Fax: +31 20 6948204 From sankarshan.mukhopadhyay at gmail.com Thu Aug 16 10:55:26 2007 From: sankarshan.mukhopadhyay at gmail.com (Sankarshan Mukhopadhyay) Date: Thu, 16 Aug 2007 16:25:26 +0530 Subject: Why does the anaconda-ks.cfg file *not* reflect the packages I chose during my manual install? In-Reply-To: <46C3AF78.8050704@ggerard.com> References: <46C3AF78.8050704@ggerard.com> Message-ID: <46C42D1E.1070008@gmail.com> -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Gregory Gerard wrote: > 1. It's incredibly annoying I cannot right-click on a top-level category > and say "Install all optional packages" but must through each > sub-category and tediously do it. Is this somewhat related to the fact that you are missing an "Install Everything" kind of installation option that users were used to ? > 2. I was hoping the ks file generated would have a good flight recording > I could then replay but it only reflects the default packages, not the > "all optional packages" I'd selected during the manual install. Isn't that how it has been all along ? > 3. While CentOS 5 configurator doesn't crash like the FC6 did (wow, that > sucked a lot of time), it doesn't allow me to select optional packages > at all making it pretty much a starting point for a boilerplate KS file > that I must then extensively hand edit. I kind of lost you here, do you mean to say system-config-kickstart does not seem to work as expected ? > 4. In my naive attempt at getting this behavior to work, I did a "rpm > -qa >> mymachine.cfg" in order to have anaconda install the RPMs that > were installed when I did the original manual install (sorry, English > grammar is elusive right now). This just resulted in the install just > sitting there pegging the CPU with "Completed Completed" being the last > message on the console. yum list (with the grep for installed) on the original machine and then passing that output through a yum install ? :) - -- You see things; and you say 'Why?'; But I dream things that never were; and I say 'Why not?' - George Bernard Shaw -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.7 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org iD8DBQFGxC0eXQZpNTcrCzMRAh2GAJ9fwi2wW9G+SvAoqUEnVFYp5u9qLACeJAg+ PaWm7Z0wYC5fHEKzBqxhvWE= =itGt -----END PGP SIGNATURE----- From clumens at redhat.com Thu Aug 16 14:31:59 2007 From: clumens at redhat.com (Chris Lumens) Date: Thu, 16 Aug 2007 10:31:59 -0400 Subject: Why does the anaconda-ks.cfg file *not* reflect the packages I chose during my manual install? In-Reply-To: <46C3AF78.8050704@ggerard.com> References: <46C3AF78.8050704@ggerard.com> Message-ID: <20070816143159.GU6172@exeter.boston.redhat.com> > 2. I was hoping the ks file generated would have a good flight recording > I could then replay but it only reflects the default packages, not the > "all optional packages" I'd selected during the manual install. Because that's a bug in 5.0. This should be much improved in 5.1. > 3. While CentOS 5 configurator doesn't crash like the FC6 did (wow, that > sucked a lot of time), it doesn't allow me to select optional packages > at all making it pretty much a starting point for a boilerplate KS file > that I must then extensively hand edit. That's unfortunately a limitation of the system-config-kickstart in RHEL5. While it's consistent with the behavior in earlier releases, it's still pretty crappy. This is fixed in the s-c-ks in F7 and later (see bug #232285, version 2.7.4-1). This is not fixed in 5.1, and most likely won't be in any update release of 5. I haven't even gotten any escalated bugs on this yet. > 4. In my naive attempt at getting this behavior to work, I did a "rpm > -qa >> mymachine.cfg" in order to have anaconda install the RPMs that > were installed when I did the original manual install (sorry, English > grammar is elusive right now). This just resulted in the install just > sitting there pegging the CPU with "Completed Completed" being the last > message on the console. Large %packages sections take a very long time to process due to the code path that anaconda+yum are using. This should also be fixed in F7 and later. I do not know the status of this in 5.1, though I lean towards it not being better there. - Chris From michael at kmaclub.com Fri Aug 17 03:34:09 2007 From: michael at kmaclub.com (Michael) Date: Thu, 16 Aug 2007 20:34:09 -0700 Subject: Adding update repos to ks.cfg Message-ID: <46C51731.3080800@kmaclub.com> I have been using kickstart for a while to do my builds. I follow up with yum on the first boot to bring everything up to date. Today, I started to ask myself why can't I point the installer to the install DVD image and updates repo at the same time and install the newest packages without having to do an update. It seems to work, but I am concerned this might be risky somehow. I added something like: repo --name localupdates --baseurl=http://my.webserver.com/updates in my ks.cfg Is this an ok or safe thing to do? I just hate installing 300 packages and then having to update 200 because of new package release. I know I could update my install tree to include the new packages, but if there is any easier way... Thoughts? Michael From kernel at linuxace.com Fri Aug 17 04:53:26 2007 From: kernel at linuxace.com (Phil Oester) Date: Thu, 16 Aug 2007 21:53:26 -0700 Subject: Adding update repos to ks.cfg In-Reply-To: <46C51731.3080800@kmaclub.com> References: <46C51731.3080800@kmaclub.com> Message-ID: <20070817045326.GA1260@linuxace.com> On Thu, Aug 16, 2007 at 08:34:09PM -0700, Michael wrote: > I have been using kickstart for a while to do my builds. I follow up > with yum on the first boot to bring everything up to date. > > Today, I started to ask myself why can't I point the installer to the > install DVD image and updates repo at the same time and install the > newest packages without having to do an update. It seems to work, but I > am concerned this might be risky somehow. Nope, it works fine, and is a real timesaver. No problems. Phil From thdxo at hotmail.com Fri Aug 17 08:53:38 2007 From: thdxo at hotmail.com (=?ks_c_5601-1987?B?vNsgxcLIrw==?=) Date: Fri, 17 Aug 2007 17:53:38 +0900 Subject: I want to add knoppix (Live CD) !! Message-ID: Hi all I want to add knoppix(LiveCD) in lan booting Is this possibility ? If this is possibility, how I do ?? please, give me information.. thank you, bye _________________________________________________________________ ??? ??? MSN ?? ?????. http://www.msn.co.kr/shopping From isamar at gmail.com Fri Aug 17 10:12:13 2007 From: isamar at gmail.com (Isamar Maia) Date: Fri, 17 Aug 2007 19:12:13 +0900 Subject: Bug in FC7 Message-ID: Hi Folks, Using FC7, If I have SCSI hard disk and I set automatic disk partitioning with kickstart for the FC7's installation, it cleans the partition table right before formatting the disks. It happens with ext3, reiserfs, vfat and ext2. With IDE disk, it's ok. This problem was reproduced with a Buslogic BT-950 SCSI Board. To where should I report this bug ? -- Isamar Maia isamar at gmail.com imtech at tim.com.br Brazil: 55-71-9146-8575 55-71-9965-6119 ??: +81-(0)3-4550-1212 From rvandolson at esri.com Fri Aug 17 14:39:27 2007 From: rvandolson at esri.com (Ray Van Dolson) Date: Fri, 17 Aug 2007 07:39:27 -0700 Subject: Adding update repos to ks.cfg In-Reply-To: <46C51731.3080800@kmaclub.com> References: <46C51731.3080800@kmaclub.com> Message-ID: <20070817143926.GA19299@esri.com> On Thu, Aug 16, 2007 at 08:34:09PM -0700, Michael wrote: > I have been using kickstart for a while to do my builds. I follow up > with yum on the first boot to bring everything up to date. > > Today, I started to ask myself why can't I point the installer to the > install DVD image and updates repo at the same time and install the > newest packages without having to do an update. It seems to work, but I > am concerned this might be risky somehow. > > I added something like: > > repo --name localupdates --baseurl=http://my.webserver.com/updates > > in my ks.cfg > > Is this an ok or safe thing to do? I just hate installing 300 packages > and then having to update 200 because of new package release. > > I know I could update my install tree to include the new packages, but > if there is any easier way... I wasn't aware of this parameter. What distros have an anaconda that supports this? Ray From michael at kmaclub.com Fri Aug 17 14:57:10 2007 From: michael at kmaclub.com (Michael) Date: Fri, 17 Aug 2007 07:57:10 -0700 Subject: Adding update repos to ks.cfg In-Reply-To: <20070817143926.GA19299@esri.com> References: <46C51731.3080800@kmaclub.com> <20070817143926.GA19299@esri.com> Message-ID: <46C5B746.7080204@kmaclub.com> Ray Van Dolson wrote: > On Thu, Aug 16, 2007 at 08:34:09PM -0700, Michael wrote: > >> I have been using kickstart for a while to do my builds. I follow up >> with yum on the first boot to bring everything up to date. >> >> Today, I started to ask myself why can't I point the installer to the >> install DVD image and updates repo at the same time and install the >> newest packages without having to do an update. It seems to work, but I >> am concerned this might be risky somehow. >> >> I added something like: >> >> repo --name localupdates --baseurl=http://my.webserver.com/updates >> >> in my ks.cfg >> >> Is this an ok or safe thing to do? I just hate installing 300 packages >> and then having to update 200 because of new package release. >> >> I know I could update my install tree to include the new packages, but >> if there is any easier way... >> > > I wasn't aware of this parameter. What distros have an anaconda that > supports this? > > Ray > It works in Fedora Core 6 and Fedora 7 that I am aware of. Michael From jkeating at j2solutions.net Fri Aug 17 14:55:03 2007 From: jkeating at j2solutions.net (Jesse Keating) Date: Fri, 17 Aug 2007 10:55:03 -0400 Subject: Adding update repos to ks.cfg In-Reply-To: <46C5B746.7080204@kmaclub.com> References: <46C51731.3080800@kmaclub.com> <20070817143926.GA19299@esri.com> <46C5B746.7080204@kmaclub.com> Message-ID: <20070817105503.55f8f0ad@ender> On Fri, 17 Aug 2007 07:57:10 -0700 Michael wrote: > It works in Fedora Core 6 and Fedora 7 that I am aware of. RHEL5 as well. Nothing older. -- Jesse Keating RHCE (jkeating.livejournal.com) Fedora Project (fedoraproject.org/wiki/JesseKeating) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From tibbs at math.uh.edu Fri Aug 17 15:17:05 2007 From: tibbs at math.uh.edu (Jason L Tibbitts III) Date: 17 Aug 2007 10:17:05 -0500 Subject: Adding update repos to ks.cfg In-Reply-To: <46C51731.3080800@kmaclub.com> References: <46C51731.3080800@kmaclub.com> Message-ID: >>>>> "M" == Michael writes: M> repo --name localupdates --baseurl=http://my.webserver.com/updates I have heard reports that this doesn't work bvecause anaconda will get confused and install some packages out of order, but frankly I have no evidence to back that up. Still, I didn't want to have to worry about any bugs in that area, so what I'm doing now is generating a complete set of packages by calling pungi -G with a manifest containing the %packages section of my kickstart file, and then calling createrepo on the resulting package set. This installs fine and actually the install goes a bit faster because the package set is minimal. - J< From jkeating at j2solutions.net Fri Aug 17 15:29:43 2007 From: jkeating at j2solutions.net (Jesse Keating) Date: Fri, 17 Aug 2007 11:29:43 -0400 Subject: Adding update repos to ks.cfg In-Reply-To: References: <46C51731.3080800@kmaclub.com> Message-ID: <20070817112943.0bfcb23b@ender> On 17 Aug 2007 10:17:05 -0500 Jason L Tibbitts III wrote: > I have heard reports that this doesn't work bvecause anaconda will get > confused and install some packages out of order, but frankly I have no > evidence to back that up. You'll only experience those bugs if your main installation source is either CD/DVD physical media, or said isos via nfs. If you're using exploaded tree NFS, or http, or ftp you'll be fine. -- Jesse Keating RHCE (jkeating.livejournal.com) Fedora Project (fedoraproject.org/wiki/JesseKeating) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From tibbs at math.uh.edu Fri Aug 17 15:46:08 2007 From: tibbs at math.uh.edu (Jason L Tibbitts III) Date: 17 Aug 2007 10:46:08 -0500 Subject: Adding update repos to ks.cfg In-Reply-To: <20070817112943.0bfcb23b@ender> References: <46C51731.3080800@kmaclub.com> <20070817112943.0bfcb23b@ender> Message-ID: >>>>> "JK" == Jesse Keating writes: JK> If you're using exploaded tree NFS, or http, or ftp you'll be JK> fine. Blagh. Oh well, at least I know how to use pungi now. Actually the install is so much faster for me (since I kickstart a minimal system and pull in the rest of the packages when the machine boots) that I think I'll stick with the pungi method. - J< From jkeating at j2solutions.net Fri Aug 17 15:58:54 2007 From: jkeating at j2solutions.net (Jesse Keating) Date: Fri, 17 Aug 2007 11:58:54 -0400 Subject: Adding update repos to ks.cfg In-Reply-To: References: <46C51731.3080800@kmaclub.com> <20070817112943.0bfcb23b@ender> Message-ID: <20070817115854.6c45d1ab@ender> On 17 Aug 2007 10:46:08 -0500 Jason L Tibbitts III wrote: > Blagh. Oh well, at least I know how to use pungi now. Actually the > install is so much faster for me (since I kickstart a minimal system > and pull in the rest of the packages when the machine boots) that I > think I'll stick with the pungi method. I don't mind that at all (: The more users the merrier. -- Jesse Keating RHCE (jkeating.livejournal.com) Fedora Project (fedoraproject.org/wiki/JesseKeating) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From clintonlee.taylor at gmail.com Fri Aug 17 16:26:15 2007 From: clintonlee.taylor at gmail.com (Clinton Lee Taylor) Date: Fri, 17 Aug 2007 18:26:15 +0200 Subject: degraded raid with kickstart ... Message-ID: Greetings ... I have asked this before, but I might have worded it just a little wrong. I was hoping somebody could help me with doing a server install using kickstart file, which starts the system with a degraded raid. Basicly starting a server with only one hard drive, which will get the second drive once the server is ready to go into use. I currently use the following in my kickstart script ... Is there an option I don't know of or could use that I have missed ... bootloader --location=mbr --driveorder=sda,sdb clearpart --all --initlabel --drives=sda,sdb # /boot part raid.01 --size=256 --ondisk=sda --asprimary part raid.02 --size=256 --ondisk=sdb --asprimary # Dos part /mnt/dos-a --fstype="vfat" --size=256 --ondisk=sda --asprimary part /mnt/dos-b --fstype="vfat" --size=256 --ondisk=sdb --asprimary # Swap part swap --size=1024 --ondisk=sda --asprimary part swap --size=1024 --ondisk=sdb --asprimary # / part raid.11 --size=2048 --ondisk=sda part raid.12 --size=2048 --ondisk=sdb # /usr part raid.21 --size=4096 --ondisk=sda part raid.22 --size=4096 --ondisk=sdb # /var part raid.31 --size=2048 --ondisk=sda part raid.32 --size=2048 --ondisk=sdb # /home part raid.51 --size=256 --grow --ondisk=sda part raid.52 --size=256 --grow --ondisk=sdb #RAID raid /boot --fstype ext3 --level=RAID1 --device=md0 raid.01 raid.02 raid / --fstype ext3 --level=RAID1 --device=md1 raid.11 raid.12 raid /usr --fstype ext3 --level=RAID1 --device=md2 raid.21 raid.22 raid /var --fstype ext3 --level=RAID1 --device=md3 raid.31 raid.32 raid /home --fstype ext3 --level=RAID1 --device=md5 raid.51 raid.52 Thanks Mailed LeeT -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdehaan at redhat.com Fri Aug 17 20:35:34 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 17 Aug 2007 16:35:34 -0400 Subject: IRC channel Message-ID: <46C60696.6060106@redhat.com> A few folks in #cobbler on freenode were asking whether kickstart-list had an IRC channel where folks could help each other with kickstart-related questions/problems. I didn't think it did, so I made one. It's #kickstart on irc.freenode.net Everyone is welcome. If such a thing already exists, I'll get rid of it and everyone there can migrate. --Michael From hhoffman at ip-solutions.net Fri Aug 17 20:46:59 2007 From: hhoffman at ip-solutions.net (Harry Hoffman) Date: Fri, 17 Aug 2007 16:46:59 -0400 Subject: possible bug in kickstart with RHEL4 and %pre scripts using LVM Message-ID: <46C60943.7030006@ip-solutions.net> Hi All, Ran into what I believe is a bug. It looks to be in the %pre section. Background: Running CentOS release 5 with cobbler 0.5.2. I have a kickstart file for installing RHEL 4.5 (see below). I have a %pre section in the kickstart file that defines partitions: # Magically figure out how to partition this thing %include /tmp/partinfo %pre # Determine how many drives we have set $(list-harddrives) let numd=$#/2 d1=$1 # This is the device of disk 1 d2=$3 # This is the device of disk 2, etc. S1=$2 # This is the size of disk 1 S2=$4 # This is the size of disk 2, etc. # This would be a partition scheme for two or more drives if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary part swap --size=512 --ondisk=$d1 --asprimary part swap --size=512 --ondisk=$d2 --asprimary part raid.01 --size=1 --grow --ondisk=$d1 part raid.02 --size=1 --grow --ondisk=$d2 raid pv.01 --fstype "physical volume (LVM)" --level=RAID1 --device=md0 raid.01 raid.02 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow EOF else cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part swap --size=1024 --ondisk=$d1 --asprimary part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow #part / --fstype ext3 --size=1 --grow --ondisk=$d1 EOF fi This pre script shows the following errors on vt3: * Running kickstart %pre script(s) let: arith: syntax error: "numd=2/2" let: arith: syntax error: "numd=2/2" [: 2: unexpected operator * All kickstart %pre script(s) have been run * Exception parsing ks.cfg: Defined PV partition pv.01 multiple times In this particular case I have only one (logical HW RAID) drive and the /tmp/partinfo is created with the lines contained in the else clause from above. The error about partition pv.01 causes kickstart to fail and die :-( Please see the full kickstart file below. I should note that if I place the part info from the else clause outside of the %pre section (i.e. just create a static partition declaration in the kickstart file) then everything works as expected. Any ideas? I'm stumped. Cheers, Harry # Kickstart file to