Workgroup Manager Crashing on connecting to server

Recently I've had an issue with Workgroup Manager crashing when it connects to the Xserver:

Error of type eMemoryAllocError (-14901) on line 401 of /SourceCache/WorkgroupManager/WorkgroupManager-361.3.1/DirNodeRefController.m

Please not I'm using:

• Workgroup Manager Version 10.6.3 (361.3.1)
• OS X 10.6.8
• OS X Server 10.4.11

Doing a copy and paste in google brings up on 1 result which happens to be on Apple Discussions.  The discussion is exactly what I was looking for. Here it is.

Basically they describe the problem after a security update (not sure whether it was on the client or server). Luckily there is a work-around by ManuelSF, but people are not really liking it. Here it is:

1. open workgroup manager
2. click cancel at the login screen
3. go to server in the menu bar and select view directories
4. you should see your directory
5. click on the lock in the upper left-hand corner to unlock and edit the directory.

Works better then it did before surprisingly!

Apple Remote Desktop Task Server trouble

Just after rebooting the xserver I noticed an error in the logs made by ARDAgent:

Jan 27 14:20:19 Xserve ARDAgent[460]: Bind error on TCP port 3283.  Error number: 48 Address already in use

Looking at remote desktop on my machine I noticed it was having trouble communicating with the task server. I tried restarting the ARD service to no avail.

I looked through some more logs as it appeared that something else had taken that port. The only logical things would have been iChat and another VNC service running. I checked IChat out and it wasn't running. I did find another log entry just after reboot, but before the ARD error which was registered to OSXVNC. Sure enough this was running. I shut down the service and made sure it wasn't starting up any more at startup.

As soon as this was done the task server was back up!

Xserver OS X 10.4.11 no ping

We've been having this problem where the server doesn't return a ping for some time. The work around is to simply unplug the network cable and plug it back in. I have noticed on two occasions (I didn't check the other occasions) that in the system log:


Jan 27 14:38:44 Xserve servermgrd: servermgr_dns: no name available via DNS for 172.25.1.81
Jan 27 14:38:44 Xserve servermgrd: servermgr_dns: no reverse DNS entry for server, various services may not function properly
Jan 27 14:38:44 Xserve servermgrd: servermgr_dns: gethostname() doesn't match the configured name - use changeip to repair

It might be worth running this "changeip"

Making a pkg file for installing sendLogon for Bloxx on OS X

In a previous post I described how we're getting a new web filtering system called Bloxx, which needs the sendLogon program to be installed and run as a login script on all the Apple Macs (OS X).

I used Iceberg free application to make the PKG file as I've used it before with relative ease, and also couldn't find the latest official Apple PKG maker app!!

Its all working now so here's how I did it:

1. Made a new project in Iceburg, and made a nice little folder to hold all the files needed (makes it easier to find things)
2. I filled in the top section "Settings"

3. Next was the "Scripts" section. I had to make a script which changed the com.apple.loginwindow plist file to add the LoginHook /usr/bin/login.sh
1. I ran nano from the command line (nano ~Desktop/sendLogonfiles/Install\ sendLogon/writeloginhook.sh)
    and typed:
2.  #!/bin/tcsh
   sudo defaults write com.apple.loginwindow LoginHook /usr/bin/login.sh
3. I found I had to put "sudo" into the script so that it had the right permissions to write to the file. Not sure why else it didn't work?
4. I saved the file and made the file executable - chmod 755 ~Desktop/sendLogonfiles/Install\ sendLogon/writeloginhook.sh
4. I then added this script to the "PostFlight" script, and set the path to "Relative". Note - I found an article here documenting the definition of each of these types of scripts. Most were concerned with whether this was an upgrade or not, and would only run once.

5. Finally was the "Files" section. 

1. I firstly put the files in the project folder.
2. I set these files with the correct permissions (as in the tutorial folder for sendLogon). chmod 755
3. Then, in Iceberg, I had to add the folders; /usr/ and /usr/bin/, which were not there.
4. I then made the "Default Destination" the /usr/bin/ folder
5. Finally I added the login.sh file and the sendLogon file.

