morganw

You'll see the kernel used during the boot process on a client, unless you've made any per host exceptions for the kernel file then it will always be the default that was included with whatever version of FOG you are using. I think if you cannot use the interface to download extra ones then you have to resort to just putting them into the kernel folder. I've always replaced the included kernel with one from version 0.24 as most of our equipment is fairly old. If I have a problem then I make an exception to use the newer default one. When you set which kernel to use you do so relative to the root of the tftpboot folder, so the value you set is "fog/kernel/kernelname".

I think I've found what was doing it, I'm doing further tests today. As far as I have found out, OS X doesn't actually do dynamic DNS updates (there seem to be a lot of people in forums struggling to achieve this behaviour) but if you are using a directory services plugin (like the Apple AD one or Centrify Direct Control) then these default to doing dynamic DNS updates. Our main DNS server isn't a Windows one and DHCP populates it with client records, after disabling the Dynamic DNS update command in the Centrify settings I think that reverse DNS records are correctly left alone by the client.

The filtering is free, the archiving isn't. I think Live@EDU is the same, free filtering.

I seem to have an intermitent problem whereby some of our Macs are deleting their reverse DNS records shorty after they have been set by DHCP. Does anyone know how to stop this behaviour?

No the user authentication has never worked (as the NFS server can't read any information from the port map server). The maps are generated by a Centrify Direct Control NIS server so this will handle the link between the unix accounts and the windows ones - I presume I don't need anything on the domain controllers. I can see the maps on the 2003R2 server with no problems and everything looks fine. Everything is part of the same domain. I think you've found the answer with the .maphosts files. I've never edited it so I presume the it's restricting the use of the user name mapping service to the localhost rather than allowing my 2008R2 server to read the data. Much obliged.

OS X 10.6 I can connect if I allow authentication for unmapped users.

2008R2

I've setup an instance of 2003R2 just to use the user name mapping for NFS, this is pulling data from an NIS server running on Debian and I can see the simple user mappings are generated successfully. The actual files I need to share via NFS with authentication are on a 2008R2 server (and it's a lot of data so I can't easily move it somewhere else). I installed the Service for NFS feature and set it up to use the 2003R2 server for user name maps but it logs this error every 30 seconds: Event ID 1005 — NFS User Name Mapping Connection I've turned the firewalls off on both servers, and tried using the hostname and FQDN of the 2003R2 server but I still get the error. Any ideas what would be stopping this from working?

The simplest way is to use stunnel on a machine that has permission to send outbound email. I think the settings would be: accept = 127.0.0.1:25 connect = smtp.gmail.com:465 If you google stunnel and gmail you'll find a lot of examples.

If you create your image on an older machine you can successfully switch between the ACPI Uniprocessor and ACPI Multiprocessor HALs just by using Sysprep, but this is the only way that is supported to work. So you need MySysprep if this isn't good enough. I do kind of the same thing as Chad, just keep a reference sysprep folder with everthing in and just add drivers to it as they are needed.

This is your friend for universal XP images Vernalex.com - Welcome Download the SysPrep Driver Scanner from the same site and it will handle the pnp driver installation, and get DriverMax to export drivers from a working system if you can't locate an actual download with the files in it. To get more control of driver selection you'll probably have to start using something like MDT and filter driver selection based on the host.

You may want to check out Clonezilla instead of WDS, FOG, Ghost etc. I've haven't used it but from what I've read it has options that can make it more suited to portable use, whereas all of the others are really designed to be permanently located on a site. Maybe not so important if you can take a VM around with you though.

Yes but you won't get much benefit from having the server actually activated unless you W7 testing goes over the grace period. You can always just install every 30 days, it would stop working after 60 days anyway as you wouldn't have enough clients contacting it to keep it running. You can run this to check exactly what the server has registered to it: cscript c:\Windows\System32\slmgr.vbs /dli

You might have to regenerate the SID for each installation for it to count but eventually the hosts get removed from KMS (after 30 days of not checking in) and if the number of installs falls below the threshold the server will stop activating. Allegedly if your KMS server is 2008R2 then it will also activate virtual machines, if you KMS server is running on anything older it will only activate physical machines.

I've got the same problem here, I think the problem is that the keychain dialogue boxes that pop require far too much thought to work out what they mean. If you press the right buttons it isn't a problem, but it seems most people just press anything just to get to start working. I think the easiest solution is to use a script to delete the keychains every night, but then you wreck it for everyone who does actually understand how to use it.

We moved our Apple computers onto the main network using some purchased software to authenticate via active directory. The home folders are on a windows file server that is in the same physical location as the old mac server that used to store them. When using the macs sometimes they will either have slow periods or lockup for a period of time, even when logged on a local user account rather than a network one. Has anyone got any experience of Apple clients connected to their network (ours is setup flat with everything in one subnet)? I've changed the tcp parameter "delayed_ack" to 0 from 3 as people have written about this causing network performance issues, but my problems also exists for a local account so I'm looking more towards things running on the workstation. I've disabled bonjour on one macbook as a test and since then I've been unable to get to have any slowdown or freeze issues, but I'm hoping that some can offer me specific advice or comment on similar issues before I spend hours disabling bonjour on everything. Also considering turning off spotlight indexing to see if that helps. Any help appreciated.

I did it by having a tidy up on the CC3 system, then wrote a script to copy the home folders via some external storage (external hard drive or anything that will just hold the files), then put them into the relevant place on the new fileserver(s). My situation was a little more complicated as some students had an additional home folder that would mapped as a drive when they logged on in the music department, so the script also consolidated that in and also drops things like application data on the way (we moved to mandatory profiles). Then when all the home folders were in place I wrote another script to recurse through the home folders and log each folder name to a text file, I then made this into a CSV file for import. To actually import the users and set up shares, security, groups etc I wrote a power shell script that took my CSV as an import. This was the part that was complicated as I wanted everything to be automated (and was more complicated again as I also wanted to provision a second home folder just for use by apple macs on a seperate fileserver).

You can try an older kernel from an older fog release, or build your own, or try one with broader hardware support from the kernel updates menu in the management interface (try a KS one). Also might be worth plugging in a CRT monitor if you are using a TFT one, some of ours won't display some screens when using WDS deployments depending on the screen resolution.

You can pick up new kernels by going to the 'Kernel Updates' section in the interface. I would try the the KS ones as I believe they are meant to be built with pretty much everything in. It might be worth finding out what the SATA controller is on your computers and then look-up Linux support for that rather than search for issues with FOG directly.

Robocopy in mirror mode. Robocopy

As a test if you disabled the onboard network card, then stuck in a network card from an older PC, I wonder if that would make a difference? If that works it would eliminate an issue with the CPU. I'm still on 0.29 and when I saw that 0.30 didn't have any fixes for the services on Windows 7, or the fsarchiver option, I decided to let the dust settle on this one.

Out of interest what are the CPU types on the computers that imaged slowly?

I wouldn't have the through the distribution you are using will make a difference, especially as you are getting massively varying speeds between computer models. Once it's going it's the client side that does all of the work, the server is just offering the images with NFS. Try setting an older kernel on a per host basis for the computers that are now taking 40 minutes to image.

Do you need admin rights for it to install itself or because it's in the users profile will it install regardless?

So do you have to aprove anything to install it or can you pick it up by just visiting certain websites?