automate the installation of Oracle 10 on RHEL4 # Harry Hoffman # Oracle 10g Kickstart file for CentOS 4.x #platform=x86, AMD64, or Intel EM64T # System authorization information auth --useshadow --enablemd5 # System bootloader configuration bootloader --location=mbr --md5pass=blah # Partition clearing information clearpart --all --initlabel # Use text mode install text # Firewall configuration firewall --enabled --port=22:tcp # Run the Setup Agent on first boot firstboot --disable # System keyboard keyboard us # System language lang en_US.UTF-8 langsupport --default=en_US.UTF-8 en_US.UTF-8 # Use network installation url --url=http://192.168.1.3/cblr/links/RHEL-4.5-i386 # Network information network --bootproto=static --ip=192.168.1.10 --netmask=255.255.240.0 --gateway=192.168.1.249 --nameserver=192.168.1.3 --hostname=db1.prod.fqdn.com --device=eth0 --onboot=on # Reboot after installation reboot #Root password rootpw --iscrypted blah # SELinux configuration selinux --enforcing # Do not configure the X Window System skipx # System timezone timezone America/New_York # Install OS instead of upgrade install # Clear the Master Boot Record zerombr # Magically figure out how to partition this thing %include /tmp/partinfo %pre # Determine how many drives we have set $(list-harddrives) let numd=$#/2 d1=$1 # This is the device of disk 1 d2=$3 # This is the device of disk 2, etc. S1=$2 # This is the size of disk 1 S2=$4 # This is the size of disk 2, etc. # This would be a partition scheme for two or more drives if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary part swap --size=512 --ondisk=$d1 --asprimary part swap --size=512 --ondisk=$d2 --asprimary part raid.01 --size=1 --grow --ondisk=$d1 part raid.02 --size=1 --grow --ondisk=$d2 raid pv.01 --fstype "physical volume (LVM)" --level=RAID1 --device=md0 raid.01 raid.02 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow EOF else cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part swap --size=1024 --ondisk=$d1 --asprimary part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow #part / --fstype ext3 --size=1 --grow --ondisk=$d1 EOF fi %packages automake bison cpp flex gcc gcc-c++ glibc-devel glibc-headers glibc-kernheaders libungif Xaw3d xorg-x11-xauth libcap ntp sudo vlock screen sysstat -sendmail postfix vim-enhanced # Oracle 10 stuff compat-db openmotif setarch libaio libaio-devel compat-gcc-32 compat-gcc-32-c++ compat-libstdc++-33 %post chvt 3 #Turn off/on services chkconfig --level 2345 isdn off chkconfig --level 2345 portmap off chkconfig --level 2345 nfslock off chkconfig --level 345 ntpd on chkconfig --level 345 cups off #Setup root mail forwarding echo -e "root:\troot at fqdn.com" >> /etc/aliases postalias /etc/aliases # Setup Oracle groupadd -r oinstall; groupadd -r dba; useradd -r -c "Oracle DB User" -g oinstall -G dba -m -p 'blah' oracle echo "# " >> /etc/sysctl.conf echo "# Required for Oracle 10g " >> /etc/sysctl.conf echo "# " >> /etc/sysctl.conf echo "kernel.shmall = 2097152" >> /etc/sysctl.conf echo "kernel.shmmax = 2147483648" >> /etc/sysctl.conf echo "kernel.shmmni = 4096" >> /etc/sysctl.conf echo "kernel.sem = 250 32000 100 128" >> /etc/sysctl.conf echo "fs.file-max = 65536" >> /etc/sysctl.conf echo "net.ipv4.ip_local_port_range = 1024 65000" >> /etc/sysctl.conf echo "net.core.rmem_default=262144" >> /etc/sysctl.conf echo "net.core.rmem_max=262144" >> /etc/sysctl.conf echo "net.core.wmem_default=262144" >> /etc/sysctl.conf echo "net.core.wmem_max=262144" >> /etc/sysctl.conf /sbin/sysctl -p echo "oracle soft nproc 2047" >> /etc/security/limits.conf echo "oracle hard nproc 16384" >> /etc/security/limits.conf echo "oracle soft nofile 1024" >> /etc/security/limits.conf echo "oracle hard nofile 65536" >> /etc/security/limits.conf # setting /etc/profile echo "if [ \$USER = \"oracle\" ]; then" >> /etc/profile echo " if [ \$SHELL = \"/bin/ksh\" ]; then" >> /etc/profile echo " ulimit -p 16384" >> /etc/profile echo " ulimit -n 65536" >> /etc/profile echo " else" >> /etc/profile echo " ulimit -u 16384 -n 65536" >> /etc/profile echo " fi" >> /etc/profile echo "fi" >> /etc/profile # Change pam to respect the new limits echo "session required /lib/security/pam_limits.so" >> /etc/pam.d/login # Make oracle directories mkdir -p /opt/oracle/product/10.2.0.1/db_1 chown -R oracle:oinstall /opt/oracle/ chmod -R 775 /opt/oracle/ # Set ORACLE_HOME in oracle's .bashrc echo "ORACLE_HOME=/opt/oracle/product/10.2.0.1/db_1" >> /home/oracle/.bashrc echo "PATH=\$PATH:\$ORACLE_HOME/bin" >> /home/oracle/.bashrc echo "export \$PATH" >> /home/oracle/.bashrc touch /etc/oratab chown oracle:oinstall /etc/oratab wget -O /home/oracle/10201_database_linux32.zip http://192.168.1.3/cblr/localmirror/10201_database_linux32.zip wget -O /home/oracle/10gR2.rsp http://192.168.1.3/cblr/localmirror/10gR2.rsp chown oracle:oinstall /home/oracle/10201_database_linux32.zip cd /home/oracle && su oracle -c "unzip 10201_database_linux32.zip" wget -O /etc/rc3.d/S99oracle_install http://192.168.1.3/cblr/localmirror/S99oracle_install chmod 755 /etc/rc3.d/S99oracle_install # Update the system # Update the system /usr/sbin/rhnreg_ks --profilename "db1.prod.fqdn.com" --username "rhuser" --password "blah" --email "root at fqdn.com" up2date -fu up2date up2date -fu wget http://192.168.1.3/cblr/watcher.py?system_done=RHEL-4-Oracle -b From Chip.Shabazian at bankofamerica.com Fri Aug 17 20:59:58 2007 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Fri, 17 Aug 2007 13:59:58 -0700 Subject: possible bug in kickstart with RHEL4 and %pre scripts using LVM In-Reply-To: <46C60943.7030006@ip-solutions.net> Message-ID: My that looks familiar ;) What happens if you don't "calculate" the number of drives. In other words, change to this: let numd=1 This can help narrow down if that is really what the problem is. Chip -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Harry Hoffman Sent: Friday, August 17, 2007 1:47 PM To: Discussion list about Kickstart Subject: possible bug in kickstart with RHEL4 and %pre scripts using LVM Hi All, Ran into what I believe is a bug. It looks to be in the %pre section. Background: Running CentOS release 5 with cobbler 0.5.2. I have a kickstart file for installing RHEL 4.5 (see below). I have a %pre section in the kickstart file that defines partitions: # Magically figure out how to partition this thing %include /tmp/partinfo %pre # Determine how many drives we have set $(list-harddrives) let numd=$#/2 d1=$1 # This is the device of disk 1 d2=$3 # This is the device of disk 2, etc. S1=$2 # This is the size of disk 1 S2=$4 # This is the size of disk 2, etc. # This would be a partition scheme for two or more drives if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary part swap --size=512 --ondisk=$d1 --asprimary part swap --size=512 --ondisk=$d2 --asprimary part raid.01 --size=1 --grow --ondisk=$d1 part raid.02 --size=1 --grow --ondisk=$d2 raid pv.01 --fstype "physical volume (LVM)" --level=RAID1 --device=md0 raid.01 raid.02 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow EOF else cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part swap --size=1024 --ondisk=$d1 --asprimary part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow #part / --fstype ext3 --size=1 --grow --ondisk=$d1 EOF fi This pre script shows the following errors on vt3: * Running kickstart %pre script(s) let: arith: syntax error: "numd=2/2" let: arith: syntax error: "numd=2/2" [: 2: unexpected operator * All kickstart %pre script(s) have been run * Exception parsing ks.cfg: Defined PV partition pv.01 multiple times In this particular case I have only one (logical HW RAID) drive and the /tmp/partinfo is created with the lines contained in the else clause from above. The error about partition pv.01 causes kickstart to fail and die :-( Please see the full kickstart file below. I should note that if I place the part info from the else clause outside of the %pre section (i.e. just create a static partition declaration in the kickstart file) then everything works as expected. Any ideas? I'm stumped. Cheers, Harry # Kickstart file to automate the installation of Oracle 10 on RHEL4 # Harry Hoffman # Oracle 10g Kickstart file for CentOS 4.x #platform=x86, AMD64, or Intel EM64T # System authorization information auth --useshadow --enablemd5 # System bootloader configuration bootloader --location=mbr --md5pass=blah # Partition clearing information clearpart --all --initlabel # Use text mode install text # Firewall configuration firewall --enabled --port=22:tcp # Run the Setup Agent on first boot firstboot --disable # System keyboard keyboard us # System language lang en_US.UTF-8 langsupport --default=en_US.UTF-8 en_US.UTF-8 # Use network installation url --url=http://192.168.1.3/cblr/links/RHEL-4.5-i386 # Network information network --bootproto=static --ip=192.168.1.10 --netmask=255.255.240.0 --gateway=192.168.1.249 --nameserver=192.168.1.3 --hostname=db1.prod.fqdn.com --device=eth0 --onboot=on # Reboot after installation reboot #Root password rootpw --iscrypted blah # SELinux configuration selinux --enforcing # Do not configure the X Window System skipx # System timezone timezone America/New_York # Install OS instead of upgrade install # Clear the Master Boot Record zerombr # Magically figure out how to partition this thing %include /tmp/partinfo %pre # Determine how many drives we have set $(list-harddrives) let numd=$#/2 d1=$1 # This is the device of disk 1 d2=$3 # This is the device of disk 2, etc. S1=$2 # This is the size of disk 1 S2=$4 # This is the size of disk 2, etc. # This would be a partition scheme for two or more drives if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary part swap --size=512 --ondisk=$d1 --asprimary part swap --size=512 --ondisk=$d2 --asprimary part raid.01 --size=1 --grow --ondisk=$d1 part raid.02 --size=1 --grow --ondisk=$d2 raid pv.01 --fstype "physical volume (LVM)" --level=RAID1 --device=md0 raid.01 raid.02 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow EOF else cat << EOF >> /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part swap --size=1024 --ondisk=$d1 --asprimary part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 volgroup internal_hd pv.01 logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow #part / --fstype ext3 --size=1 --grow --ondisk=$d1 EOF fi %packages automake bison cpp flex gcc gcc-c++ glibc-devel glibc-headers glibc-kernheaders libungif Xaw3d xorg-x11-xauth libcap ntp sudo vlock screen sysstat -sendmail postfix vim-enhanced # Oracle 10 stuff compat-db openmotif setarch libaio libaio-devel compat-gcc-32 compat-gcc-32-c++ compat-libstdc++-33 %post chvt 3 #Turn off/on services chkconfig --level 2345 isdn off chkconfig --level 2345 portmap off chkconfig --level 2345 nfslock off chkconfig --level 345 ntpd on chkconfig --level 345 cups off #Setup root mail forwarding echo -e "root:\troot at fqdn.com" >> /etc/aliases postalias /etc/aliases # Setup Oracle groupadd -r oinstall; groupadd -r dba; useradd -r -c "Oracle DB User" -g oinstall -G dba -m -p 'blah' oracle echo "# " >> /etc/sysctl.conf echo "# Required for Oracle 10g " >> /etc/sysctl.conf echo "# " >> /etc/sysctl.conf echo "kernel.shmall = 2097152" >> /etc/sysctl.conf echo "kernel.shmmax = 2147483648" >> /etc/sysctl.conf echo "kernel.shmmni = 4096" >> /etc/sysctl.conf echo "kernel.sem = 250 32000 100 128" >> /etc/sysctl.conf echo "fs.file-max = 65536" >> /etc/sysctl.conf echo "net.ipv4.ip_local_port_range = 1024 65000" >> /etc/sysctl.conf echo "net.core.rmem_default=262144" >> /etc/sysctl.conf echo "net.core.rmem_max=262144" >> /etc/sysctl.conf echo "net.core.wmem_default=262144" >> /etc/sysctl.conf echo "net.core.wmem_max=262144" >> /etc/sysctl.conf /sbin/sysctl -p echo "oracle soft nproc 2047" >> /etc/security/limits.conf echo "oracle hard nproc 16384" >> /etc/security/limits.conf echo "oracle soft nofile 1024" >> /etc/security/limits.conf echo "oracle hard nofile 65536" >> /etc/security/limits.conf # setting /etc/profile echo "if [ \$USER = \"oracle\" ]; then" >> /etc/profile echo " if [ \$SHELL = \"/bin/ksh\" ]; then" >> /etc/profile echo " ulimit -p 16384" >> /etc/profile echo " ulimit -n 65536" >> /etc/profile echo " else" >> /etc/profile echo " ulimit -u 16384 -n 65536" >> /etc/profile echo " fi" >> /etc/profile echo "fi" >> /etc/profile # Change pam to respect the new limits echo "session required /lib/security/pam_limits.so" >> /etc/pam.d/login # Make oracle directories mkdir -p /opt/oracle/product/10.2.0.1/db_1 chown -R oracle:oinstall /opt/oracle/ chmod -R 775 /opt/oracle/ # Set ORACLE_HOME in oracle's .bashrc echo "ORACLE_HOME=/opt/oracle/product/10.2.0.1/db_1" >> /home/oracle/.bashrc echo "PATH=\$PATH:\$ORACLE_HOME/bin" >> /home/oracle/.bashrc echo "export \$PATH" >> /home/oracle/.bashrc touch /etc/oratab chown oracle:oinstall /etc/oratab wget -O /home/oracle/10201_database_linux32.zip http://192.168.1.3/cblr/localmirror/10201_database_linux32.zip wget -O /home/oracle/10gR2.rsp http://192.168.1.3/cblr/localmirror/10gR2.rsp chown oracle:oinstall /home/oracle/10201_database_linux32.zip cd /home/oracle && su oracle -c "unzip 10201_database_linux32.zip" wget -O /etc/rc3.d/S99oracle_install http://192.168.1.3/cblr/localmirror/S99oracle_install chmod 755 /etc/rc3.d/S99oracle_install # Update the system # Update the system /usr/sbin/rhnreg_ks --profilename "db1.prod.fqdn.com" --username "rhuser" --password "blah" --email "root at fqdn.com" up2date -fu up2date up2date -fu wget http://192.168.1.3/cblr/watcher.py?system_done=RHEL-4-Oracle -b _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From mdehaan at redhat.com Fri Aug 17 21:06:45 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 17 Aug 2007 17:06:45 -0400 Subject: Other stupid web tricks: kickstart-list search engine Message-ID: <46C60DE5.5050001@redhat.com> Google Custom Search, to save some typing. It just mines the archives. http://tinyurl.com/2usop9 --Michael DeHaan From martin.hooper at gmail.com Sat Aug 18 12:43:28 2007 From: martin.hooper at gmail.com (Martin Hooper) Date: Sat, 18 Aug 2007 13:43:28 +0100 Subject: Kickstart file to do everything automatically apart from partitioning Message-ID: <635a92990708180543h593d79a8tea6f5ac72f090b4f@mail.gmail.com> http://www.martinjh.myby.co.uk/ks.cfg Given the above KS file what would I need to change to make a Fedora 7 automatic install which only does manual partitioning? Also is that a package list that will install all of the required base libraries and all of kDE?? -- Martin Hooper http://www.martinjh.myby.co.uk From martin.hooper at gmail.com Sat Aug 18 14:29:53 2007 From: martin.hooper at gmail.com (Martin Hooper) Date: Sat, 18 Aug 2007 15:29:53 +0100 Subject: Kisckstart file to make an unattended installation Message-ID: <635a92990708180729p5511d7echf5a67f71ea51f0b4@mail.gmail.com> Given this Kickstart file: http://martinjh.myby.co.uk/ks.cfg How do I make it unattended apart from the partitioning? Also I would like to know wether the package list is fine for a desktop PC with printing and sound..? -- Martin Hooper http://www.martinjh.myby.co.uk From jkeating at j2solutions.net Sat Aug 18 15:10:43 2007 From: jkeating at j2solutions.net (Jesse Keating) Date: Sat, 18 Aug 2007 11:10:43 -0400 Subject: Kisckstart file to make an unattended installation In-Reply-To: <635a92990708180729p5511d7echf5a67f71ea51f0b4@mail.gmail.com> References: <635a92990708180729p5511d7echf5a67f71ea51f0b4@mail.gmail.com> Message-ID: <20070818111043.32a091d9@ender> On Sat, 18 Aug 2007 15:29:53 +0100 "Martin Hooper" wrote: > How do I make it unattended apart from the partitioning? Just remove the bootloader and clearpart directives and you should be prompted for those during install. -- Jesse Keating RHCE (jkeating.livejournal.com) Fedora Project (fedoraproject.org/wiki/JesseKeating) GPG Public Key (geek.j2solutions.net/jkeating.j2solutions.pub) -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: not available URL: From martin.hooper at gmail.com Sun Aug 19 07:44:34 2007 From: martin.hooper at gmail.com (Martin Hooper) Date: Sun, 19 Aug 2007 08:44:34 +0100 Subject: Kisckstart file to make an unattended installation In-Reply-To: <20070818111043.32a091d9@ender> References: <635a92990708180729p5511d7echf5a67f71ea51f0b4@mail.gmail.com> <20070818111043.32a091d9@ender> Message-ID: <635a92990708190044s7dc72b1peeca134268855643@mail.gmail.com> On 18/08/07, Jesse Keating wrote: > On Sat, 18 Aug 2007 15:29:53 +0100 > "Martin Hooper" wrote: > > > How do I make it unattended apart from the partitioning? > > Just remove the bootloader and clearpart directives and you should be > prompted for those during install. That worked thanks! -- Martin Hooper http://www.martinjh.myby.co.uk From martin.hooper at gmail.com Sun Aug 19 19:38:46 2007 From: martin.hooper at gmail.com (Martin Hooper) Date: Sun, 19 Aug 2007 20:38:46 +0100 Subject: Kisckstart file to make an unattended installation In-Reply-To: <635a92990708190044s7dc72b1peeca134268855643@mail.gmail.com> References: <635a92990708180729p5511d7echf5a67f71ea51f0b4@mail.gmail.com> <20070818111043.32a091d9@ender> <635a92990708190044s7dc72b1peeca134268855643@mail.gmail.com> Message-ID: <635a92990708191238l6c1dffc1p1d108a17be07511e@mail.gmail.com> Just a quick question - What is the codes needed for UK Language and UK Keyboard? -- Martin Hooper http://www.martinjh.myby.co.uk From richzendy at gmail.com Mon Aug 20 04:44:31 2007 From: richzendy at gmail.com (Edwind Richzendy Contreras Soto) Date: Mon, 20 Aug 2007 00:44:31 -0400 Subject: Kisckstart file to make an unattended installation In-Reply-To: <635a92990708191238l6c1dffc1p1d108a17be07511e@mail.gmail.com> References: <635a92990708180729p5511d7echf5a67f71ea51f0b4@mail.gmail.com> <20070818111043.32a091d9@ender> <635a92990708190044s7dc72b1peeca134268855643@mail.gmail.com> <635a92990708191238l6c1dffc1p1d108a17be07511e@mail.gmail.com> Message-ID: <90ba020d0708192144x39af6b23h2defab0e4582c166@mail.gmail.com> 2007/8/19, Martin Hooper : > > Just a quick question - What is the codes needed for UK Language and > UK Keyboard? try system-config-kickstart set your language/keyboard preferences and see this params in a example kickstart -- > Martin Hooper > http://www.martinjh.myby.co.uk > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From sys.mailing at gmail.com Mon Aug 20 19:48:30 2007 From: sys.mailing at gmail.com (Bjorn Oglefjorn) Date: Mon, 20 