I've just tested this through the Apple Remote Desktop Task Server on a single MacBook Pro and worked a charm (had to ring ICT though to add its' IP address to the Bloxx list of computers with sendLogon working)

Note: There is potential also for adding to this a little script which changes the proxy settings to the system and maybe also Firefox, but it would be easier to control these settings through a .pac file or .wpad file.

Deploy Studio RC20 MultiCasting on OS X 10.4.11 xserver

After finally finishing the Master image it came in at 72GB! Takes about 3 hours to image one machine and 6 hours for two. So I thought I'd give Multicasting another go (I tried it a couple of years ago with lots of lost packets!). Also just heard they refitted the server room so maybe this'll help.

It was set up a few months back when we had DHCP problems caused by upgrading DeployStudio to RC20. You get the option to set up Multicasting when you set up DeployStudio Server. Here's my post

So taking a new iMac 24" and an old white 21" iMac I thought I'd set up Multicasting. First in DeployStudio admin I clicked the image and clicked the Multicast button at the bottom (looks like the Airport icon)

Airport Icon at bottom denotes MultiCasting

This now shows it's multicasting in the Activity section.

I've set up a workflow to restore an image to a volume, and in this set up some Multicast options.

Setting up Multicast options in a workflow

So far I have tried:

• Attempt 1
• Stream Rate 6MB/s
• Clients Disk Speed 120MB/s (I wasn't sure about the older 21" iMac so I slowed it from 200MB/s)
• Start Trigger 2 client
• Result:
• Stream Rate 5MB/s
• Disk Rate changes from 5MB/s to 15MB/s

I've been writing this as the process is going, and just noticed at 12:48 that some errors occurred. I noticed that this happened around the time that I loaded DeployStudio Admin on my machine, so I think this might have caused a slight interruption on the server (and it is 7 years old now). I do remember that I once remote'd into a machine whilst imaging which interrupted it. So the process is a little delicate. I shall wait to see if it has effected the process or whether it can cope with a few byte missing here and there?

After waiting until the 100% marker (which took 3 hours) it seemed to have stopped with no more errors, although it didn't quite finish. It said something about "starting" but just stopped there. I waited for another hour with no more progress so decided to reboot. Upon reboot the startup disk was not set, and the came up with the white screen with a question mark!

My initial conclusion was that the process had failed, but in hindsight I think that maybe the computers were waiting the data that was lost to be broadcast again, but this might be me clutching at straws. I shall look into it at the end of August. For now we'll just have the old method to use.

Imaging OS X 10.6.6 on OS X server 10.4.11

I have over 100 Macs which need reimaging this summer. To make it nice and simple they will all have 10.6.6 installed. Unfortunately I only have a 10.4.11 server (PPC) which I don't think will be upgraded before summer ends! At the moment I have 9 brand new 27" iMacs which don't like the current version of Deploy Studio RC20.

The reason I only have RC20 installed it that this was the last version to support OS X 10.4.11. Now, I also have 9 24" new'ish iMacs which do work with Deploy Studio RC20, and some relatively new Mac Pro's which definitely work with Deploy Studio RC20. So it is still worth perhaps using it for the 90 or so Macs and resort to other techniques to reimage the 9 which don't work. Of course over time I will need a new system but to have a really good new system we'll need time.

So the plan is to carry on using deploy studio for 90% of the Macs, and resort to something else for the last 10%. Alternatives are:

• Carbon Copy Cloner
• SuperDuper
• The install disk
• the install disk on a removable hard drive
• A bootable 10.6.6 on a removable hard drive

Well, I've looked up CCC and can't find any reference to it having some bootable interface. It simply copies a hard drive to another, which to be fair could be bootable, so I might do this.

I'm currently trying the easiest of these which is to use the install disk. I've managed to find the command line for mounting an AFP share. So I've open terminal from the booted up install disk and typed:

mkdir /Volumes/tmp
mount -t afp afp://username:password@server.somewhere.ac.uk/Images/ /Volumes/tmp/

The image that was taken from the 10.6.6 27" iMac was either done in DeployStudio helper or Disk Utility. When trying to restore this image to a drive in Disk Utility it has come up with an error message:

Restore Failure Could not find any scan information. The source image needs to be imagescanned/scanned for restore

I did a quick search and came up with this. I do now remember that to restore off a network share using disk utility it does need to be scanned for restore. Put the image in Disk Utility and go to: Image > Scan for Restore. This takes a long time on the old Apple Server G4, but not so long on the new 27" iMacs!

