Bug 392021
Summary: | Impossible to use non-predefined static ip adresses with kickstart | ||
---|---|---|---|
Product: | [Fedora] Fedora | Reporter: | Olivier Galibert <galibert> |
Component: | anaconda | Assignee: | David Cantrell <dcantrell> |
Status: | CLOSED RAWHIDE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> |
Severity: | high | Docs Contact: | |
Priority: | low | ||
Version: | 10 | CC: | chris, cristian.ciupitu, deknuydt, felmasper, gustavo, herrold, jansen, k.georgiou, lance, marcus.moeller, matteo.faleschini, mglantz, mwest, ron |
Target Milestone: | --- | Keywords: | Reopened |
Target Release: | --- | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Fixed In Version: | Doc Type: | Bug Fix | |
Doc Text: | Story Points: | --- | |
Clone Of: | Environment: | ||
Last Closed: | 2009-02-21 05:03:01 UTC | Type: | --- |
Regression: | --- | Mount Type: | --- |
Documentation: | --- | CRM: | |
Verified Versions: | Category: | --- | |
oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
Cloudforms Team: | --- | Target Upstream Version: | |
Embargoed: |
Description
Olivier Galibert
2007-11-20 12:12:09 UTC
It also happily ignored a: network --bootproto=static --device=eth0 --ip=192.44.78.42 --netmask=255.255.255.0 --gateway=192.44.78.22 --nameserver=192.44.78.6 I've just added in ks.cfg. Joy. How are you grabbing your kickstart config? If over the network, then it's always been a matter of you either use DHCP or you pass all of the ip=, etc flags on the kernel command line. In the initrd: default vfc8i64 append initrd=ifc8i64.img text reiserfs ks=file:/ks.cfg This change is stopping us from upgrading the machines in my lab using kickstart as we've done since the Fedora 1 days. We boot with IP=... on the commandline and fetch the ks.cfg from a webserver. I don't control the network and can't run a DHCP server accessible by all lab machines, so this is a major inconvenience. It's always worked so well! Now it's back to hand-installs for us. Similar problem here: when a system needs to be reinstalled (computer lab, or OS upgrade), I have a script that creates a kickstart file on disk, copies the installer kernel and initrd, and adds an entry to grub.conf, then reboots to this installer. Worked like a charm in previous versions, but with Fedora 8, I'm stuck on the fact that it always requires dhcp. Is this a design decission to drop support for everything but dhcp? If so, why isn't it in the release notes? If not, what is to be done about it? I'm using a workaround. In loader.c - doLoaderMain - case STEP_IFACE I commented out this row: loaderData->ipinfo_set = 0; I did it because in net.c - setupNetworkDeviceConfig network options are copied to netDev structure only of ipinfo_set != 0 I have the impression that kickstart loading process sets this variable to 1 and only AFTER this, loader inits this variable to 0 overwriting kickstart static request. I tested Matteo Faleschini's workaround, and I rebuilt "boot.iso" from the F8 installer DVD ISO image (ie: images/boot.iso). It works every time. I now have several shiny new F8 nodes whirring away. Same bug existst in Fedora 9, but it seems the workaround from Comment #6 no longer does the trick I think this bug was introduced by the patch for BUG #233152. Moving this to F9 bugs as the workaround for F8 no longer works. I can confirm this bug is present in F9. Also, this seems related to bug #451711 I think I've fixed this in anaconda's git repo. The next build of anaconda (anaconda-11.4.1.49-1) will have the fix. That that version appears in the rawhide tree, please test if you can. The functionality prior to Fedora 8 has been restored. If you specify ks= and give it a network location, but do not specify IP configuration parameters, anaconda assumes DHCP. Likewise, if you specify ks= as a network location and specify ip=, netmask=, gateway=, and so on, anaconda will configure the interface with those parameters and DHCP is never run. network lines in kickstart files work as well, at least in my local tests. I need you guys to test rawhide once anaconda-11.4.1.49-1 lands in rawhide. Thanks for all the reports. The functionality prior to Fedora 8 is not restored. In F10 it is impossible to avoid dhcp when doing a nfs install from a non-predefined static ip. I have an equivalent of boot.iso on a usb key, with a kickstart file added in the initrd. The kickstart file includes: nfs --server=m19.limsi.fr --dir=/corpora-backup/system/%%REPO%% With: not "network" line -> dhcp network bootproto=static -> dhcp network bootproto=ask -> error window, then dhcp, then error windows again then reboot network bootproto=query -> error window, then dhcp, then error windows again then reboot network bootproto=static ip=.... -> static ip used Is there something I can put in my kickstart file so that I get asked for what I want and I don't have a dhcp systematically done? OG. I performed an installation of rawhide under the following conditions: 1) Booted using boot.iso 2) Put my kickstart file on a USB jumpdrive and passed ks=hd:sdc1:/ks.cfg on the boot line. 3) Removed the 'network' line from my kickstart file. Installation proceeds as expected and the interface is not brought up via DHCP. I am prompted to configure the network interface when it gets to package selection. Same on RHEL 5.4 Best Regards Marcus Is there a boot image one could use to install RHEL 5 with kickstart from network and static IP? Again verified on RHEL 5.4. |