Aug 2007 15:48:30 -0400 Subject: possible bug in kickstart with RHEL4 and %pre scripts using LVM In-Reply-To: <46C60943.7030006@ip-solutions.net> References: <46C60943.7030006@ip-solutions.net> Message-ID: <926ab61b0708201248s7eac4258paeed2123e62769ff@mail.gmail.com> Here you go: --- partinfo 2007-08-20 15:47:18.000000000 -0400 +++ partinfo.new 2007-08-20 15:47:44.000000000 -0400 @@ -10,7 +10,7 @@ S2=$4 # This is the size of disk 2, etc. # This would be a partition scheme for two or more drives -if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo +if [ $numd -ge 2 ]; then cat << EOF > /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary part swap --size=512 --ondisk=$d1 --asprimary @@ -25,7 +25,7 @@ else -cat << EOF >> /tmp/partinfo +cat << EOF > /tmp/partinfo part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary part swap --size=1024 --ondisk=$d1 --asprimary part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 --BO On 8/17/07, Harry Hoffman wrote: > > Hi All, > > Ran into what I believe is a bug. It looks to be in the %pre section. > > Background: > > Running CentOS release 5 with cobbler 0.5.2. I have a kickstart file for > installing RHEL 4.5 (see below). > > I have a %pre section in the kickstart file that defines partitions: > > > # Magically figure out how to partition this thing > %include /tmp/partinfo > > %pre > # Determine how many drives we have > set $(list-harddrives) > let numd=$#/2 > d1=$1 # This is the device of disk 1 > d2=$3 # This is the device of disk 2, etc. > S1=$2 # This is the size of disk 1 > S2=$4 # This is the size of disk 2, etc. > > # This would be a partition scheme for two or more drives > if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo > part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary > part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary > part swap --size=512 --ondisk=$d1 --asprimary > part swap --size=512 --ondisk=$d2 --asprimary > part raid.01 --size=1 --grow --ondisk=$d1 > part raid.02 --size=1 --grow --ondisk=$d2 > raid pv.01 --fstype "physical volume (LVM)" --level=RAID1 --device=md0 > raid.01 raid.02 > volgroup internal_hd pv.01 > logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow > EOF > > else > > cat << EOF >> /tmp/partinfo > part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary > part swap --size=1024 --ondisk=$d1 --asprimary > part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 > volgroup internal_hd pv.01 > logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow > #part / --fstype ext3 --size=1 --grow --ondisk=$d1 > EOF > > fi > > > This pre script shows the following errors on vt3: > * Running kickstart %pre script(s) > let: arith: syntax error: "numd=2/2" > > let: arith: syntax error: "numd=2/2" > > [: 2: unexpected operator > > * All kickstart %pre script(s) have been run > * Exception parsing ks.cfg: Defined PV partition pv.01 multiple times > > > In this particular case I have only one (logical HW RAID) drive and the > /tmp/partinfo is created with the lines contained in the else clause > from above. > > The error about partition pv.01 causes kickstart to fail and die :-( > > Please see the full kickstart file below. > > I should note that if I place the part info from the else clause outside > of the %pre section (i.e. just create a static partition declaration in > the kickstart file) then everything works as expected. > > Any ideas? I'm stumped. > > Cheers, > Harry > > > # Kickstart file to automate the installation of Oracle 10 on RHEL4 > # Harry Hoffman > # Oracle 10g Kickstart file for CentOS 4.x > #platform=x86, AMD64, or Intel EM64T > # System authorization information > auth --useshadow --enablemd5 > # System bootloader configuration > bootloader --location=mbr --md5pass=blah > # Partition clearing information > clearpart --all --initlabel > # Use text mode install > text > # Firewall configuration > firewall --enabled --port=22:tcp > # Run the Setup Agent on first boot > firstboot --disable > # System keyboard > keyboard us > # System language > lang en_US.UTF-8 > langsupport --default=en_US.UTF-8 en_US.UTF-8 > # Use network installation > url --url=http://192.168.1.3/cblr/links/RHEL-4.5-i386 > # Network information > network --bootproto=static --ip=192.168.1.10 --netmask=255.255.240.0 > --gateway=192.168.1.249 --nameserver=192.168.1.3 > --hostname=db1.prod.fqdn.com --device=eth0 --onboot=on > # Reboot after installation > reboot > > #Root password > rootpw --iscrypted blah > # SELinux configuration > selinux --enforcing > # Do not configure the X Window System > skipx > # System timezone > timezone America/New_York > # Install OS instead of upgrade > install > # Clear the Master Boot Record > zerombr > > > > > # Magically figure out how to partition this thing > %include /tmp/partinfo > > %pre > # Determine how many drives we have > set $(list-harddrives) > let numd=$#/2 > d1=$1 # This is the device of disk 1 > d2=$3 # This is the device of disk 2, etc. > S1=$2 # This is the size of disk 1 > S2=$4 # This is the size of disk 2, etc. > > # This would be a partition scheme for two or more drives > if [ $numd -ge 2 ]; then cat << EOF >> /tmp/partinfo > part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary > part /boot2 --fstype ext3 --size=100 --ondisk=$d2 --asprimary > part swap --size=512 --ondisk=$d1 --asprimary > part swap --size=512 --ondisk=$d2 --asprimary > part raid.01 --size=1 --grow --ondisk=$d1 > part raid.02 --size=1 --grow --ondisk=$d2 > raid pv.01 --fstype "physical volume (LVM)" --level=RAID1 --device=md0 > raid.01 raid.02 > volgroup internal_hd pv.01 > logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow > EOF > > else > > cat << EOF >> /tmp/partinfo > part /boot --fstype ext3 --size=100 --ondisk=$d1 --asprimary > part swap --size=1024 --ondisk=$d1 --asprimary > part pv.01 --fstype "physical volume (LVM)" --size=1 --grow --ondisk=$d1 > volgroup internal_hd pv.01 > logvol / --fstype ext3 --name=slash --vgname=internal_hd --size=1 --grow > #part / --fstype ext3 --size=1 --grow --ondisk=$d1 > EOF > > fi > > > %packages > automake > bison > cpp > flex > gcc > gcc-c++ > glibc-devel > glibc-headers > glibc-kernheaders > libungif > Xaw3d > xorg-x11-xauth > libcap > ntp > sudo > vlock > screen > sysstat > -sendmail > postfix > vim-enhanced > # Oracle 10 stuff > compat-db > openmotif > setarch > libaio > libaio-devel > compat-gcc-32 > compat-gcc-32-c++ > compat-libstdc++-33 > > %post > chvt 3 > > > #Turn off/on services > chkconfig --level 2345 isdn off > chkconfig --level 2345 portmap off > chkconfig --level 2345 nfslock off > chkconfig --level 345 ntpd on > chkconfig --level 345 cups off > > #Setup root mail forwarding > echo -e "root:\troot at fqdn.com" >> /etc/aliases > postalias /etc/aliases > > > # Setup Oracle > groupadd -r oinstall; > groupadd -r dba; > useradd -r -c "Oracle DB User" -g oinstall -G dba -m -p 'blah' oracle > > echo "# " >> /etc/sysctl.conf > echo "# Required for Oracle 10g " >> /etc/sysctl.conf > echo "# " >> /etc/sysctl.conf > echo "kernel.shmall = 2097152" >> /etc/sysctl.conf > echo "kernel.shmmax = 2147483648" >> /etc/sysctl.conf > echo "kernel.shmmni = 4096" >> /etc/sysctl.conf > echo "kernel.sem = 250 32000 100 128" >> /etc/sysctl.conf > echo "fs.file-max = 65536" >> /etc/sysctl.conf > echo "net.ipv4.ip_local_port_range = 1024 65000" >> /etc/sysctl.conf > echo "net.core.rmem_default=262144" >> /etc/sysctl.conf > echo "net.core.rmem_max=262144" >> /etc/sysctl.conf > echo "net.core.wmem_default=262144" >> /etc/sysctl.conf > echo "net.core.wmem_max=262144" >> /etc/sysctl.conf > /sbin/sysctl -p > > > echo "oracle soft nproc 2047" >> > /etc/security/limits.conf > echo "oracle hard nproc 16384" >> > /etc/security/limits.conf > echo "oracle soft nofile 1024" >> > /etc/security/limits.conf > echo "oracle hard nofile 65536" >> > /etc/security/limits.conf > > # setting /etc/profile > echo "if [ \$USER = \"oracle\" ]; then" >> /etc/profile > echo " if [ \$SHELL = \"/bin/ksh\" ]; then" >> /etc/profile > echo " ulimit -p 16384" >> /etc/profile > echo " ulimit -n 65536" >> /etc/profile > echo " else" >> /etc/profile > echo " ulimit -u 16384 -n 65536" >> /etc/profile > echo " fi" >> /etc/profile > echo "fi" >> /etc/profile > > # Change pam to respect the new limits > echo "session required /lib/security/pam_limits.so" >> > /etc/pam.d/login > > # Make oracle directories > mkdir -p /opt/oracle/product/10.2.0.1/db_1 > chown -R oracle:oinstall /opt/oracle/ > chmod -R 775 /opt/oracle/ > > # Set ORACLE_HOME in oracle's .bashrc > echo "ORACLE_HOME=/opt/oracle/product/10.2.0.1/db_1" >> > /home/oracle/.bashrc > echo "PATH=\$PATH:\$ORACLE_HOME/bin" >> /home/oracle/.bashrc > echo "export \$PATH" >> /home/oracle/.bashrc > > touch /etc/oratab > chown oracle:oinstall /etc/oratab > > > wget -O /home/oracle/10201_database_linux32.zip > http://192.168.1.3/cblr/localmirror/10201_database_linux32.zip > wget -O /home/oracle/10gR2.rsp > http://192.168.1.3/cblr/localmirror/10gR2.rsp > > chown oracle:oinstall /home/oracle/10201_database_linux32.zip > > cd /home/oracle && su oracle -c "unzip 10201_database_linux32.zip" > > wget -O /etc/rc3.d/S99oracle_install > http://192.168.1.3/cblr/localmirror/S99oracle_install > chmod 755 /etc/rc3.d/S99oracle_install > > > # Update the system > # Update the system > /usr/sbin/rhnreg_ks --profilename "db1.prod.fqdn.com" --username > "rhuser" --password "blah" --email "root at fqdn.com" > up2date -fu up2date > up2date -fu > > > wget http://192.168.1.3/cblr/watcher.py?system_done=RHEL-4-Oracle -b > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From michael at kmaclub.com Mon Aug 20 21:53:52 2007 From: michael at kmaclub.com (Michael) Date: Mon, 20 Aug 2007 14:53:52 -0700 Subject: Installing minimal number of packages Message-ID: <46CA0D70.6040502@kmaclub.com> Hello, I am doing a fc6 install for testing an automated app server install. In the packages section, I just have %packages @core sox screen mysql-server I would expect a few small number of packages ( around 100 ). However, I am getting 550 installed. I don't really understand why so many packages are getting pulled in. How can I debug this? Is @core no longer the right directive? Michael From Chip.Shabazian at bankofamerica.com Mon Aug 20 22:16:14 2007 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Mon, 20 Aug 2007 15:16:14 -0700 Subject: Installing minimal number of packages In-Reply-To: <46CA0D70.6040502@kmaclub.com> Message-ID: You still get base. A truly minimal install would be %packages --no-base But that would result in a largely unusable system, so you need to add packages back in from that point. NOTE: I have not tested this on FC6, but expect it to be the same. -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael Sent: Monday, August 20, 2007 2:54 PM To: Discussion list about Kickstart Subject: Installing minimal number of packages Hello, I am doing a fc6 install for testing an automated app server install. In the packages section, I just have %packages @core sox screen mysql-server I would expect a few small number of packages ( around 100 ). However, I am getting 550 installed. I don't really understand why so many packages are getting pulled in. How can I debug this? Is @core no longer the right directive? Michael _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From michael at kmaclub.com Mon Aug 20 22:55:22 2007 From: michael at kmaclub.com (Michael) Date: Mon, 20 Aug 2007 15:55:22 -0700 Subject: Installing minimal number of packages In-Reply-To: References: Message-ID: <46CA1BDA.2060407@kmaclub.com> Shabazian, Chip wrote: > You still get base. A truly minimal install would be > > %packages --no-base > > But that would result in a largely unusable system, so you need to add > packages back in from that point. > > NOTE: I have not tested this on FC6, but expect it to be the same. > > Is there not a happy medium somewhere in betwen? I definitely want a stripped system that includes rpm, and basic functionality but I don't want X, or lots of bulk. Doing --no-base and adding back all the packages just to make it function correctly seems tedious to get correct. If that is the best way, can someone give me an example of what they add back? Michael Michael From Chip.Shabazian at bankofamerica.com Mon Aug 20 23:32:09 2007 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Mon, 20 Aug 2007 16:32:09 -0700 Subject: Installing minimal number of packages In-Reply-To: <46CA1BDA.2060407@kmaclub.com> Message-ID: Outside of doing a genhd and creating your own comps.xml file (complicated, but getting easier), you can either start with --no-base and add packages back in, or keep base and remove packages you don't want: %packages -remove_this -and_this I remove packages, but do end up with a large packages section full of removals. It's been a while since I've done it, but if you play with it, I have been able to whittle an install down to just under 100 rpm's. -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael Sent: Monday, August 20, 2007 3:55 PM To: Discussion list about Kickstart Subject: Re: Installing minimal number of packages Shabazian, Chip wrote: > You still get base. A truly minimal install would be > > %packages --no-base > > But that would result in a largely unusable system, so you need to add > packages back in from that point. > > NOTE: I have not tested this on FC6, but expect it to be the same. > > Is there not a happy medium somewhere in betwen? I definitely want a stripped system that includes rpm, and basic functionality but I don't want X, or lots of bulk. Doing --no-base and adding back all the packages just to make it function correctly seems tedious to get correct. If that is the best way, can someone give me an example of what they add back? Michael Michael _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From jscott at cpinternet.com Mon Aug 20 23:44:24 2007 From: jscott at cpinternet.com (J.J. Scott) Date: Mon, 20 Aug 2007 18:44:24 -0500 Subject: Installing minimal number of packages In-Reply-To: <46CA1BDA.2060407@kmaclub.com> Message-ID: <042201c7e384$0ab6aa40$3200000a@rogue> I just PXE booted a VM instance from this tiny.ks ( only changed the root password ) this is what it ends up being... 248 Packages ( rpm -qa | wc -l ) Disk Space ( du -k --max-depth=1 | sort -n ) 0 ./selinux 0 ./sys 8 ./home 8 ./initrd 8 ./mnt 8 ./opt 8 ./srv 16 ./lost+found 16 ./tmp 24 ./media 152 ./dev 192 ./root 5364 ./bin 5768 ./boot 8988 ./etc 12644 ./var 12704 ./sbin 87972 ./lib 262427 ./proc 510904 ./usr 907223 . ---------------------------------------------------------------------------- ------- tiny.ks ---------------------------------------------------------------------------- ------- lang en_US langsupport en_US keyboard us mouse timezone America/Chicago rootpw --iscrypted $$$$$$$$$$$$$$$$$$$$$$$$$$$$$ reboot text interactive install url --url http://10.0.0.1/mrepo/centos4-i386/ zerombr yes clearpart --all --initlabel part /boot --fstype ext3 --size 256 part swap --size 512 part / --fstype ext3 --size 1 --grow auth --useshadow --enablemd5 network --bootproto=dhcp firewall --disabled xconfig --depth=16 --resolution=1024x768 --defaultdesktop=GNOME firstboot --reconfig %packages --resolvedeps sudo kernel grub openssh-server openssh openssh-clients joe yum curl openldap-clients xorg-x11-xauth ntp sysstat sharutils -anacron -apmd -autofs -bluez-libs -bluez-bluefw -bluez-hcidump -bluez-utils -comps -cups -cups-libs -desktop-file-utils -dhcpv6_client -diskdumputils -dmraid -eject -finger -lftp -logwatch -rpmdb-CentOS -fbset -freetype -fontconfig -htmlview -ipsec-tools -iptables -irda-utils -isdn4k-utils -lockdev -mailcap -mdadm -mgetty -minicom -mt-st -nano -nc -netdump -nfs-utils -quota -pcmcia-cs -pinfo -portmap -rdist -rmt -rp-pppoe -rsh -statserial -setserial -slocate -specspo -stunnel -sysreport -system-config-securitylevel-tui -system-config-network-tui -talk -tcpdump -up2date -vconfig -wvdial -wireless-tools -ypbind -yp-tools -wireless-tools -NetworkManager -redhat-lsb system-config-mouse -pyxf86config -rhpl -libwvstreams -ppp -utemper -wireless-tools %post mkdir -p /root/bin; cd /root/bin wget -q http://10.0.0.1/mrepo/centos4-i386/install/post-install.sh sh /root/bin/post-install.sh 2>&1 | tee >> /root/post-install.log > -----Original Message----- > From: kickstart-list-bounces at redhat.com > [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael > Sent: Monday, August 20, 2007 5:55 PM > To: Discussion list about Kickstart > Subject: Re: Installing minimal number of packages > > Shabazian, Chip wrote: > > You still get base. A truly minimal install would be > > > > %packages --no-base > > > > But that would result in a largely unusable system, so you > need to add > > packages back in from that point. > > > > NOTE: I have not tested this on FC6, but expect it to be the same. > > > > > > Is there not a happy medium somewhere in betwen? I > definitely want a stripped system that includes rpm, and > basic functionality but I don't want X, or lots of bulk. > > Doing --no-base and adding back all the packages just to make > it function correctly seems tedious to get correct. > > If that is the best way, can someone give me an example of > what they add back? > > Michael > > Michael > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > No virus found in this incoming message. > Checked by AVG Free Edition. > Version: 7.5.484 / Virus Database: 269.12.0/961 - Release > Date: 8/19/2007 7:27 AM > > No virus found in this outgoing message. Checked by AVG Free Edition. Version: 7.5.484 / Virus Database: 269.12.0/961 - Release Date: 8/19/2007 7:27 AM From Joe_Wulf at yahoo.com Mon Aug 20 23:24:54 2007 From: Joe_Wulf at yahoo.com (Joe_Wulf) Date: Mon, 20 Aug 2007 19:24:54 -0400 Subject: Installing minimal number of packages In-Reply-To: <46CA1BDA.2060407@kmaclub.com> References: <46CA1BDA.2060407@kmaclub.com> Message-ID: I REALLY like RedHat, as opposed to other distributions. Works well for me both at work and at home. I hear you, Michael with regards to the difficulties in getting to a truly "MINIMAL" installation. I've got quite a bit of experience working with RedHat Enterprise Linux, Advanced Server v4.0. The 'extra' stuff that gets installed is almost 150 packages that add no value and have to be weeded out. I've found about the same perspective with the Fedora line as well. The method I've