Red Hat Bugzilla – Bug 226814
anaconda "loader" issue when using pxe kickstart w/ dhcp
Last modified: 2007-11-16 20:14:55 EST
Description of problem:
I know another kickstart/dhcp bug exists but I'm not sure if this is the same
issue (since they seem to have working work arounds). When attempting to
install the OS via pxe boot, the loader reads the kickstart file but continues
to ask for the network configuration after attempting to get dhcp
information. This occurs whether setting --bootproto to dhcp OR static
(should I assume static options don't work when using pxe?). I tried a
variety of kernel arguments including; setting the linksleep to 30, 50, 80...
etc, setting the nicdelay= option to multiple times, and using the
eth0_ethtool="speed=100 autoneg=off duplex=full delay_link=1" option. I've
tried this with RHEL4U1-4 using x86_64 and intel architectures on multiple
switches with no success. I really can't believe this is still an issue given
the current trend toward large HPC clusters and since Redhat also prides
itself on simplicity to install. Please advise me on any other options, keep
in mind I will not have the ability to change switch settings.
Version-Release number of selected component (if applicable):
RHEL4 (UPDATE 1 through 4)
Steps to Reproduce:
1. Create a dhcpd server with bootp and configure to install with kickstart
(nsf or http). I can only use Cisco switches - with default configurations -
in this environment.
2. Boot the client machine w/ pxe, it recognizes dhcp and begins the install.
A window shows the client is re-getting dhcp info. Then the manual network
card input menu appears (suggesting dhcp failed).
An input menu appears, dhcp fails.
No input menus - an entirely automated install process.
*** This bug has been marked as a duplicate of 189795 ***