Sean Alderman wrote:
> My Menu file was created by the add_install_client script specifically for
> the MAC addr of the machine. As far as hostname -> IP addresses go, the last
> octet of the IP follows the numbering convention for the host names. The
> Menu file looks like so:
>
>
Ok. I always skip add_install_client, and make the menu.lst myself. I
only make 1 for each Solaris version I want to be able to install, and
then I use DHCP option 150 (which pxegrub will look for) to tell grub
which menu.lst file to load for each client.
> [root at c3keval03:/tftpboot]# uname -a
> SunOS c3keval03 5.10 Generic_137138-09 i86pc i386 i86pc
>
>
Which update of s10 are you trying to boot?
> [root at c3keval03:/tftpboot]# cat menu.lst.010017A4770004
> default=0
> timeout=5
> title Solaris_10 Jumpstart
> kernel /I86PC.Solaris_10-1/multiboot kernel/unix - install dhcp -B \
> install_config=192.168.1.3:/export/config, \
> sysid_config=192.168.1.3:/export/config/c3keval05, \
> install_media=192.168.1.3:/export/install, \
> install_boot=192.168.1.3:/export/install/boot
> module /I86PC.Solaris_10-1/x86.miniroot
>
>
I've never heard of install_boot= before, but if it was put in there for
you, then it might be needed. I don't use it, but worse case it it gets
ignored.
> Overall, my problem is not booting. It's that I can not seem to figure out a
> way to determine what is happening after the machine boots. I PXE boot, GRUB
> sends the Kernel down, the machine loads the kernel configures devices, then
> prompts me with a 6 option menu 4 options for installing Solaris, 1 option to
> load a driver, and 1 option for a shell.
This menu is an x86 only thing, and I don't remember what triggers it
getting skipped. I thought it was the presence of 'install' in the
kernel args.
I notice you have both 'install' and 'dhcp', I know in my experience,
using 'dhcp' without all the other parameters never works like it used
to. I don't think I've tried it like you have it. I know for me it works
without 'dhcp', so while it might not fix your problem, it might be
worth testing it with out it.
> This exactly the problem I am having in my heterogeneous environment. I
> only started experimenting on the Blade platform because I can completely
> isolate it from my network and I have 6 blades to work with. I needed to
> rule out issues like : NFS served by our EMC NS500G being incompatible with
> the jumpstart kernel or even NFS served by my TFTP server being incompatible
> with the jumpstart kernel.
>
> What I have now is a completely isolated jumpstart environment that is 100%
> Solaris 10 built from the scripts in the Tools Directory on the DVD and it
> still doesn't work.
>
>
I don't use any of those tools anymore. A long long (solaris 2.5) time
ago I dug through them and learned what they did, and found easier (and
better IMO) ways to do all of it myself.
hmm. I know I figured out how to make that menu go straight to 'custom
jumpstart'... I swear it was only the '- install' that did it, but maybe
I'm forgetting something. I'll have to try to see if I can make it
appear here again.
> Below are the rest of the files associated with the jumpstart server:
>
> [root at c3keval03:/export/config]# cat rules.ok
> probe cpq
> cpq cpq begin.cpq hp_blade finish.cpq
> # version=2 checksum=4309
>
> [root at c3keval03:/export/config]# cat custom_probes.ok
> probe_cpq() {
> SI_CPQ=`modinfo |grep cpq |nawk '{print $6}'`
> export SI_CPQ
> }
> cmp_cpq() {
> probe_cqp
> if [ "X${SI_CPQ}" = X${1}" ]; then
> return 0
> else
> return 1
> fi
> }
> # version=2 checksum=14265
>
> [root at c3keval03:/export/config]# cat c3keval05/sysidcfg
> system_locale=en_US
> timezone=UCT
> terminal=sun-cmd
> timeserver=localhost
> network_interface=bnx0 {primary
> hostname=c3keval05.private
> ip_address=192.168.1.5
> netmask=255.255.255.0
> protocol_ipv6=no}
> network_interface=bnx1 {secondary
> hostname=c3keval05.public
> ip_address=172.18.71.25
> netmask=255.255.255.0
> default_route=172.18.71.1
> protocol_ipv6=no}
> name_service=DNS {domain_name=public
> name_server=172.18.4.5,172.18.12.2
> search=public}
> root_password=uHzDlckCn2mIk
> security_policy=NONE
>
>
To each his own, but I've always tried to have as little in the sysidcfg
file as possible, so that one can be share by all.
I bet that's not causing this problem though.
> [root at c3keval03:/export/config]# cat hp_blade
> install_type initial_install
> system_type server
> pool rpool auto 8g 8g mirror any any
> bootenv installbe bename s10u6-zfsroot dataset /var
> cluster SUNWCreq
> package CPQary3 add nfs://192.168.1.3:/export/pkg/drivers
> package SUNWzip add
> package SUNWbash add
> package SUNWsshr add
> package SUNWsshu add
> package SUNWsshdu add
> package SUNWsshdr add
> package SUNWsshcu add
> package SUNWman add
> package SUNWdoc add
> package SUNWless add
> package SFWfile add
> nfs://192.168.1.3:/export/pkg/companion/Solaris_i386/Packages
> package SFWgfind add
> nfs://192.168.1.3:/export/pkg/companion/Solaris_i386/Packages
> package SFWsed add
> nfs://192.168.1.3:/export/pkg/companion/Solaris_i386/Packages
> package SFWsudo add
> nfs://192.168.1.3:/export/pkg/companion/Solaris_i386/Packages
> package SFWtop add
> nfs://192.168.1.3:/export/pkg/companion/Solaris_i386/Packages
> package SUNWntpr add
> package SUNWntpu add
>
I can't see any thing wrong here either.... Hmm.
I'll try to poke around here and see if I can get any more info.
-Kyle