found to achieve a functional system that has minimal packages installed is to install it by hand once, utilize the RPM commands to get info on the packages, then build a list of what I want to keep and what I want to remove. I've then coded a script and included with it a package disposition text file. The script removes all the excess while leaving a functional system in place. My rule was an objective evaluation of every package---if I could honestly see 70 percent of the systems actually use packageX, in any given enterprise, then the package stayed in the baseline I was building. That is a pretty rigorous methodology, yet has achieve stable results. I'm currently working on translating that to a kickstart process, as well as apply these lessons learned to RHEL v5. hooo boy, is THAT gonna be fun! I've posted to some of the RedHat and fedora lists in the past about providing significant fidelity to the package installation process. So far, there seems to be great interest in allowing the existing frustrations and confusion to reign. Puzzling and sad. In these days of system exploitations, I would have thought RedHat would desire to lead the charge for a system that is easy to install, minimizes the attack surface miscreants have open to them, robustly secures all facets of the OS from a holistic perspective. I wish you luck. But, realize too, that in the end, you'll know a hell of a lot more about the RedHat OS than you knew before. R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael Sent: Monday, August 20, 2007 18:55 To: Discussion list about Kickstart Subject: Re: Installing minimal number of packages Shabazian, Chip wrote: > You still get base. A truly minimal install would be > > %packages --no-base > > But that would result in a largely unusable system, so you need to add > packages back in from that point. > > NOTE: I have not tested this on FC6, but expect it to be the same. > > Is there not a happy medium somewhere in betwen? I definitely want a stripped system that includes rpm, and basic functionality but I don't want X, or lots of bulk. Doing --no-base and adding back all the packages just to make it function correctly seems tedious to get correct. If that is the best way, can someone give me an example of what they add back? Michael Michael _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From mdehaan at redhat.com Tue Aug 21 14:38:17 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Tue, 21 Aug 2007 10:38:17 -0400 Subject: Installing minimal number of packages In-Reply-To: References: Message-ID: <46CAF8D9.6070207@redhat.com> Shabazian, Chip wrote: > > Outside of doing a genhd and creating your own comps.xml file > (complicated, but getting easier), you can either start with --no-base > and add packages back in, or keep base and remove packages you don't > want Cobbler ships with one example kickstart that uses the following: %packages --nobase crontabs dhclient dhcpv6_client nfs-utils openssh-clients openssh-server yum That's fairly minimal, but still includes yum so you can install what you might have accidentally left out. Work up from there. --Michael DeHaan From Chip.Shabazian at bankofamerica.com Tue Aug 21 15:33:14 2007 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Tue, 21 Aug 2007 08:33:14 -0700 Subject: Installing minimal number of packages In-Reply-To: Message-ID: Joe, I feel (and have lived through) your pain. Our objective was to create a "core image" that fills 80% - 90% of all non application needs on build, and we have been quite successful with it, Starting with RHEL 2.1. With RHEL 5, I have seen more improvements (personally) with kickstart than with any previous release, going back to RHL 7.0 which is when I started doing significant work with kickstart. Many of the challenges that have existed for years, are now resolved either through kickstart or other tools. Some highlights: You can now specify yum repo's in your command section so that systems can be built with current releases from a tested stable build All kickstart tools now use the same UI for package selection (it used to be three different code bases to maintain) Ability to have multiple %post sections (thus allowing you to do work in --no-chroot, then work in a chrooted environment) Tools such as Revisor make it almost TOO simple to create your own "single install CD" or live CD with what you want Anyway, there has been significant work on Red Hat's part to make improvements to kickstart, and it really shows in RHEL 5. I for one am happy to see the progress and can't wait until we can move forward with RHEL 5 deployments and get off 3 and 4. I covered the above, and some others in my recent Linuxworld presentation which you can download at http://www.shabazian.com/lw2007.pdf -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Joe_Wulf Sent: Monday, August 20, 2007 4:25 PM To: 'Discussion list about Kickstart' Subject: RE: Installing minimal number of packages I REALLY like RedHat, as opposed to other distributions. Works well for me both at work and at home. I hear you, Michael with regards to the difficulties in getting to a truly "MINIMAL" installation. I've got quite a bit of experience working with RedHat Enterprise Linux, Advanced Server v4.0. The 'extra' stuff that gets installed is almost 150 packages that add no value and have to be weeded out. I've found about the same perspective with the Fedora line as well. The method I've found to achieve a functional system that has minimal packages installed is to install it by hand once, utilize the RPM commands to get info on the packages, then build a list of what I want to keep and what I want to remove. I've then coded a script and included with it a package disposition text file. The script removes all the excess while leaving a functional system in place. My rule was an objective evaluation of every package---if I could honestly see 70 percent of the systems actually use packageX, in any given enterprise, then the package stayed in the baseline I was building. That is a pretty rigorous methodology, yet has achieve stable results. I'm currently working on translating that to a kickstart process, as well as apply these lessons learned to RHEL v5. hooo boy, is THAT gonna be fun! I've posted to some of the RedHat and fedora lists in the past about providing significant fidelity to the package installation process. So far, there seems to be great interest in allowing the existing frustrations and confusion to reign. Puzzling and sad. In these days of system exploitations, I would have thought RedHat would desire to lead the charge for a system that is easy to install, minimizes the attack surface miscreants have open to them, robustly secures all facets of the OS from a holistic perspective. I wish you luck. But, realize too, that in the end, you'll know a hell of a lot more about the RedHat OS than you knew before. R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael Sent: Monday, August 20, 2007 18:55 To: Discussion list about Kickstart Subject: Re: Installing minimal number of packages Shabazian, Chip wrote: > You still get base. A truly minimal install would be > > %packages --no-base > > But that would result in a largely unusable system, so you need to add > packages back in from that point. > > NOTE: I have not tested this on FC6, but expect it to be the same. > > Is there not a happy medium somewhere in betwen? I definitely want a stripped system that includes rpm, and basic functionality but I don't want X, or lots of bulk. Doing --no-base and adding back all the packages just to make it function correctly seems tedious to get correct. If that is the best way, can someone give me an example of what they add back? Michael Michael _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From Joe_Wulf at yahoo.com Wed Aug 22 13:31:25 2007 From: Joe_Wulf at yahoo.com (Joe_Wulf) Date: Wed, 22 Aug 2007 09:31:25 -0400 Subject: Installing minimal number of packages In-Reply-To: References: Message-ID: Chip, Reviewed your article, very informative from what I can see on the slides. Would benefit from the 'audio' that goes with each slide, I'm sure. I would have liked to have seen or found a 'users manual' for kickstart. Something that explains the entire process from end to end. Several things you mention in your presentation I've never seen stated anywhere else. Other questions: - When I build a system with KS, why does the /root/anaconda-ks.cfg file NOT reflect exactly what the source ks.cfg looked like? - I want to build many systems from many versions of the RedHat OS's. Why doesn't the kickstart configuration NATURALLY understand each OS I've copied into my ks server and build ks-cfg files that work natively with that version of the OS. A: I know, it wasn't designed to. The real question is why hasn't that been done yet! I've got others but don't have the time this moment to pull them out of my brain. With RHEL5, do you have any issues, tips, tricks and/or solutions? Also, I'd like to get with you off-list, and chat about this, too. Let me know if you would as well (Joe_Wulf at yahoo.com). R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Shabazian, Chip Sent: Tuesday, August 21, 2007 11:33 To: Discussion list about Kickstart Subject: RE: Installing minimal number of packages Joe, I feel (and have lived through) your pain. Our objective was to create a "core image" that fills 80% - 90% of all non application needs on build, and we have been quite successful with it, Starting with RHEL 2.1. With RHEL 5, I have seen more improvements (personally) with kickstart than with any previous release, going back to RHL 7.0 which is when I started doing significant work with kickstart. Many of the challenges that have existed for years, are now resolved either through kickstart or other tools. Some highlights: You can now specify yum repo's in your command section so that systems can be built with current releases from a tested stable build All kickstart tools now use the same UI for package selection (it used to be three different code bases to maintain) Ability to have multiple %post sections (thus allowing you to do work in --no-chroot, then work in a chrooted environment) Tools such as Revisor make it almost TOO simple to create your own "single install CD" or live CD with what you want Anyway, there has been significant work on Red Hat's part to make improvements to kickstart, and it really shows in RHEL 5. I for one am happy to see the progress and can't wait until we can move forward with RHEL 5 deployments and get off 3 and 4. I covered the above, and some others in my recent Linuxworld presentation which you can download at http://www.shabazian.com/lw2007.pdf -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Joe_Wulf Sent: Monday, August 20, 2007 4:25 PM To: 'Discussion list about Kickstart' Subject: RE: Installing minimal number of packages I REALLY like RedHat, as opposed to other distributions. Works well for me both at work and at home. I hear you, Michael with regards to the difficulties in getting to a truly "MINIMAL" installation. I've got quite a bit of experience working with RedHat Enterprise Linux, Advanced Server v4.0. The 'extra' stuff that gets installed is almost 150 packages that add no value and have to be weeded out. I've found about the same perspective with the Fedora line as well. The method I've found to achieve a functional system that has minimal packages installed is to install it by hand once, utilize the RPM commands to get info on the packages, then build a list of what I want to keep and what I want to remove. I've then coded a script and included with it a package disposition text file. The script removes all the excess while leaving a functional system in place. My rule was an objective evaluation of every package---if I could honestly see 70 percent of the systems actually use packageX, in any given enterprise, then the package stayed in the baseline I was building. That is a pretty rigorous methodology, yet has achieve stable results. I'm currently working on translating that to a kickstart process, as well as apply these lessons learned to RHEL v5. hooo boy, is THAT gonna be fun! I've posted to some of the RedHat and fedora lists in the past about providing significant fidelity to the package installation process. So far, there seems to be great interest in allowing the existing frustrations and confusion to reign. Puzzling and sad. In these days of system exploitations, I would have thought RedHat would desire to lead the charge for a system that is easy to install, minimizes the attack surface miscreants have open to them, robustly secures all facets of the OS from a holistic perspective. I wish you luck. But, realize too, that in the end, you'll know a hell of a lot more about the RedHat OS than you knew before. R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael Sent: Monday, August 20, 2007 18:55 To: Discussion list about Kickstart Subject: Re: Installing minimal number of packages Shabazian, Chip wrote: > You still get base. A truly minimal install would be > > %packages --no-base > > But that would result in a largely unusable system, so you need to add > packages back in from that point. > > NOTE: I have not tested this on FC6, but expect it to be the same. > > Is there not a happy medium somewhere in betwen? I definitely want a stripped system that includes rpm, and basic functionality but I don't want X, or lots of bulk. Doing --no-base and adding back all the packages just to make it function correctly seems tedious to get correct. If that is the best way, can someone give me an example of what they add back? Michael Michael _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From mdehaan at redhat.com Wed Aug 22 14:43:59 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 22 Aug 2007 10:43:59 -0400 Subject: Installing minimal number of packages In-Reply-To: References: Message-ID: <46CC4BAF.7090003@redhat.com> Joe_Wulf wrote: > > I would have liked to have seen or found a 'users manual' for kickstart. > http://www.redhat.com/docs/manuals/enterprise/RHEL-5-manual/Installation_Guide-en-US/index.html http://www.redhat.com/docs/manuals/enterprise/RHEL-5-manual/Installation_Guide-en-US/s1-kickstart2-options.html and "yum install system-config-kickstart" > - When I build a system with KS, why does the /root/anaconda-ks.cfg file > NOT reflect exactly what the source ks.cfg looked like? > Anaconda records all of it's choices, including choices made interactively in the installer. The anaconda.ks file can be used to reproduce the install choices, including questions not posed in the original kickstart. Perhaps you could give a concrete example of something showing up that you did not expect and was not what you wanted? > - I want to build many systems from many versions of the RedHat OS's. Why > doesn't the kickstart configuration NATURALLY understand each OS I've > copied into my ks server and build ks-cfg files that work natively with > that version of the OS. > Each version of the OS has new features that cannot be expressed in previous versions. An example is the "repo" functionality that was added for RHEL 5 / FC 6. This functionality is very powerful but was not available in EL 4. Ergo, if you request to use it in your kickstart, that should be an error, else you won't be able to install packages from external repos that you may have mentioned in the packages section. Keeping seperate kickstarts for seperate OS's is desired behavior, because they are seperate OS's and therefore require different answer files. If you want something to manage those configurations, look at something like http://cobbler.et.redhat.com, use RHN, or write your templating system, however simple or complex you'd like it to be. --Michael From rigg0022 at umn.edu Wed Aug 22 15:18:27 2007 From: rigg0022 at umn.edu (Riggs, Ben) Date: Wed, 22 Aug 2007 10:18:27 -0500 Subject: Any ideas why this fails?: In-Reply-To: <46C60DE5.5050001@redhat.com> References: <46C60DE5.5050001@redhat.com> Message-ID: <46CC53C3.4090008@umn.edu> %include /tmp/network-include %pre --interpreter /usr/bin/python #find-ethN configures network info import os #Determine which ethN interface matches mac address supplied by ldap network_statement = "network --bootproto=dhcp --onboot=on --noipv6" pipe = os.popen("ifconfig -a", 'r') for line in pipe.readlines(): if '00:e0:81:33:d6:c7' in line.lower(): network_statement = "".join((network_statement, " --device=", line.split()[0].strip(), "\n")) file = open("/tmp/network-include", 'w') file.write(network_statement) file.close() I can't find anything glaringly obvious. I know the python works. When anaconda runs, it still asks for network info manually. From Matt.Fahrner at coat.com Wed Aug 22 15:28:23 2007 From: Matt.Fahrner at coat.com (Matt Fahrner) Date: Wed, 22 Aug 2007 11:28:23 -0400 Subject: Any ideas why this fails?: In-Reply-To: <46CC53C3.4090008@umn.edu> References: <46C60DE5.5050001@redhat.com> <46CC53C3.4090008@umn.edu> Message-ID: <46CC5617.1070809@coat.com> Are you sure "ifconfig" is in your path? Riggs, Ben wrote: > > %include /tmp/network-include > > %pre --interpreter /usr/bin/python > #find-ethN configures network info > import os > > #Determine which ethN interface matches mac address supplied by ldap > network_statement = "network --bootproto=dhcp --onboot=on --noipv6" > pipe = os.popen("ifconfig -a", 'r') > for line in pipe.readlines(): > if '00:e0:81:33:d6:c7' in line.lower(): > network_statement = "".join((network_statement, " --device=", > line.split()[0].strip(), "\n")) > file = open("/tmp/network-include", 'w') > file.write(network_statement) > file.close() > > > I can't find anything glaringly obvious. I know the python works. When > anaconda runs, it still asks for network info manually. > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list -- --------------------------------------------------------------------- Matt Fahrner 2 South Park St. Chief Systems Architect Willis House Burlington Coat Factory Warehouse Lebanon, N.H. 03766 Tel: (603) 448-4100 x5150 USA Fax: (603) 443-6190 Matt.Fahrner at COAT.COM --------------------------------------------------------------------- From rigg0022 at umn.edu Wed Aug 22 15:30:17 2007 From: rigg0022 at umn.edu (Riggs, Ben) Date: Wed, 22 Aug 2007 10:30:17 -0500 Subject: Any ideas