It should be ready for restore now...

Restored 70GB compressed in 1 hour 45 over the old network, which isn't too bad. I don't think it'll be able to do multiple machines very efficiently though.

The next process is to try and create or use an old netboot image to netboot the slightly older machines. So it only gets easier from this moment onwards.

Netboot services "greyed out"

Services and/or Sharing greyed out:

The Netboot or other service list items in Server Admin Computers & Services column may be dimmed ("grayed out"). Dimmed items cannot be administered in Server Admin.

Solution

   1. On the affected server, quit Server Admin.

   2. Delete the /var/servermgrd/ directory.

   3. In Terminal, execute this command:

      sudo launchctl stop com.apple.servermgrd

   4. Restart Server Admin.

Also, if Netboot still isn't working:

sudo killall mountd

sudo rm /var/db/mountdtab

sudo mountd

Mac OS X Default Paper Size plist file

For some reason in one room in particular when someone prints to the Dell Colour Laser 3110 the default paper size is Letter. I thought I'd set it up a long time ago not to be. So I've found this thread which mentions the file location for default paper size:

~/Library/Preferences/com.apple.print.PrintingPrefs.plist

It contains:

DefaultPaperID
iso-a4

Add this to a plist file to the "details" tab in a computer list group in workgroup manager and see what happens.

Deploy Studio Set up RC20

So after my trials and tribulations with Deploy Studio RC20 we ended up uninstalling it completely so as to rule it out. But alas, it was needed and so upon reinstalling it, I found that I needed to set it up again. So refering back to my notes........ doh, didn't make many notes (in fact I only made one note), and so I am writing this 'note' on how I did it.

So, I worked out that you need to start the service before setting it up. It's because it gives itself it's name based on the server address, and does other things. So, after this I refered to the pdf I downloaded, which is a little out of date and NOT made by the people who made Deploy Studio (I have no idea why?). So basically:

This network sharepoint was setup previously

These groups were also setup previously

These settings are the default settings and haven't been tested. It did state in the PDF that the IP address should be that of the server, but this is incorrect. It has to do with multicasting.

So, hopefully this helps in the future. Let it be known that I have taken notes!

Deploy Studio set up and DHCP problems

I've been using Deploy Studio server for at least 1 year, perhaps 2, and it's fantastic. It's streamlined imaging perfectly, with workflows helping with different setups.

But all good things have their drawbacks, and upon upgrading Deploy Studio server from RC14 to RC20 everything appeared good, until two days later the Network Administrator rings me up and asks if he can unplug the Xserver, doh! It appeared to be the cause of much dismay for those two days, constantly giving all computers wrong DHCP information, for which is was not even meant to be giving!

We looked into it and the Xserver was responding to DHCP requests and giving "no-dns-server.example.com" as the dns server, overriding what the computers normally had set, which caused much much confusion. It was of course quite amazing that the network administrator managed to track it down to this problem, but I do suppose it took two days.

Upon closer examination, we noticed the DHCP service was "apparently" not running, but when we switched the "Netboot' service off it no longer responded to DHCP requests. So it was the Netboot service which was using part of the DHCP service, which Alan was sure was not 'usual'. Of course we were not sure what it did before the trouble started, whether or not it responded before of not, and whether it was just the fact it was dishing out the wrong DHCP information (IP Stack someone later told me) rather then no information.

We looked at the stopped DHCP service settings and saw one entry, which did have "no-dns-server.example.com" in it's properties. We deleted this entry, which was the only entry, but we still received responses. We measured these responses and it was now giving out the right information, which was better, but we would have preferred it not to give any information.

So to conclude, I whittled it down to the update from Deploy Studio RC14 to RC20, where Deploy Studio had changed, and perhaps added the "no-dns-server.example.com" entry the way Netboot used the  DHCP service.