[coolpor2]

My main advice is once you find a way that finally works stick with it often minor things are significant. And don’t be afraid to restart from scratch. i.e set up a new deloyment server (this would be last resort in your case as it seems it’s screwing up during actual image creation), and reinstall everything on the reference computer starting with the OS if need be.

If it helps I was running osx server 10.4.9, and the image was created from a G5 running 10.4.7. And I’m sticking with those versions because now for us everything just works.

If you’re sure that it isn’t a network issue i.e the usual suspects dhcp, dns and nfs are behaving then just be really methodical with recreating the image.

If you are going to create an install image from a dmg file make sure you check and verify permiissions on the volume…I’m surprised the direct firewire capture didn’t work the reason i avoided that method in the first place was because i was being impatient and a lazy ass.

Once i got everything working something else cropped up to put a spanner in the works. When I tried to deploy to some machines the following day it would hang at the blue screen. It turns out I needed to a quick re-erase AND re-format of the volume on each affected machine using my trusted osx statup disks – this cleared it but be prepared as this might happen to you.

HTH

[coolpor2]

Thanks for everyones responses. It appears that the main problem was with SIU modifying the dmg disk image during the netinstall image creation process.
It managed to skip certain files….
I’m not quite sure if the RPC timeout error was related but I didn’t get that error again, but the upshot was that I had to reimage the reference machine. But this time I did a direct capture within SIU via firewire rather than my original attempt which was :

Create image using diskutil then create netinstall image in SIU from image file created in diskutil.

It appears SIU is a bit hit and miss if you’re not creating a netboot/netinstall image from a FW target-disk-mode mounted volume.

It’s all very well Apple talking about wiki servers and all the other new features in leopard server, but I really hope they make significant improvements in Netboot and SIU – so that we don’t have to turn to third-party tools for deployment in preference because things that should ‘just work’ in 10.4 don’t. I think fundamentally they’re good tools – but as I say very temperamental. Apple have done a good job with the improvements in ARD 3 now let’s hope they don’t sit still with NB/SIU in leopard.

Anyway, i’ve managed to deploy the OS….eventually. Thanks for everyones input.

[coolpor2]

Mactroll;

I’ve managed to get verbose output from the NB startup on the client. It’s gets an IP from our dhcp server, locates the path of the image on the server
but then spits out RPC timeout error. On Bombich.com he says that this is likely a NFS issue, but I’ve managed to run the mnt commands he suggest
to check nfs connectivity when booted locally on the client machines. aaargh!!! this problem is driving me round th bend.

Also, to the last poster I have created the image from an imac G5 running 10.4.9

[coolpor2]

Hi

Thanks for the reply. I can connect to the server via tftp from the terminal on one of the clients. I can also
mount the NetBootSP0 share via NFS so no problems there.

I’m still having difficulty getting verbose output for the NB startup. I’ve tried pressing CMD V and N a the same time but
all I get is the verbose ouput for the normal boot process, and this doesn’t really tell me anything.

[coolpor2]

The problem of creating network home dirs cleared up after a reboot of the server and all machines. I don’t know how
and why it was fixed.. i’m none the wiser except that homedir creation is now working without fault, however now i’m gettting an even more serious problem as a result of the restart, the problem is that network accounts are failing at login with the following message;

"Unable to login at this time….The home folder for the user account is located on an AFP or SMB server."

This is not a machine specific problem but it is account specific – the majority of accounts are affected but some are not.
Here are some checks i ran at the server using the dscl commands;

1) The hostname and NFSHomeDirectory records for the users DO match up

2)When using dscl /LDAPv3/127.0.0.1/Mounts i get the following two options;

macsvr1.mydomain.com:/Volumes/Data/Home
macsvr1.local:/Volumes/Data/shared

i then look at the macsvr1.mydomain.com in more detail, and heres what i see;

cn: macsvr1.mydomain.com:/Volumes/Data/Home
mountDirectory: /Network/Servers/
mountOption: net url==afp://;AUTH=NO%20USER%[email protected]/Home
mountType: url
objectClass: mount top
AppleMetaNodeLocation: /LDAPv3/127.0.0.1
PasswordPlus: ********
RecordName: macsvr1.mydomain.com:/Volumes/Data/Home
RecordType: dsRecTypeStandard:Mounts
VFSLinkDir: /Network/Servers/
VFSOpts: net url==afp://;AUTH=NO%20USER%[email protected]/Home
VFSType: url

All these options look fine to me, although i’m not sure about the net url== entries.
Everything looks ok in WGM inspector for each user pointing to the correct network shares.

Any ideas ? I’m sure it’s probably something very simple.

[Author]

Posts