why this fails?: In-Reply-To: <46CC5617.1070809@coat.com> References: <46C60DE5.5050001@redhat.com> <46CC53C3.4090008@umn.edu> <46CC5617.1070809@coat.com> Message-ID: <46CC5689.8060803@umn.edu> Good question. I hadn't thought that it might not be in anaconda's path. Matt Fahrner wrote: > Are you sure "ifconfig" is in your path? > > Riggs, Ben wrote: >> >> %include /tmp/network-include >> >> %pre --interpreter /usr/bin/python >> #find-ethN configures network info >> import os >> >> #Determine which ethN interface matches mac address supplied by ldap >> network_statement = "network --bootproto=dhcp --onboot=on --noipv6" >> pipe = os.popen("ifconfig -a", 'r') >> for line in pipe.readlines(): >> if '00:e0:81:33:d6:c7' in line.lower(): >> network_statement = "".join((network_statement, " --device=", >> line.split()[0].strip(), "\n")) >> file = open("/tmp/network-include", 'w') >> file.write(network_statement) >> file.close() >> >> >> I can't find anything glaringly obvious. I know the python works. When >> anaconda runs, it still asks for network info manually. >> >> _______________________________________________ >> Kickstart-list mailing list >> Kickstart-list at redhat.com >> https://www.redhat.com/mailman/listinfo/kickstart-list > From Matt.Fahrner at coat.com Wed Aug 22 15:39:37 2007 From: Matt.Fahrner at coat.com (Matt Fahrner) Date: Wed, 22 Aug 2007 11:39:37 -0400 Subject: Any ideas why this fails?: In-Reply-To: <46CC5689.8060803@umn.edu> References: <46C60DE5.5050001@redhat.com> <46CC53C3.4090008@umn.edu> <46CC5617.1070809@coat.com> <46CC5689.8060803@umn.edu> Message-ID: <46CC58B9.6080806@coat.com> The other thing to do is get it to kickstart boot one way or another up through "stage2.img" and then try to run the script manually on the virtual console with a shell prompt. That's how I debugged my "%pre" scripts... - Matt Riggs, Ben wrote: > Good question. I hadn't thought that it might not be in anaconda's path. > > Matt Fahrner wrote: >> Are you sure "ifconfig" is in your path? >> >> Riggs, Ben wrote: >>> >>> %include /tmp/network-include >>> >>> %pre --interpreter /usr/bin/python >>> #find-ethN configures network info >>> import os >>> >>> #Determine which ethN interface matches mac address supplied by ldap >>> network_statement = "network --bootproto=dhcp --onboot=on --noipv6" >>> pipe = os.popen("ifconfig -a", 'r') >>> for line in pipe.readlines(): >>> if '00:e0:81:33:d6:c7' in line.lower(): >>> network_statement = "".join((network_statement, " --device=", >>> line.split()[0].strip(), "\n")) >>> file = open("/tmp/network-include", 'w') >>> file.write(network_statement) >>> file.close() >>> >>> >>> I can't find anything glaringly obvious. I know the python works. >>> When anaconda runs, it still asks for network info manually. >>> >>> _______________________________________________ >>> Kickstart-list mailing list >>> Kickstart-list at redhat.com >>> https://www.redhat.com/mailman/listinfo/kickstart-list >> > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list -- --------------------------------------------------------------------- Matt Fahrner 2 South Park St. Chief Systems Architect Willis House Burlington Coat Factory Warehouse Lebanon, N.H. 03766 Tel: (603) 448-4100 x5150 USA Fax: (603) 443-6190 Matt.Fahrner at COAT.COM --------------------------------------------------------------------- From Joe_Wulf at yahoo.com Wed Aug 22 15:11:24 2007 From: Joe_Wulf at yahoo.com (Joe_Wulf) Date: Wed, 22 Aug 2007 11:11:24 -0400 Subject: Installing minimal number of packages In-Reply-To: <46CC4BAF.7090003@redhat.com> References: <46CC4BAF.7090003@redhat.com> Message-ID: Michael, thank you very much for the feedback. Those PDF's I'll surely take a look at. The /root/anaconda-ks.cfg does seem to pick up some of the unique qualities of the HW installed on. But, seems to loose specifics, such as anything in %post. Also, the adding of a user doesn't seem to have an equivalent during the install or post sections. I'm coming at this still pretty new to kickstart. So I wish to be clear that I'm voicing concerns and frustrations.... not throwing stones. Doing things the kickstart way certainly has been entertaining, to say the least. So far I've been blessed by having the luxury of doing things at a sedate pace, in a lab environment. The learning curve to get kickstart going and viable in a production environment, at the pace mission systems NEED to be pushed out the door would have taken a lot longer, at least for me. I'll keep learning and hopefully contributing some in return. One thing that is still very difficult for me to 'get' is how to manipulate the firstboot 'stuff'. I've been developing a very robust secure hardened version of linux. I have several places where reboots exist, yet have not found good info in any of the KS docs I've looked at that describes how to build a system, and manipulate firstboot to meet my needs. Maybe that isn't a kickstart 'thing' either. Don't know yet. R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -----Original Message----- From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Michael DeHaan Sent: Wednesday, August 22, 2007 10:44 To: Discussion list about Kickstart Subject: Re: Installing minimal number of packages Joe_Wulf wrote: > > I would have liked to have seen or found a 'users manual' for kickstart. > http://www.redhat.com/docs/manuals/enterprise/RHEL-5-manual/Installation_Guide-en -US/index.html http://www.redhat.com/docs/manuals/enterprise/RHEL-5-manual/Installation_Guide-en -US/s1-kickstart2-options.html and "yum install system-config-kickstart" > - When I build a system with KS, why does the /root/anaconda-ks.cfg file > NOT reflect exactly what the source ks.cfg looked like? > Anaconda records all of it's choices, including choices made interactively in the installer. The anaconda.ks file can be used to reproduce the install choices, including questions not posed in the original kickstart. Perhaps you could give a concrete example of something showing up that you did not expect and was not what you wanted? > - I want to build many systems from many versions of the RedHat OS's. Why > doesn't the kickstart configuration NATURALLY understand each OS I've > copied into my ks server and build ks-cfg files that work natively with > that version of the OS. > Each version of the OS has new features that cannot be expressed in previous versions. An example is the "repo" functionality that was added for RHEL 5 / FC 6. This functionality is very powerful but was not available in EL 4. Ergo, if you request to use it in your kickstart, that should be an error, else you won't be able to install packages from external repos that you may have mentioned in the packages section. Keeping seperate kickstarts for seperate OS's is desired behavior, because they are seperate OS's and therefore require different answer files. If you want something to manage those configurations, look at something like http://cobbler.et.redhat.com, use RHN, or write your templating system, however simple or complex you'd like it to be. --Michael _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list From mdehaan at redhat.com Wed Aug 22 18:21:32 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Wed, 22 Aug 2007 14:21:32 -0400 Subject: Installing minimal number of packages In-Reply-To: References: <46CC4BAF.7090003@redhat.com> Message-ID: <46CC7EAC.8010704@redhat.com> Joe_Wulf wrote: > Michael, thank you very much for the feedback. > Those PDF's I'll surely take a look at. > > The /root/anaconda-ks.cfg does seem to pick up some of the unique qualities of > the HW > installed on. But, seems to loose specifics, such as anything in %post. Also, > the > adding of a user doesn't seem to have an equivalent during the install or post > sections. > Ok, I wasn't aware of that. Maybe some one has an answer for that, though you might want to file an RFE in bugzilla about keeping the %post and user stuff. > I've looked at that describes how to build a system, and manipulate firstboot to > meet > my needs. Maybe that isn't a kickstart 'thing' either. Don't know yet. > Typically I disable firstboot in kickstart. (firstboot --disable). One thing you can do if you want is to create a service script in /etc/rc.d that runs what you want on boot, and then the last line of that script you can disable the script by either moving/deleting it or running chkconfig to disable it. Effectively you've done the same as firstboot, without using firstboot. That being said, I'm sure there are others that get good mileage out of it. The other way to go would be to pick up a configuration management system (puppet, cfengine, bcfg2, etc) and let the configuration management system do the post-install setup you require. That is really the ideal approach. From muksyed at stanford.edu Thu Aug 23 05:23:00 2007 From: muksyed at stanford.edu (Mukarram Syed) Date: Wed, 22 Aug 2007 22:23:00 -0700 Subject: Installing RedHat on a SAN disk using kickstart Message-ID: <0eda01c7e545$acdda720$6401a8c0@stanford.edu> Hi, Any suggestions would be appreciated on this Subject of Installing RedHat on a SAN disk using Kickstart: I have been using Kickstart to install RedHat Linux on local disks. But a requirement arose in our department to install Linux on a SAN disk which I have never done before. We will be using Emulex Card with lpfc drivers. How could one incorporate loading the lpfc driver upon bootup so that it could recognize the SAN disk and install OS on this. Should I rebuild the kernel or the initrd file to load the lpfc drivers? If yes, how? Which partitions will it use to install the OS (sda/sdb or something else)? Please advice. Thanks Much in advance -Mukarram. -------------- next part -------------- An HTML attachment was scrubbed... URL: From richzendy at gmail.com Thu Aug 23 21:40:54 2007 From: richzendy at gmail.com (Edwind Richzendy Contreras Soto) Date: Thu, 23 Aug 2007 17:40:54 -0400 Subject: Installing RedHat on a SAN disk using kickstart In-Reply-To: <0eda01c7e545$acdda720$6401a8c0@stanford.edu> References: <0eda01c7e545$acdda720$6401a8c0@stanford.edu> Message-ID: <90ba020d0708231440u3b817048y9c659e7c7a27d1b5@mail.gmail.com> 2007/8/23, Mukarram Syed : > > > > Hi, > > Any suggestions would be appreciated on this Subject of Installing RedHat > on a SAN disk using Kickstart: > > > > I have been using Kickstart to install RedHat Linux on local disks. But a > requirement arose in our department to install Linux on a SAN disk which I > have never done before. > > We will be using Emulex Card with lpfc drivers. How could one incorporate > loading the lpfc driver upon bootup so that it could recognize the SAN disk > and install OS on this. Should I rebuild the kernel or the initrd file to > load the lpfc drivers? If yes, how? > > Which partitions will it use to install the OS (sda/sdb or something > else)? > > Please advice. > Try mount your SAN disk in a Linux Box ( full operative ) and use NFS/http/ftp for export this directory to network. Thanks Much in advance > > > > -Mukarram. > > > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From muksyed at stanford.edu Thu Aug 23 22:39:22 2007 From: muksyed at stanford.edu (Mukarram Syed) Date: Thu, 23 Aug 2007 15:39:22 -0700 Subject: Installing RedHat on a SAN disk using kickstart In-Reply-To: <90ba020d0708231440u3b817048y9c659e7c7a27d1b5@mail.gmail.com> Message-ID: <0f9701c7e5d6$7413b880$6401a8c0@stanford.edu> Edwin, Thanks for the reply. But I don't quite understand. You mean install Linux on the local drive, get the SAN disk installed and then export which directory to network? I kinda lost you there. Please clarify. -Mukarram. Try mount your SAN disk in a Linux Box ( full operative ) and use NFS/http/ftp for export this directory to network. _____ From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Edwind Richzendy Contreras Soto Sent: Thursday, August 23, 2007 2:41 PM To: Discussion list about Kickstart Subject: Re: Installing RedHat on a SAN disk using kickstart 2007/8/23, Mukarram Syed : Hi, Any suggestions would be appreciated on this Subject of Installing RedHat on a SAN disk using Kickstart: I have been using Kickstart to install RedHat Linux on local disks. But a requirement arose in our department to install Linux on a SAN disk which I have never done before. We will be using Emulex Card with lpfc drivers. How could one incorporate loading the lpfc driver upon bootup so that it could recognize the SAN disk and install OS on this. Should I rebuild the kernel or the initrd file to load the lpfc drivers? If yes, how? Which partitions will it use to install the OS (sda/sdb or something else)? Please advice. Try mount your SAN disk in a Linux Box ( full operative ) and use NFS/http/ftp for export this directory to network. Thanks Much in advance -Mukarram. _______________________________________________ Kickstart-list mailing list Kickstart-list at redhat.com https://www.redhat.com/mailman/listinfo/kickstart-list -------------- next part -------------- An HTML attachment was scrubbed... URL: From pasik at iki.fi Fri Aug 24 08:57:23 2007 From: pasik at iki.fi (Pasi =?iso-8859-1?Q?K=E4rkk=E4inen?=) Date: Fri, 24 Aug 2007 11:57:23 +0300 Subject: Installing RedHat on a SAN disk using kickstart In-Reply-To: <0eda01c7e545$acdda720$6401a8c0@stanford.edu> References: <0eda01c7e545$acdda720$6401a8c0@stanford.edu> Message-ID: <20070824085723.GF5028@edu.joroinen.fi> On Wed, Aug 22, 2007 at 10:23:00PM -0700, Mukarram Syed wrote: > > > Hi, > > Any suggestions would be appreciated on this Subject of Installing RedHat on > a SAN disk using Kickstart: > Download driver update disk for the HBA, and boot rhel installer with "linux dd" to load it. See documentation for details. If your HBA vendor is not providing dd, you can also create it yourself. -- Pasi > > > I have been using Kickstart to install RedHat Linux on local disks. But a > requirement arose in our department to install Linux on a SAN disk which I > have never done before. > > We will be using Emulex Card with lpfc drivers. How could one incorporate > loading the lpfc driver upon bootup so that it could recognize the SAN disk > and install OS on this. Should I rebuild the kernel or the initrd file to > load the lpfc drivers? If yes, how? > > Which partitions will it use to install the OS (sda/sdb or something else)? > > Please advice. > > > > Thanks Much in advance > > > > -Mukarram. > > From Joe_Wulf at yahoo.com Tue Aug 28 05:12:16 2007 From: Joe_Wulf at yahoo.com (Joe_Wulf) Date: Tue, 28 Aug 2007 01:12:16 -0400 Subject: NFS not mountable when kickstarting RHEL5x32 via Fedora 7 Message-ID: I have a weird problem. I have Fedora 7 built up as my DNS, tftp, DHCP and kickstart server. I can successfully build RHEL AS4u5. I've now tried RHEL5.... it boots up initially, gets to the anaconda screen and freezes, never to go any further. When checking the box to be built (actually a VM), console 4 gives continuous errors regarding unable to NFS mount the kickstart source. HALP. This is weird, as I have a manually built VM with RHEL 5, full system install, and it can moments later successfully mount the same location as the KS'd VM is attempting. R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: ProSync_SignatureItem_Blue.jpg Type: image/jpeg Size: 3249 bytes Desc: not available URL: From Joe_Wulf at yahoo.com Wed Aug 29 14:24:49 2007 From: Joe_Wulf at yahoo.com (Joe_Wulf) Date: Wed, 29 Aug 2007 10:24:49 -0400 Subject: NFS not mountable when kickstarting RHEL5x32 via Fedora 7 In-Reply-To: References: Message-ID: Just to let everyone know, I've solved the problem. In my ks.cfg file, it helps if you put the ip address of the system you are building, vice that of your NFS server. Once I fixed that little detail, RHEL5 built nicely. The next problem is that the virtual machine wouldn't fully boot. It stays at the grub menu wanting booting instructions. I've checked the kind of VM I built and its hard disk drive and did it with SCSI. Does RHEL5 have any issues with SCSI? The manually built system that worked for me and is fully functional was built with IDE drives. I wonder what the difference is? R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com _____ From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Joe_Wulf Sent: Tuesday, August 28, 2007 01:12 To: kickstart-list at redhat.com Subject: NFS not mountable when kickstarting RHEL5x32 via Fedora 7 I have a weird problem. I have Fedora 7 built up as my DNS, tftp, DHCP and kickstart server. I can successfully build RHEL AS4u5. I've now tried RHEL5.... it boots up initially, gets to the anaconda screen and freezes, never to go any further. When checking the box to be built (actually a VM), console 4 gives continuous errors regarding unable to NFS mount the kickstart source. HALP. This is weird, as I have a manually built VM with RHEL 5, full system install, and it can moments later successfully mount the same location as the KS'd VM is attempting. R, -Joe Wulf, CISSP, USN(RET) Senior IA Engineer ProSync Technology Group, LLC www.prosync.com -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: ProSync_SignatureItem_Blue.jpg Type: image/jpeg Size: 3249 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: ProSync_SignatureItem_Blue.jpg Type: image/jpeg Size: 3249 bytes Desc: not available URL: From kstube at gmail.com Thu Aug 30 21:02:50 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 16:02:50 -0500 Subject: static network configuration ask Message-ID: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> Is there an option that will allow me to automate everything in kickstart except just the network portion so that I can have the same kickstart config for a bunch of clustered hosts but input the network information myself? The reason I ask is because I would like to automate a lot of other stuff in %post or on first boot but need networking to be set up properly by the time that stuff runs or it will screw all of that up. Thanks in advance Kit Stube -------------- next part -------------- An HTML attachment was scrubbed... URL: From rigg0022 at umn.edu Thu Aug 30 21:05:00 2007 From: rigg0022 at umn.edu (Riggs, Ben) Date: Thu, 30 Aug 2007 16:05:00 -0500 Subject: static network configuration ask In-Reply-To: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> Message-ID: <46D730FC.8040007@umn.edu> Check out http://cobbler.et.redhat.com It's made our life here tons easier. Ben Kit Stube wrote: > Is there an option that will allow me to automate everything in > kickstart except just the network portion so that I can have the same > kickstart config for a bunch of clustered hosts but input the network > information myself? > > The reason I ask is because I would like to automate a lot of other > stuff in %post or on first boot but need networking to be set up > properly by the time that stuff runs or it will screw all of that up. > > Thanks in advance > Kit Stube > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From Chip.Shabazian at bankofamerica.com Thu Aug 30 21:23:26 2007 From: Chip.Shabazian at bankofamerica.com (Shabazian, Chip) Date: Thu, 30 Aug 2007 14:23:26 -0700 Subject: static network configuration ask In-Reply-To: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> Message-ID: If you can use DHCP to assign the network address, even for just the first boot, that makes life the easiest. Otherwise, you can create a kickstart file that simply takes the values entered on the boot: line and creates a static network entry for them in %pre that is then included in the command section. Search the archives and you will find lots of references to both of the above. Chip ________________________________ From: kickstart-list-bounces at redhat.com [mailto:kickstart-list-bounces at redhat.com] On Behalf Of Kit Stube Sent: Thursday, August 30, 2007 2:03 PM To: kickstart-list at redhat.com Subject: static network configuration ask Is there an option that will allow me to automate everything in kickstart except just the network portion so that I can have the same kickstart config for a bunch of clustered hosts but input the network information myself? The reason I ask is because I would like to automate a lot of other stuff in %post or on first boot but need networking to be set up properly by the time that stuff runs or it will screw all of that up. Thanks in advance Kit Stube -------------- next part -------------- An HTML attachment was scrubbed... URL: From Pablo.Iranzo at uv.es Thu Aug 30 21:30:26 2007 From: Pablo.Iranzo at uv.es (=?UTF-8?Q?Pablo_Iranzo_G=C3=B3mez?=) Date: Thu, 30 Aug 2007 23:30:26 +0200 (CEST) Subject: static network configuration ask In-Reply-To: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> Message-ID: Hi - Just don't fill it and you'll get prompted for it - Send parameters on command line, process /proc/commandline, and write "semi-automatically" the network configuration files Regards Pablo -- Pablo Iranzo G?mez (http://Alufis35.uv.es/~iranzo/) (PGPKey Available on http://www.uv.es/~iranzop/PGPKey.pgp) -- Postulado de Boling sobre la Ley de Murphy: Si se encuentra bien, no se preocupe. Se le pasar? On Thu, 30 Aug 2007, Kit Stube wrote: > Is there an option that will allow me to automate everything in kickstart > except just the network portion so that I can have the same kickstart config > for a bunch of clustered hosts but input the network information myself? > > The reason I ask is because I would like to automate a lot of other stuff in > %post or on first boot but need networking to be set up properly by the > time that stuff runs or it will screw all of that up. > > Thanks in advance > Kit Stube > From kstube at gmail.com Thu Aug 30 21:38:21 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 16:38:21 -0500 Subject: static network configuration ask In-Reply-To: <46D730FC.8040007@umn.edu> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D730FC.8040007@umn.edu> Message-ID: <695c4edd0708301438j5adbc83evcf1bfa6964d6ff17@mail.gmail.com> Cobbler looks cool. Kinda like a front end for pxelinux and isc dhcpd? The problem we have is we don't want to have to look up the mac address for a host in order to configure it. Cobbler seems to rely on that. Currently we have our dhcp server on an install network dump anyone that netboots on that network to a default pxelinux menu. That default menu allows a person to chose an os to install. Chosing one kicks them into one of several different default kickstart files. If I could get kickstart to ask for static network information like it would if I was installing from CD, I would be done. On 8/30/07, Riggs, Ben wrote: > > Check out http://cobbler.et.redhat.com It's made our life here tons > easier. > > Ben > > Kit Stube wrote: > > Is there an option that will allow me to automate everything in > > kickstart except just the network portion so that I can have the same > > kickstart config for a bunch of clustered hosts but input the network > > information myself? > > > > The reason I ask is because I would like to automate a lot of other > > stuff in %post or on first boot but need networking to be set up > > properly by the time that stuff runs or it will screw all of that up. > > > > Thanks in advance > > Kit Stube > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From ebrown at lanl.gov Thu Aug 30 21:45:37 2007 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 30 Aug 2007 15:45:37 -0600 Subject: static network configuration ask In-Reply-To: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> Message-ID: <46D73A81.9050704@lanl.gov> I'm here reviewing my kickstart file because I think the answer is much simpler than suggestions so far, and I'm wondering if I'm forgetting something. I don't think so. This is the network line in the kickstart file: network --bootproto=static Then don't wait for anaconda to prompt for parameters, just enter them at the 'boot:' prompt: boot: ip=blah gateway=blah netmask=blah dns=blah I don't think it's necessary to do any processing of /proc/cmdline or includes etc, this is it. -Ed Kit Stube wrote: > Is there an option that will allow me to automate everything in kickstart > except just the network portion so that I can have the same kickstart config > for a bunch of clustered hosts but input the network information myself? > > The reason I ask is because I would like to automate a lot of other stuff in > %post or on first boot but need networking to be set up properly by the > time that stuff runs or it will screw all of that up. > > Thanks in advance > Kit Stube > > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From ebrown at lanl.gov Thu Aug 30 21:55:48 2007 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 30 Aug 2007 15:55:48 -0600 Subject: static network configuration ask In-Reply-To: <46D73A81.9050704@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> Message-ID: <46D73CE4.5030805@lanl.gov> > boot: ip=blah gateway=blah netmask=blah dns=blah This is not quite complete. And actually, we make a custom boot cd, with multiple targets that contain the common parameters, so all that's needed at the boot prompt is target, ip and gateway. You'll probably need at least this: boot: linux ks=http://url_of_kickstart_file ip=... gateway=... netmask=... dns=... -Ed From mdehaan at redhat.com Thu Aug 30 22:11:49 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Thu, 30 Aug 2007 18:11:49 -0400 Subject: static network configuration ask In-Reply-To: <695c4edd0708301438j5adbc83evcf1bfa6964d6ff17@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D730FC.8040007@umn.edu> <695c4edd0708301438j5adbc83evcf1bfa6964d6ff17@mail.gmail.com> Message-ID: <46D740A5.2050707@redhat.com> Kit Stube wrote: > Cobbler looks cool. Kinda like a front end for pxelinux and isc > dhcpd? The problem we have is we don't want to have to look up the > mac address for a host in order to configure it. Cobbler seems to > rely on that. Nah, you don't :) At least not in newer versions ... Starting with 0.6.x (0.6.1 is out today -- I recommend installing that), you can name a system anything arbitrarily. If you know the MAC address, it helps to use that so we can assign a specific kickstart -- you can also use the ip if you know that instead. Otherwise you can pick a specific profile and just assign that to be the "default" for all PXE booting systems. You can also use it to create a PXE menu, so you can walk up to a machine while it's booting, type "menu", and pick what you want. > > Currently we have our dhcp server on an install network dump anyone > that netboots on that network to a default pxelinux menu. That > default menu allows a person to chose an os to install. Chosing one > kicks them into one of several different default kickstart files. If > I could get kickstart to ask for static network information like it > would if I was installing from CD, I would be done. I should have read further before typing. Yeah, we can do that :) The MAC address way of assigning the static info works really really well though, if you can start doing it. I like kickstarts to be fully interactive because I can start them and not have to baby-sit. Plus mapping a system to a new profile is as easy as: cobbler edit --name= --profile= --Michael > > > On 8/30/07, *Riggs, Ben* > > wrote: > > Check out http://cobbler.et.redhat.com It's made our life here > tons easier. > > Ben > > Kit Stube wrote: > > Is there an option that will allow me to automate everything in > > kickstart except just the network portion so that I can have the > same > > kickstart config for a bunch of clustered hosts but input the > network > > information myself? > > > > The reason I ask is because I would like to automate a lot of other > > stuff in %post or on first boot but need networking to be set up > > properly by the time that stuff runs or it will screw all of > that up. > > > > Thanks in advance > > Kit Stube > > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From kstube at gmail.com Thu Aug 30 22:14:36 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 17:14:36 -0500 Subject: static network configuration ask In-Reply-To: <46D73CE4.5030805@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> Message-ID: <695c4edd0708301514ic51b8c7g88fa3f3b92b6782b@mail.gmail.com> Thanks Ed. I'll try that and report back. It's still a little more complicated than I would have liked as this will be used by people other than myself. It really simplifies all the scripting I was doing in %post to force it to work despite the difficulties. :-) It would be much slicker if there were an "ask" option to the network or if it didn't default to dhcp :-( On 8/30/07, Ed Brown wrote: > > > boot: ip=blah gateway=blah netmask=blah dns=blah > > This is not quite complete. And actually, we make a custom boot cd, > with multiple targets that contain the common parameters, so all > that's needed at the boot prompt is target, ip and gateway. You'll > probably need at least this: > > boot: linux ks=http://url_of_kickstart_file ip=... gateway=... > netmask=... dns=... > > -Ed > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From ebrown at lanl.gov Thu Aug 30 22:30:52 2007 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 30 Aug 2007 16:30:52 -0600 Subject: static network configuration ask In-Reply-To: <695c4edd0708301514ic51b8c7g88fa3f3b92b6782b@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301514ic51b8c7g88fa3f3b92b6782b@mail.gmail.com> Message-ID: <46D7451C.50909@lanl.gov> Kit Stube wrote: > Thanks Ed. I'll try that and report back. It's still a little more > complicated than I would have liked as this will be used by people other > than myself. It really simplifies all the scripting I was doing in %post > to force it to work despite the difficulties. :-) > > It would be much slicker if there were an "ask" option to the network or if > it didn't default to dhcp :-( Inputting the parameters at the boot prompt prevents defaulting to dhcp initially. Leaving them out means defaulting to dhcp attempts, but then you WILL get "ask"ed. The problem there is that I don't think those answers are properly turned into final network configurations (ifcfg-ethN, /etc/sysconfig/network, /etc/hosts), without some %pre/%post help. -Ed From kstube at gmail.com Thu Aug 30 22:33:32 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 17:33:32 -0500 Subject: static network configuration ask In-Reply-To: <46D740A5.2050707@redhat.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D730FC.8040007@umn.edu> <695c4edd0708301438j5adbc83evcf1bfa6964d6ff17@mail.gmail.com> <46D740A5.2050707@redhat.com> Message-ID: <695c4edd0708301533t60eaef59r80abed944271336b@mail.gmail.com> As long as we don't have to enter a mac address it sounds good to me. The problem with mac addresses is that our servers boot so fast it's hard to catch the config menu as it flies by. We're normally looking at them through an ASM or IpKVM of some type which just makes it harder..... Our build process is already obnoxious enough already. I'll look into that further after I get this first batch of servers delivered to the poor folks who are waiting on me right now. :-( Thanks. Kit On 8/30/07, Michael DeHaan wrote: > > Kit Stube wrote: > > Cobbler looks cool. Kinda like a front end for pxelinux and isc > > dhcpd? The problem we have is we don't want to have to look up the > > mac address for a host in order to configure it. Cobbler seems to > > rely on that. > > Nah, you don't :) At least not in newer versions ... > > Starting with 0.6.x (0.6.1 is out today -- I recommend installing that), > you can name a system anything arbitrarily. If you know the MAC > address, it helps to use that so we can assign a specific kickstart -- > you can also use the ip if you know that instead. Otherwise you can > pick a specific profile and just assign that to be the "default" for all > PXE booting systems. You can also use it to create a PXE menu, so you > can walk up to a machine while it's booting, type "menu", and pick what > you want. > > > > > Currently we have our dhcp server on an install network dump anyone > > that netboots on that network to a default pxelinux menu. That > > default menu allows a person to chose an os to install. Chosing one > > kicks them into one of several different default kickstart files. If > > I could get kickstart to ask for static network information like it > > would if I was installing from CD, I would be done. > > I should have read further before typing. Yeah, we can do that :) > > The MAC address way of assigning the static info works really really > well though, if you can start doing it. I like kickstarts to be fully > interactive because I can start them and not have to baby-sit. Plus > mapping a system to a new profile is as easy as: > > cobbler edit --name= --profile= > > --Michael > > > > > > > > On 8/30/07, *Riggs, Ben* > > > wrote: > > > > Check out http://cobbler.et.redhat.com It's made our life here > > tons easier. > > > > Ben > > > > Kit Stube wrote: > > > Is there an option that will allow me to automate everything in > > > kickstart except just the network portion so that I can have the > > same > > > kickstart config for a bunch of clustered hosts but input the > > network > > > information myself? > > > > > > The reason I ask is because I would like to automate a lot of > other > > > stuff in %post or on first boot but need networking to be set up > > > properly by the time that stuff runs or it will screw all of > > that up. > > > > > > Thanks in advance > > > Kit Stube > > > > > > > > > > > > ------------------------------------------------------------------------ > > > > > > _______________________________________________ > > > Kickstart-list mailing list > > > Kickstart-list at redhat.com > > > https://www.redhat.com/mailman/listinfo/kickstart-list > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From kstube at gmail.com Thu Aug 30 22:39:20 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 17:39:20 -0500 Subject: static network configuration ask In-Reply-To: <46D73CE4.5030805@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> Message-ID: <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> That worked for the most part. Ed, your a genious! The only problem now is that it didn't set the hostname. Is there some way to use something like $hostname in post so I can set all that up? Perhaps that was what Pablo was trying to tell me? I don't understand how to process /proc/commandline. Does anyone have an example? I would be greatly appreciative. Kit On 8/30/07, Ed Brown wrote: > > > boot: ip=blah gateway=blah netmask=blah dns=blah > > This is not quite complete. And actually, we make a custom boot cd, > with multiple targets that contain the common parameters, so all > that's needed at the boot prompt is target, ip and gateway. You'll > probably need at least this: > > boot: linux ks=http://url_of_kickstart_file ip=... gateway=... > netmask=... dns=... > > -Ed > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From ebrown at lanl.gov Thu Aug 30 22:44:23 2007 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 30 Aug 2007 16:44:23 -0600 Subject: static network configuration ask In-Reply-To: <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> Message-ID: <46D74847.3090609@lanl.gov> hostname won't be set during the %post scriptlet without some help, but it should be set on reboot just fine. To set it for the sake of commands in %post (for example, if you need to run cfengine): hostname `grep HOSTNAME /etc/sysconfig/network |cut -d'=' -f2` But as the above implies, it should already be configured on reboot. -Ed Kit Stube wrote: > That worked for the most part. Ed, your a genious! > > The only problem now is that it didn't set the hostname. Is there some way > to use something like $hostname in post so I can set all that up? > Perhaps that was what Pablo was trying to tell me? I don't understand how > to process /proc/commandline. Does anyone have an example? > I would be greatly appreciative. > > Kit > > > > On 8/30/07, Ed Brown wrote: >>> boot: ip=blah gateway=blah netmask=blah dns=blah >> This is not quite complete. And actually, we make a custom boot cd, >> with multiple targets that contain the common parameters, so all >> that's needed at the boot prompt is target, ip and gateway. You'll >> probably need at least this: >> >> boot: linux ks=http://url_of_kickstart_file ip=... gateway=... >> netmask=... dns=... >> >> -Ed >> >> _______________________________________________ >> Kickstart-list mailing list >> Kickstart-list at redhat.com >> https://www.redhat.com/mailman/listinfo/kickstart-list >> > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From Pablo.Iranzo at uv.es Thu Aug 30 22:42:29 2007 From: Pablo.Iranzo at uv.es (=?UTF-8?Q?Pablo_Iranzo_G=C3=B3mez?=) Date: Fri, 31 Aug 2007 00:42:29 +0200 (CEST) Subject: static network configuration ask In-Reply-To: <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> Message-ID: Hi On Thu, 30 Aug 2007, Kit Stube wrote: > That worked for the most part. Ed, your a genious! > > The only problem now is that it didn't set the hostname. Is there some way > to use something like $hostname in post so I can set all that up? > Perhaps that was what Pablo was trying to tell me? I don't understand how > to process /proc/commandline. Does anyone have an example? > I would be greatly appreciative. If you put on commandline ip=123.123.123.123 netmask=234.234.234.234 hostname=host123 On %post you could do: HOST=`cat /proc/cmdline|tr " " "\n"|grep hostname|cut -d "=" -f 2` And when creating the custom /etc/sysconfig/network, fill in the value from $HOST Regards Pablo From ebrown at lanl.gov Thu Aug 30 22:50:47 2007 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 30 Aug 2007 16:50:47 -0600 Subject: static network configuration ask In-Reply-To: <46D74847.3090609@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> Message-ID: <46D749C7.9020304@lanl.gov> To say this another way, it is not necessary to enter hostname as a boot parameter, assuming the ip address can be resolved. Ed Brown wrote: > hostname won't be set during the %post scriptlet without some help, but > it should be set on reboot just fine. To set it for the sake of > commands in %post (for example, if you need to run cfengine): > > hostname `grep HOSTNAME /etc/sysconfig/network |cut -d'=' -f2` > > But as the above implies, it should already be configured on reboot. > > -Ed > > > > Kit Stube wrote: >> That worked for the most part. Ed, your a genious! >> >> The only problem now is that it didn't set the hostname. Is there >> some way >> to use something like $hostname in post so I can set all that up? >> Perhaps that was what Pablo was trying to tell me? I don't >> understand how >> to process /proc/commandline. Does anyone have an example? >> I would be greatly appreciative. >> >> Kit >> >> >> >> On 8/30/07, Ed Brown wrote: >>>> boot: ip=blah gateway=blah netmask=blah dns=blah >>> This is not quite complete. And actually, we make a custom boot cd, >>> with multiple targets that contain the common parameters, so all >>> that's needed at the boot prompt is target, ip and gateway. You'll >>> probably need at least this: >>> >>> boot: linux ks=http://url_of_kickstart_file ip=... gateway=... >>> netmask=... dns=... >>> >>> -Ed >>> >>> _______________________________________________ >>> Kickstart-list mailing list >>> Kickstart-list at redhat.com >>> https://www.redhat.com/mailman/listinfo/kickstart-list >>> >> >> >> ------------------------------------------------------------------------ >> >> _______________________________________________ >> Kickstart-list mailing list >> Kickstart-list at redhat.com >> https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From kstube at gmail.com Thu Aug 30 23:03:52 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 18:03:52 -0500 Subject: static network configuration ask In-Reply-To: <46D74847.3090609@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> Message-ID: <695c4edd0708301603m5b38cee1x9b41d42b19d4c402@mail.gmail.com> For whatever reason, the hostname is not getting put in /etc/sysconfig/network. I do not see it after a reboot. Are you sure this is supposed to happen? On 8/30/07, Ed Brown wrote: > > hostname won't be set during the %post scriptlet without some help, > but it should be set on reboot just fine. To set it for the sake of > commands in %post (for example, if you need to run cfengine): > > hostname `grep HOSTNAME /etc/sysconfig/network |cut -d'=' -f2` > > But as the above implies, it should already be configured on reboot. > > -Ed > > > > Kit Stube wrote: > > That worked for the most part. Ed, your a genious! > > > > The only problem now is that it didn't set the hostname. Is there some > way > > to use something like $hostname in post so I can set all that up? > > Perhaps that was what Pablo was trying to tell me? I don't understand > how > > to process /proc/commandline. Does anyone have an example? > > I would be greatly appreciative. > > > > Kit > > > > > > > > On 8/30/07, Ed Brown wrote: > >>> boot: ip=blah gateway=blah netmask=blah dns=blah > >> This is not quite complete. And actually, we make a custom boot cd, > >> with multiple targets that contain the common parameters, so all > >> that's needed at the boot prompt is target, ip and gateway. You'll > >> probably need at least this: > >> > >> boot: linux ks=http://url_of_kickstart_file ip=... gateway=... > >> netmask=... dns=... > >> > >> -Ed > >> > >> _______________________________________________ > >> Kickstart-list mailing list > >> Kickstart-list at redhat.com > >> https://www.redhat.com/mailman/listinfo/kickstart-list > >> > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From kstube at gmail.com Thu Aug 30 23:05:11 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 18:05:11 -0500 Subject: static network configuration ask In-Reply-To: References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> Message-ID: <695c4edd0708301605hc910bd7j9ef0af0eae390c35@mail.gmail.com> I see what you mean now. I know what to do with this information at this point. Can I do the same thing in %pre? Kit On 8/30/07, Pablo Iranzo G?mez wrote: > > Hi > > On Thu, 30 Aug 2007, Kit Stube wrote: > > > That worked for the most part. Ed, your a genious! > > > > The only problem now is that it didn't set the hostname. Is there some > way > > to use something like $hostname in post so I can set all that up? > > Perhaps that was what Pablo was trying to tell me? I don't understand > how > > to process /proc/commandline. Does anyone have an example? > > I would be greatly appreciative. > > If you put on commandline ip=123.123.123.123 > netmask=234.234.234.234 hostname=host123 > > On %post you could do: > > HOST=`cat /proc/cmdline|tr " " "\n"|grep hostname|cut -d "=" > -f 2` > > And when creating the custom /etc/sysconfig/network, fill in the > value from $HOST > > Regards > Pablo > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From kstube at gmail.com Thu Aug 30 23:09:49 2007 From: kstube at gmail.com (Kit Stube) Date: Thu, 30 Aug 2007 18:09:49 -0500 Subject: static network configuration ask In-Reply-To: <46D749C7.9020304@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> <46D749C7.9020304@lanl.gov> Message-ID: <695c4edd0708301609h1294e205g24078adf2574a294@mail.gmail.com> Come to think of it, I have seen the behavior where the system appears to have done a reverse lookup of its self and set it's own hostname. :-) That's perfect. At what point during the install process does it do that? Thanks again for all of the great info everyone. Kit On 8/30/07, Ed Brown wrote: > > To say this another way, it is not necessary to enter hostname as a > boot parameter, assuming the ip address can be resolved. > > Ed Brown wrote: > > hostname won't be set during the %post scriptlet without some help, but > > it should be set on reboot just fine. To set it for the sake of > > commands in %post (for example, if you need to run cfengine): > > > > hostname `grep HOSTNAME /etc/sysconfig/network |cut -d'=' -f2` > > > > But as the above implies, it should already be configured on reboot. > > > > -Ed > > > > > > > > Kit Stube wrote: > >> That worked for the most part. Ed, your a genious! > >> > >> The only problem now is that it didn't set the hostname. Is there > >> some way > >> to use something like $hostname in post so I can set all that up? > >> Perhaps that was what Pablo was trying to tell me? I don't > >> understand how > >> to process /proc/commandline. Does anyone have an example? > >> I would be greatly appreciative. > >> > >> Kit > >> > >> > >> > >> On 8/30/07, Ed Brown wrote: > >>>> boot: ip=blah gateway=blah netmask=blah dns=blah > >>> This is not quite complete. And actually, we make a custom boot cd, > >>> with multiple targets that contain the common parameters, so all > >>> that's needed at the boot prompt is target, ip and gateway. You'll > >>> probably need at least this: > >>> > >>> boot: linux ks=http://url_of_kickstart_file ip=... gateway=... > >>> netmask=... dns=... > >>> > >>> -Ed > >>> > >>> _______________________________________________ > >>> Kickstart-list mailing list > >>> Kickstart-list at redhat.com > >>> https://www.redhat.com/mailman/listinfo/kickstart-list > >>> > >> > >> > >> > ------------------------------------------------------------------------ > >> > >> _______________________________________________ > >> Kickstart-list mailing list > >> Kickstart-list at redhat.com > >> https://www.redhat.com/mailman/listinfo/kickstart-list > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From Pablo.Iranzo at uv.es Thu Aug 30 23:07:36 2007 From: Pablo.Iranzo at uv.es (=?UTF-8?Q?Pablo_Iranzo_G=C3=B3mez?=) Date: Fri, 31 Aug 2007 01:07:36 +0200 (CEST) Subject: static network configuration ask In-Reply-To: <695c4edd0708301605hc910bd7j9ef0af0eae390c35@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <695c4edd0708301605hc910bd7j9ef0af0eae390c35@mail.gmail.com> Message-ID: Hi On Thu, 30 Aug 2007, Kit Stube wrote: > I see what you mean now. I know what to do with this information at this > point. Can I do the same thing in %pre? ?Where are you going to put the configuration files if you still don't have any filesystem? ;) Is it possible to do something at that point, but should be kept in /tmp until filesystems are created and you can copy that to the installed system... so I don't see the point in doing it at %pre instead of doing at %post Regards Pablo From ebrown at lanl.gov Thu Aug 30 23:26:18 2007 From: ebrown at lanl.gov (Ed Brown) Date: Thu, 30 Aug 2007 17:26:18 -0600 Subject: static network configuration ask In-Reply-To: <695c4edd0708301609h1294e205g24078adf2574a294@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> <46D749C7.9020304@lanl.gov> <695c4edd0708301609h1294e205g24078adf2574a294@mail.gmail.com> Message-ID: <46D7521A.2020207@lanl.gov> It does the reverse lookup even before it gets the kickstart file, and flashes a screen telling you it's doing that, though often that goes by so fast you don't see it. You can also watch for it on alt-F3 (or 4?) though. I don't know when it writes /etc/sysconfig/network, but it's there in %post without my doing anything. You do have have the 'network --bootproto=static' line in the kickstart file? -ed Kit Stube wrote: > Come to think of it, I have seen the behavior where the system appears to > have done a reverse lookup of its self and set it's own hostname. :-) > That's perfect. At what point during the install process does it do that? > > Thanks again for all of the great info everyone. > > Kit > > > On 8/30/07, Ed Brown wrote: >> To say this another way, it is not necessary to enter hostname as a >> boot parameter, assuming the ip address can be resolved. >> >> Ed Brown wrote: >>> hostname won't be set during the %post scriptlet without some help, but >>> it should be set on reboot just fine. To set it for the sake of >>> commands in %post (for example, if you need to run cfengine): >>> >>> hostname `grep HOSTNAME /etc/sysconfig/network |cut -d'=' -f2` >>> >>> But as the above implies, it should already be configured on reboot. >>> >>> -Ed >>> >>> >>> >>> Kit Stube wrote: >>>> That worked for the most part. Ed, your a genious! >>>> >>>> The only problem now is that it didn't set the hostname. Is there >>>> some way >>>> to use something like $hostname in post so I can set all that up? >>>> Perhaps that was what Pablo was trying to tell me? I don't >>>> understand how >>>> to process /proc/commandline. Does anyone have an example? >>>> I would be greatly appreciative. >>>> >>>> Kit >>>> >>>> >>>> >>>> On 8/30/07, Ed Brown wrote: >>>>>> boot: ip=blah gateway=blah netmask=blah dns=blah >>>>> This is not quite complete. And actually, we make a custom boot cd, >>>>> with multiple targets that contain the common parameters, so all >>>>> that's needed at the boot prompt is target, ip and gateway. You'll >>>>> probably need at least this: >>>>> >>>>> boot: linux ks=http://url_of_kickstart_file ip=... gateway=... >>>>> netmask=... dns=... >>>>> >>>>> -Ed >>>>> >>>>> _______________________________________________ >>>>> Kickstart-list mailing list >>>>> Kickstart-list at redhat.com >>>>> https://www.redhat.com/mailman/listinfo/kickstart-list >>>>> >>>> >>>> >> ------------------------------------------------------------------------ >>>> _______________________________________________ >>>> Kickstart-list mailing list >>>> Kickstart-list at redhat.com >>>> https://www.redhat.com/mailman/listinfo/kickstart-list >>> _______________________________________________ >>> Kickstart-list mailing list >>> Kickstart-list at redhat.com >>> https://www.redhat.com/mailman/listinfo/kickstart-list >> _______________________________________________ >> Kickstart-list mailing list >> Kickstart-list at redhat.com >> https://www.redhat.com/mailman/listinfo/kickstart-list >> > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From dan at half-asleep.com Fri Aug 31 02:13:55 2007 From: dan at half-asleep.com (Daniel Segall) Date: Thu, 30 Aug 2007 22:13:55 -0400 Subject: Using Kickstart to deploy Debian servers? Message-ID: <46D77963.7020508@half-asleep.com> I feel dirty even asking this, but here it goes. I have been tasked with the unlucky job of setting up a deployment solution for Debian servers. I have an existing Kickstart server that I call via altiris (we mostly use blades). Does anyone know if there is a relatively easy way to create a kickstart config for Debian "etch"? I really don't want to setup FAI if I don't have to. Thanks, -Dan From kstube at gmail.com Fri Aug 31 14:36:23 2007 From: kstube at gmail.com (Kit Stube) Date: Fri, 31 Aug 2007 09:36:23 -0500 Subject: static network configuration ask In-Reply-To: References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <695c4edd0708301605hc910bd7j9ef0af0eae390c35@mail.gmail.com> Message-ID: <695c4edd0708310736m11ceb2aeu5e5816b1709abb0e@mail.gmail.com> I really just wanted to grab variables that I could use in the main config section. ie: system architecture. Is there a list of global variables that are present during the config? Kit On 8/30/07, Pablo Iranzo G?mez wrote: > > Hi > > On Thu, 30 Aug 2007, Kit Stube wrote: > > > I see what you mean now. I know what to do with this information at > this > > point. Can I do the same thing in %pre? > > ?Where are you going to put the configuration files if you still > don't have any filesystem? ;) > > Is it possible to do something at that point, but should be kept > in /tmp until filesystems are created and you can copy that to the > installed system... so I don't see the point in doing it at %pre instead > of doing at %post > > Regards > Pablo > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From mdehaan at redhat.com Fri Aug 31 15:33:10 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 31 Aug 2007 11:33:10 -0400 Subject: static network configuration ask In-Reply-To: <695c4edd0708301533t60eaef59r80abed944271336b@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D730FC.8040007@umn.edu> <695c4edd0708301438j5adbc83evcf1bfa6964d6ff17@mail.gmail.com> <46D740A5.2050707@redhat.com> <695c4edd0708301533t60eaef59r80abed944271336b@mail.gmail.com> Message-ID: <46D834B6.7030505@redhat.com> Kit Stube wrote: > As long as we don't have to enter a mac address it sounds good to > me. The problem with mac addresses is that our servers boot so fast > it's hard to catch the config menu as it flies by. We're normally > looking at them through an ASM or IpKVM of some type which just makes > it harder..... Our build process is already obnoxious enough > already. I'll look into that further after I get this first batch > of servers delivered to the poor folks who are waiting on me right > now. :-( Thanks. > Cool, FWIW you can usually get the MAC from the BIOS ... or be really clever and make a default kickstart profile that does a minimal install with the sole purpose of emailing you the mac addresses for easier future maintaince :) Lots of fun can be had... > Kit > > > > On 8/30/07, *Michael DeHaan* > wrote: > > Kit Stube wrote: > > Cobbler looks cool. Kinda like a front end for pxelinux and isc > > dhcpd? The problem we have is we don't want to have to look up the > > mac address for a host in order to configure it. Cobbler seems to > > rely on that. > > Nah, you don't :) At least not in newer versions ... > > Starting with 0.6.x (0.6.1 is out today -- I recommend installing > that), > you can name a system anything arbitrarily. If you know the MAC > address, it helps to use that so we can assign a specific kickstart -- > you can also use the ip if you know that instead. Otherwise you can > pick a specific profile and just assign that to be the "default" > for all > PXE booting systems. You can also use it to create a PXE menu, so you > can walk up to a machine while it's booting, type "menu", and pick > what > you want. > > > > > Currently we have our dhcp server on an install network dump anyone > > that netboots on that network to a default pxelinux menu. That > > default menu allows a person to chose an os to install. Chosing one > > kicks them into one of several different default kickstart > files. If > > I could get kickstart to ask for static network information like it > > would if I was installing from CD, I would be done. > > I should have read further before typing. Yeah, we can do that :) > > The MAC address way of assigning the static info works really really > well though, if you can start doing it. I like kickstarts to be fully > interactive because I can start them and not have to baby-sit. Plus > mapping a system to a new profile is as easy as: > > cobbler edit --name= --profile= > > --Michael > > > > > > > > On 8/30/07, *Riggs, Ben* >> > > wrote: > > > > Check out http://cobbler.et.redhat.com It's made our life here > > tons easier. > > > > Ben > > > > Kit Stube wrote: > > > Is there an option that will allow me to automate > everything in > > > kickstart except just the network portion so that I can > have the > > same > > > kickstart config for a bunch of clustered hosts but input the > > network > > > information myself? > > > > > > The reason I ask is because I would like to automate a lot > of other > > > stuff in %post or on first boot but need networking to be > set up > > > properly by the time that stuff runs or it will screw all of > > that up. > > > > > > Thanks in advance > > > Kit Stube > > > > > > > > > > > > ------------------------------------------------------------------------ > > > > > > _______________________________________________ > > > Kickstart-list mailing list > > > Kickstart-list at redhat.com > > > > > > https://www.redhat.com/mailman/listinfo/kickstart-list > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > > > https://www.redhat.com/mailman/listinfo/kickstart-list > > > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > > > ------------------------------------------------------------------------ > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list From mdehaan at redhat.com Fri Aug 31 15:38:07 2007 From: mdehaan at redhat.com (Michael DeHaan) Date: Fri, 31 Aug 2007 11:38:07 -0400 Subject: Using Kickstart to deploy Debian servers? In-Reply-To: <46D77963.7020508@half-asleep.com> References: <46D77963.7020508@half-asleep.com> Message-ID: <46D835DF.7090809@redhat.com> Daniel Segall wrote: > I feel dirty even asking this, but here it goes. I have been tasked > with the unlucky job of setting up a deployment solution for Debian > servers. I have an existing Kickstart server that I call via altiris > (we mostly use blades). Does anyone know if there is a relatively easy > way to create a kickstart config for Debian "etch"? I really don't > want to setup FAI if I don't have to. > > Thanks, > -Dan > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list Debian has it's own system of answer files that are different from kickstart. The kernel options change from "ks=blah" to "auto=true url=blah". I'm told (user patch, haven't tested) it actually works with "cobbler distro add" if you specify "--breed=debian", though you could just as easily do this all yourself. You may also want to look at FAI -- http://www.informatik.uni-koeln.de/fai/ But yes, wrong list :) --Michael From Darrick.Waller at baesystems.com Fri Aug 31 16:10:59 2007 From: Darrick.Waller at baesystems.com (Waller, Darrick J (US SSA)) Date: Fri, 31 Aug 2007 11:10:59 -0500 Subject: rhel5 always comes up with dhcp Message-ID: <04799F26A23174449975FB1913E3A02F155C70@gldms20030.goldlnk.rootlnka.net> I'm using the following script to kickstart a rhel5 workstation. When the machine comes up, the /etc/sysconfig/network-scripts/ifcfg-eth0 file has been moved to /etc/sysconfig/network-scripts/ifcfg-eth0.bak, and the new file specifies the bootproto as dhcp. Any ideas would be much appreciated. Kickstart.cfg: # Embed pxelinux.cfg file here #PXE default linux #PXE label linux #PXE kernel vmlinuz-REDHATEL5_x64 #PXE append ksdevice=link ip= netmask= gateway= dns=1 load_ramdisk=1 initrd=initrd.img-REDHATEL5_x64 network ks=nfs::/export/Corp01/install/kickstart/hosts/^HOSTNAME^ # Install Red Hat Linux install # Installation Number key # System Language lang en_US.UTF-8 # Language Modules to Install langsupport --default en_US en_US # System Keyboard keyboard us # System Mouse mouse --emulthree # Skip XWindows configuration skipx # Setup Networking network --device eth0 --bootproto static --ip ^IPADDR^ --netmask --gateway ^GATEWAY^ --nameserver --hostname ^HOSTNAME^ --noipv6 # Set root password rootpw --iscrypted # Disable Firewall firewall --disabled # Disable SELinux selinux --disabled # Enable Security Features authconfig --enableshadow --enablemd5 # Set timezone timezone --utc America/Chicago # Create a Bootloader bootloader --location=mbr --driveorder=sda --append="rhgb quiet" --md5pass= # Reboot after install reboot # Use text mode to install text # Use NFS installation media nfs --server --dir /export/Corp01/install/kickstart/REDHATEL5_Desktop_x64 # Clear the MBR zerombr # Clear all partitions clearpart --linux --drives=sda # Create Partitions part / --fstype ext3 --size=8192 part swap --size=1024 part /var --fstype ext3 --size=100 --grow %packages @admin-tools @base @core @development-libs @development-tools @editors @legacy-software-development @printing tcl.i386 kernel-devel nautilus-open-terminal xinetd -kdenetwork -NetworkManager -kdenetwork-devel -dhclient -NetworkManager-gnome -dhcdbd %post ####################################################################### # Setup Variables ####################################################################### DOMAIN=^DOMAIN^ HOSTNAME=^HOSTNAME^ IPADDR=^IPADDR^ GATEWAY=^GATEWAY^ KICKSTART_SERVER= KICKSTART_DIR=/export/Corp01/install/kickstart CFENGINE_SERVER= CFENGINE_MNT_DIR=/spool/cfengine CFENGINE_INST_DIR=/export/JS/cfengine/local LOG_DIR=/var/log/kickstart LOG_FILE=${LOG_DIR}/kickstart_install.log KICK_DIR=/mnt SCRIPTS_DIR=${KICK_DIR}/scripts FILES_DIR=${KICK_DIR}/files export DOMAIN HOSTNAME IPADDR GATEWAY KICKSTART_SERVER CFENGINE_SERVER CFENGINE_MNT_DIR CFENGINE_INST_DIR LOG_DIR LOG_FILE KICK_DIR SCRIPTS_DIR FILES_DIR ######################################################################## mkdir ${LOG_DIR} # Mount up the kickstart server ######################################################################## echo "Mounting ${KICKSTART_SERVER}:${KICKSTART_DIR} -> ${KICK_DIR}" >> ${LOG_FILE} mount -o ro,nolock ${KICKSTART_SERVER}:${KICKSTART_DIR} ${KICK_DIR} >> ${LOG_FILE} 2>&1 ######################################################################## ######################################################################## # Select which services you want on/off at boot ######################################################################## chkconfig sendmail off chkconfig iptables off chkconfig ntpd on chkconfig autofs on chkconfig cups on chkconfig telnet off chkconfig rlogin on chkconfig rsh on chkconfig rexec on chkconfig rsync on chkconfig snmpd off chkconfig snmptrapd off chkconfig yumupdatesd off ######################################################################## ######################################################################## # Select which options you want setup ######################################################################## ${SCRIPTS_DIR}/Network_setup.sh ${SCRIPTS_DIR}/NTP_setup.sh ${SCRIPTS_DIR}/CFengine_install.sh ${SCRIPTS_DIR}/NIS_setup.sh ${SCRIPTS_DIR}/xorg_install.sh ${SCRIPTS_DIR}/Autofs_setup.sh ######################################################################## echo "unmounting ${KICK_DIR}" >> ${LOG_FILE} umount ${KICK_DIR} >> ${LOG_FILE} 2>&1 -------------- next part -------------- An HTML attachment was scrubbed... URL: From srobson at cadence.com Fri Aug 31 16:27:22 2007 From: srobson at cadence.com (Steve Robson) Date: Fri, 31 Aug 2007 17:27:22 +0100 Subject: rhel5 always comes up with dhcp In-Reply-To: <20070831161145.052F4732F5@hormel.redhat.com> References: <20070831161145.052F4732F5@hormel.redhat.com> Message-ID: <46D8416A.7020108@cadence.com> > From: "Waller, Darrick J \(US SSA\)" > Date: Fri, 31 Aug 2007 11:10:59 -0500 > > I?m using the following script to kickstart a rhel5 workstation. When > the machine comes up, the /etc/sysconfig/network-scripts/ifcfg-eth0 file > has been moved to /etc/sysconfig/network-scripts/ifcfg-eth0.bak, and the > new file specifies the bootproto as dhcp. Any ideas would be much > appreciated. This is probably being done by your Network_setup.sh script. > KICK_DIR=/mnt > SCRIPTS_DIR=${KICK_DIR}/scripts > ${SCRIPTS_DIR}/Network_setup.sh -- Regards, Steve IT Support - UNIX/Linux Cadence Design Systems Bagshot Road Bracknell BERKSHIRE RG12 0PH UK From kstube at gmail.com Fri Aug 31 17:56:02 2007 From: kstube at gmail.com (Kit Stube) Date: Fri, 31 Aug 2007 12:56:02 -0500 Subject: static network configuration ask In-Reply-To: <46D7521A.2020207@lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> <46D749C7.9020304@lanl.gov> <695c4edd0708301609h1294e205g24078adf2574a294@mail.gmail.com> <46D7521A.2020207@lanl.gov> Message-ID: <695c4edd0708311056y527a6d47i3811db25eb2df0b9@mail.gmail.com> You are correct. I have bootproto=static. At this point, this is what I have: text install key dfsdfgsdggsdg lang en_US.UTF-8 langsupport --default en_US.UTF-8 en_US.UTF-8 keyboard us mouse genericwheelps/2 --device psaux skipx nfs --server 192.168.49.26 --dir /vol/vol1/vmimages/kickstart/RH5-server-i386 network --bootproto static --device eth0 --nameserver 192.168.49.107--netmask 255.255.255.0 rootpw --iscrypted $1$rGjAQdLQm$LNgsdfsVF/TjfszJB6T firewall --disabled authconfig --enableshadow --enableldap --enableldapauth --ldapserver ldaps.ourdomain.com --ldapbasedn dc=ourdomain,dc=com selinux --disabled timezone --utc America/Chicago bootloader --location=mbr repo --name=repo --baseurl=http://repo.ourdomain.com/linux/4/$ARCH ### This part is still broken. reboot %packages %include /mnt/config/includes/RHEL5_packages firstboot %pre # nfs mount the config directory so that we can use %include # in this file to make things cleaner. mkdir /mnt/config mount -t nfs 192.168.49.26:/vol/vol1/vmimages/kickstart/configs /mnt/config %post --nochroot cp /mnt/config/includes/* /mnt/sysimage/root chmod +x /mnt/sysimage/root/system_setup.sh #system_setup.sh used to live here in %post and I will be moving it back in with a %include rather than copying the file to the host if that will #work. For some reason, my hostname still does not get set up in /etc/sysconfig/network even though I pass hostname, ip, and Kit On 8/30/07, Ed Brown wrote: > > It does the reverse lookup even before it gets the kickstart file, > and flashes a screen telling you it's doing that, though often that > goes by so fast you don't see it. You can also watch for it on alt-F3 > (or 4?) though. I don't know when it writes /etc/sysconfig/network, > but it's there in %post without my doing anything. You do have have > the 'network --bootproto=static' line in the kickstart file? > > -ed > > > Kit Stube wrote: > > Come to think of it, I have seen the behavior where the system appears > to > > have done a reverse lookup of its self and set it's own hostname. :-) > > That's perfect. At what point during the install process does it do > that? > > > > Thanks again for all of the great info everyone. > > > > Kit > > > > > > On 8/30/07, Ed Brown wrote: > >> To say this another way, it is not necessary to enter hostname as a > >> boot parameter, assuming the ip address can be resolved. > >> > >> Ed Brown wrote: > >>> hostname won't be set during the %post scriptlet without some help, > but > >>> it should be set on reboot just fine. To set it for the sake of > >>> commands in %post (for example, if you need to run cfengine): > >>> > >>> hostname `grep HOSTNAME /etc/sysconfig/network |cut -d'=' -f2` > >>> > >>> But as the above implies, it should already be configured on reboot. > >>> > >>> -Ed > >>> > >>> > >>> > >>> Kit Stube wrote: > >>>> That worked for the most part. Ed, your a genious! > >>>> > >>>> The only problem now is that it didn't set the hostname. Is there > >>>> some way > >>>> to use something like $hostname in post so I can set all that up? > >>>> Perhaps that was what Pablo was trying to tell me? I don't > >>>> understand how > >>>> to process /proc/commandline. Does anyone have an example? > >>>> I would be greatly appreciative. > >>>> > >>>> Kit > >>>> > >>>> > >>>> > >>>> On 8/30/07, Ed Brown wrote: > >>>>>> boot: ip=blah gateway=blah netmask=blah dns=blah > >>>>> This is not quite complete. And actually, we make a custom boot cd, > >>>>> with multiple targets that contain the common parameters, so all > >>>>> that's needed at the boot prompt is target, ip and gateway. You'll > >>>>> probably need at least this: > >>>>> > >>>>> boot: linux ks=http://url_of_kickstart_file ip=... gateway=... > >>>>> netmask=... dns=... > >>>>> > >>>>> -Ed > >>>>> > >>>>> _______________________________________________ > >>>>> Kickstart-list mailing list > >>>>> Kickstart-list at redhat.com > >>>>> https://www.redhat.com/mailman/listinfo/kickstart-list > >>>>> > >>>> > >>>> > >> > ------------------------------------------------------------------------ > >>>> _______________________________________________ > >>>> Kickstart-list mailing list > >>>> Kickstart-list at redhat.com > >>>> https://www.redhat.com/mailman/listinfo/kickstart-list > >>> _______________________________________________ > >>> Kickstart-list mailing list > >>> Kickstart-list at redhat.com > >>> https://www.redhat.com/mailman/listinfo/kickstart-list > >> _______________________________________________ > >> Kickstart-list mailing list > >> Kickstart-list at redhat.com > >> https://www.redhat.com/mailman/listinfo/kickstart-list > >> > > > > > > ------------------------------------------------------------------------ > > > > _______________________________________________ > > Kickstart-list mailing list > > Kickstart-list at redhat.com > > https://www.redhat.com/mailman/listinfo/kickstart-list > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: From dmc.fedora at filteredperception.org Fri Aug 31 18:30:57 2007 From: dmc.fedora at filteredperception.org (Douglas McClendon) Date: Fri, 31 Aug 2007 13:30:57 -0500 Subject: Using Kickstart to deploy Debian servers? In-Reply-To: <46D835DF.7090809@redhat.com> References: <46D77963.7020508@half-asleep.com> <46D835DF.7090809@redhat.com> Message-ID: <46D85E61.5080600@filteredperception.org> Michael DeHaan wrote: > Daniel Segall wrote: >> I feel dirty even asking this, but here it goes. I have been tasked >> with the unlucky job of setting up a deployment solution for Debian >> servers. I have an existing Kickstart server that I call via altiris >> (we mostly use blades). Does anyone know if there is a relatively easy >> way to create a kickstart config for Debian "etch"? I really don't >> want to setup FAI if I don't have to. >> >> Thanks, >> -Dan >> >> _______________________________________________ >> Kickstart-list mailing list >> Kickstart-list at redhat.com >> https://www.redhat.com/mailman/listinfo/kickstart-list > > Debian has it's own system of answer files that are different from > kickstart. Well, actually... ubuntu isn't debian, but... https://help.ubuntu.com/6.10/ubuntu/installation-guide/i386/automatic-install.html#kickstart (I have no personal experience with this. Most likely this is nowhere near 'relatively easy' to use with etch yet, but maybe in the future?) > The kernel options change from "ks=blah" to "auto=true url=blah". > > I'm told (user patch, haven't tested) it actually works with "cobbler > distro add" if you specify "--breed=debian", though you could just as > easily do this all yourself. > > You may also want to look at FAI -- http://www.informatik.uni-koeln.de/fai/ I think he specifically wanted to avoid that if possible. -dmc From ebrown at lanl.gov Fri Aug 31 18:56:19 2007 From: ebrown at lanl.gov (Edward F. Brown) Date: Fri, 31 Aug 2007 12:56:19 -0600 (MDT) Subject: static network configuration ask In-Reply-To: <695c4edd0708311056y527a6d47i3811db25eb2df0b9@mail.gmail.com> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> <46D749C7.9020304@lanl.gov> <695c4edd0708301609h1294e205g24078adf2574a294@mail.gmail.com> <46D7521A.2020207@lanl.gov> <695c4edd0708311056y527a6d47i3811db25eb2df0b9@mail.gmail.com> Message-ID: <52141.128.165.0.81.1188586579.squirrel@webmail.lanl.gov> On Fri, August 31, 2007 11:56 am, Kit Stube wrote: > At this point, this is what I have: > network --bootproto static --device eth0 --nameserver > 192.168.49.107--netmask > 255.255.255.0 If you aren't passing the nameserver('dns=') and netmask parameters at the boot prompt, this is probably why the hostname is not being configured. The reverse lookup happens before the kickstart file is read. Again, you can customize syslinux.cfg within boot.iso, if you use cd's to initiate the install, and have as little as just a target name and ip address to enter at the boot prompt. Advantages: absolutely no files to edit to kick a new system, no mac addresses to determine... Disadvantages: need to recreate cd with update level changes... -Ed From kstube at gmail.com Fri Aug 31 22:17:47 2007 From: kstube at gmail.com (Kit Stube) Date: Fri, 31 Aug 2007 17:17:47 -0500 Subject: static network configuration ask In-Reply-To: <52141.128.165.0.81.1188586579.squirrel@webmail.lanl.gov> References: <695c4edd0708301402s4a5433e2lb127fc4c063dceca@mail.gmail.com> <46D73A81.9050704@lanl.gov> <46D73CE4.5030805@lanl.gov> <695c4edd0708301539i69e06e56x3c41eea92ceb9116@mail.gmail.com> <46D74847.3090609@lanl.gov> <46D749C7.9020304@lanl.gov> <695c4edd0708301609h1294e205g24078adf2574a294@mail.gmail.com> <46D7521A.2020207@lanl.gov> <695c4edd0708311056y527a6d47i3811db25eb2df0b9@mail.gmail.com> <52141.128.165.0.81.1188586579.squirrel@webmail.lanl.gov> Message-ID: <695c4edd0708311517gb8eae12o1f2886d1136c426b@mail.gmail.com> After much messing around. I figured out a way to get it to go interactive for the network config. On top of that, it appears to persist once the host is rebooted. I pass it the option ip=192.168.50 and it kicks out to the network configurator every time! So far giving it a broken config is the only way I have found to do this. They really need an "ask" option to the network config stuff to make it go interactive during that portion of the setup. Thank you gentlemen for your help. I am very glad to have found the answer to this one. On 8/31/07, Edward F. Brown wrote: > > On Fri, August 31, 2007 11:56 am, Kit Stube wrote: > > At this point, this is what I have: > > > network --bootproto static --device eth0 --nameserver > > 192.168.49.107--netmask > > 255.255.255.0 > > If you aren't passing the nameserver('dns=') and netmask parameters at the > boot prompt, this is probably why the hostname is not being configured. > The reverse lookup happens before the kickstart file is read. > > Again, you can customize syslinux.cfg within boot.iso, if you use cd's to > initiate the install, and have as little as just a target name and ip > address to enter at the boot prompt. Advantages: absolutely no files to > edit to kick a new system, no mac addresses to determine... Disadvantages: > need to recreate cd with update level changes... > > -Ed > > _______________________________________________ > Kickstart-list mailing list > Kickstart-list at redhat.com > https://www.redhat.com/mailman/listinfo/kickstart-list > -------------- next part -------------- An HTML attachment was scrubbed... URL: