Keys Corner » Apple

Time Machine backup stalls if NAS volumes are mounted

John — Sat, 21 May 2011 05:49:51 +0000

I have had a problem on my Mac mini for several months. Time Machine hangs after backing up a small amount of data, and refuses to continue. I can leave the machine running for days and nothing happens. This is the sort of thing I see when I click on the Time Machine icon in the menu bar:

Time Machine stalled - the figures in the screen shot never change

It is rather strange as we have an iMac running in the same network, on the same version of OS X (Snow Leopard – Version 10.6.7) and Time Machine runs just fine on that Mac.

By chance I have found the solution to the problem, and I haven’t seen a description of it anywhere on the internet, so I thought I’d post it here.

Our configuration is that we have a (wired) network with the two Macs and a Synology DS210+ NAS server connected. The DS210+ stores all our data, iTunes music and so on and when the Macs boot, several disk volumes on the DS210+ are automatically mounted on their desktop. These volumes are all stored on the internal drives of the NAS. We also have an external drive attached to the NAS via USB, which is exclusively used by Time Machine for its backups.

Solution:
If, on the Mac mini, I unmount the volumes stored on the internal drives of the NAS, Time Machine immediately continues and completes the backup. If I remount any of those volumes, Time Machine stalls indefinitely until I unmount them. Simple. It works every time. I have no idea why I only need to do this on the Mac mini, and not on the other machine.

If you find this useful, I’d appreciate a short message either as an answer to this post, or by e-mail (see the “Contact us” page) – I’d be interested to see if others find this solution works for them too!

Relocating referenced master images in Aperture

John — Sun, 07 Nov 2010 18:02:04 +0000

I have done this a couple of times in the past, and the solution is not well documented by Apple:

I use Aperture 3.0 to manage my digital photographs (tag them, edit them, organize them, and so on). I don’t like the idea of storing my photographs inside the Aperture library. An Aperture bug could destroy or damage them them, and recovering them from a backup could mean loosing my latest pictures. So I store them externally (on a server in my case, but it could also be on an external USB-drive attached to my computer) and link to the original images when using Aperture. That is what Apple calls “using referenced masters”.

What I have done in the past, and again a couple of weeks ago, is to replace the server that the photos are stored on.

The new server has a different name and network address, so of course Aperture can’t find the referenced masters of my photographs. This needn’t be obvious immediately – Aperture keeps a lower resolution copy of each photograph in its library, so until you need to access the full-resolution version of the picture, you don’t see that the connection to the external master copy has been lost. By the time that last happened, I had re-formatted the original disks and sold the server.

Really, the only thing I needed to do was to globally edit the old path to the top directory where the photos were stored – replace the server name or network address with the new one, and everything would have been OK. Scanning the documentation and the Apple discussion forums didn’t give me any help on how to do this. Apple foresees you wanting to move referenced master photos from one drive to another, but assumes that you still have the old drive and that you want Aperture to do the moving for you. If that is the case, you can use the “Relocate Masters” command. But there is no obvious way to change the address of the masters if you don’t have the old drive.

Some experimentation led me to this solution, which worked well for me. I used the “Locate Referenced Files” command as follows. There are two ways you can approach this – you can process one project at a time (which is what I did, since I was trying it out and didn’t want to have fix a problem with 11000 photos if something went wrong!) or if you are feeling adventurous, you can select all your photos in the library by clicking on “Library / Photos” in the vertical menu on the left of the Aperture window.

Choose a photo (framed in yellow in the screenshot) in either a project or in Library/Photos

Above, you can see one photo has been selected (yellow frame) and the text “Offline” that appears when the mouse hovers over it shows that Aperture can’t find the original drive that it is stored on.

Go to the menu at the top of the window; click Select All

Right-click on any of the selected pictures, click Locate Referenced Files... in the pop-up menu; a new window appears (see next screenshot)

We'll discuss this in more detail below...

When you click “Locate referenced Files” a new window opens – it’s a bit small in the shot above, but in the top part you see the list of files which you have selected, and where Aperture thinks they are stored. In the lower half you can navigate along the path to the folder where the pictures are now stored. You can see the details in the two screenshots below. If you match up a “lost” photo in the upper half of the window with its actual location on the new drive in the bottom half of the window, and click on the “Reconnect All” button at the bottom right of the window, Aperture will correct the path to all the photos that you selected – even if they are in different folders!

In the top half of the window select any one photo which needs relocation

In the lower half of the window, navigate to the actual location of that photo now

Here you can see the path being corrected for all the files in one project

The process is quite quick and worked without any problems at all in my case.

Resetting Software Update to use Apple’s servers

John — Sun, 02 May 2010 11:31:20 +0000

Using a Snow Leopard Server, (OS X 10.6.x) it seems that administrators have a problem if they have used Apple’s Managed Client for OS X (MCX) to provide software updates to their clients from their local server and then want to switch back to having the clients get the updates from Apple’s servers. The problem seems to lie with Apple’s Managed Preferences, which is part of MCX.

Here is what I understand happens, and a solution:

Apple allows system administrators to restrict which preferences and system settings a user on a local client may manage themselves. The restricted services are managed centrally by the system administrator. He does this by setting up one or more groups of clients or users which are managed by the server and the changing of these services by users on these computers may then be locked out by the administrator using the Workgroup Manager from the server.

If the administrator removes members from the group, the restrictions should cease to apply. At least that is what I and other users expect. That does not seem to be the case. This means that once the software update source has been changed to the local server, people have problems to reset the software update source to Apple’s own servers.

There are various solutions described in the Apple support forums, but you need to pay close attention – the way that the software update in MCX is managed changed for each recent version of OS X and none of the solutions I found works permanently for Snow Leopard.

Here is what I have found to work. The solution allows you to switch back to sourcing from the Apple servers, but seems to reset itself periodically, so that it wants to use the local server.

In the Workgroup Manager Set the preference for the Software Update for the group being managed to have an update frequency of “Never” and clear the field specifying the location of the software updates on the local server.
If you don’t have any other need to manage the clients using MCX, you need to delete the complete Managed Preferences directory at /Library/Managed Preferences/

I used a user with administrator rights to do this, and was asked for my password before the files could be moved to the trash can.
I also removed the clients from the group defining which clients should be managed for the software update, managed in the Workgroup Manager.

If you have other managed preferences which you wish to keep, you will need to look for the specific files – identifiable by their names containing “system update” – which are associated with the System Update function.

Various forums recommend deleting the client cache ~/Library/Caches/com.apple.softwareupdate and ~/Library/Preferences/com.apple.SoftwareUpdate.plist but deleting these alone didn’t help in my case under Snow Leopard.

If I find out what needs deleting to stop the source reverting periodically to the local server, I will update this note.

Additional Notes:
Finding information about MCX is not easy – it is not covered in the Apple Technical Documentation for OS X Server.
There is an introduction to MCX in Mac Tech here. It was written for OS X 10.5 (Leopard), so is reasonably up to date.
There is an older article introducing MCX also in MacTech here.

Alice on the iPad

John — Tue, 13 Apr 2010 11:53:02 +0000

Alice in Wonderland on the iPad

The iPad seems to have quite a potential to shake up the market for eBooks.
(Click the pic to view).

On almost the same subject – we saw the Alice film in 3D at the weekend – and thoroughly enjoyed it!

(Via Daring Fireball)

Using Snow Leopard instead of Snow Leopard Server?

John — Sat, 27 Mar 2010 16:58:07 +0000

I have seen several articles over the years indicating that there are no problems in using a standard / client version of OS X as a server instead of using the OS X Server version, which is a darn sight more expensive. And indeed I have used a Mac mini with a client version of OS X for several years as a file and music server. It needs very little power, is small and just sits there unobtrusively doing it’s job 24×7.

However since I migrated all our Macs, including the server, to Snow Leopard (OS X V10.6), the Mac that I use as my work station has been driving me crazy. Any time I connect it to a network drive, it hangs itself up. Not straight away, but often after it has gone to sleep for a while. After that, as soon as I access a network drive again, the Finder hangs with a spinning beach ball and all the other processes become unresponsive. Usually the only solution is to reboot. Very frustrating, and bad enough to make me think seriously whether I wouldn’t be better migrating to Linux, since Apple has in the mean time issued two updates to Snow Leopard and the problem is as bad as ever. Ruth has had similar problems, but not so bad.

Three weeks ago I saw someone offering a new copy of Snow Leopard Server on EBay and I decided to bid for it. I ended up getting it for 177 Euro instead of the list price of 499 Euro. I installed it on the Mac mini server and to my surprise, all our Macs are now running rock solid again. The problems accessing network drives that Ruth and I had have vanished completely.

So, although many experts confidently insist that you can use the OS X client version as a home server (I don’t think anyone would propose to run it in a production environment), that doesn’t seem to be true in the case of Snow Leopard, although I have previously had no problems whatsoever for several years using the Tiger and Leopard clients as servers. I thought I would post this, as I haven’t seen anyone report this experience previously – maybe it will help someone else having similar problems. (I suspect a cheaper alternative might be to migrate a problematic Snow Leopard home server back to a Leopard client, but haven’t tried that.)

Recreating the correct links to files in iTunes

John — Mon, 15 Mar 2010 17:35:57 +0000

The Mac mini I set up with Snow Leopard Server had previously been acting as a server using the client version of OS X. While I was setting up the server again after installing the Server operating system, I decided to also combine the existing external data drive with a second drive and make a RAID array. That meant reformatting both drives before they could be mapped as a RAID set. So I backed up my data (to a third drive) and then created the RAID set and copied my data back onto the new RAID drives.

I only realised some time later that I had used different names for the data partition before and after setting up the RAID array. So the iTunes library was full of little grey exclamation marks, indicating the track can no longer be found. If you double click on the track, iTunes will allow you to navigate to the new location of the music and will re-link the meta data in iTunes to the file containing the track. But with over 1000 albums and a lot of podcasts, that was not practicable:

The easiest way to fix the problem, that I am aware of, is to export the complete library as it is to your desktop.

How to export your library

This produces a file containing XML statements, however, you don’t need to be able to do more than identify your path to the tracks in this file. You need that because you are going to substitute the new path for the old one wherever it occurs. You can edit the XML file using, for example, TextWrangler and simply make a global replacement of the old path (up to, but not including the file name) with the new path. This might take several minutes – in my case the library file was 37 MB and that does take a time to substitute all the occurrences.

When the file has been edited and saved, you can import it into iTunes.

How to import the edited XML library file

The only small problem is that you will now have all your tunes listed twice, once with an exclamation mark and once without. However that is no big thing – you can eliminate the “wrong” tracks (with exclamation marks) by downloading and running the Super Remove Dead Tracks script from Doug’s Apple Scripts web site.

Being a script it is very slow, taking about 4 hours to run in my case. However, once it has finished its magic, you will have a clean iTunes library again.

Installing Snow Leopard Server on a Mac Mini

John — Sun, 14 Mar 2010 23:50:05 +0000

I picked up a new copy of OS X 10.6 Server on EBay for 177 Euro (instead of the 499 Euro it costs if you buy it new from Apple) and have installed it on my 2006 Intel Mac mini. I had one issue at the start of the installation, whose solution is not obvious.

I started a headless installation from a second Mac. To do this, install the Server Admin application from the Admin Tools disk on the Mac you will be using to control the server from, and insert the OS X Server disk in the DVD drive of the server-to-be. The Server Admin tool will find your other Mac and ask for a serial number. Apple’s documentation says to enter the first eight characters of the server hardware serial number, which on an Apple X-Serve is on a label on the server body.

In the case of the Mac mini, there is no label, but if you can display the serial number using the “About This Mac/More Info” menu point under the Apple logo in the main menu. See the screen shot below.

Showing the serial number of the Mac mini

However entering the first 8 characters just produces an error message saying the the serial number is invalid. In cases where there is no hardware serial number, you are supposed to be able to enter “12345678″, but this also produces the same error message.

I eventually hit on the idea of entering the complete serial number, which in my case was 11 characters long, and bingo! The installation started correctly. After the initial installation is completed, you can remotely access the server via the Server Admin tool using the first eight characters of the server’s serial number

Apple Packaging

John — Sat, 05 Dec 2009 13:36:09 +0000

The Apple Store yesterday in Frankfurt’s Fressgass’
If you have ever bought an Apple product, you know their packaging is usually exquisite. Well, the same applies to their stores before they open. The Frankfurt store pictured above opens at the beginning of January, and work is still going on inside. However, the logo is already installed, and the whole shop front is covered in black foil to keep our prying eyes from seeing the preparations.

So where’s the backup going to be?

John — Tue, 18 Aug 2009 19:47:00 +0000

Apple is building a new 1 billion dollar data center, five times larger than it’s existing one in California. Supposedly, to start offering “cloud computing” services (i.e. allowing users to store their data on the web, or use web-based applications).

Everyone’s major concern about cloud computing, is “What happens to my data if the service goes down?”. (Another one, especially for companies, is “How secure is my data if its not in my data center?”). Apple, being Apple, is sure to have addressed that issue. So if they are going to offer cloud computing services, where is their back-up data center going to be? Unless a second location for a similarly sized data center emerges, I would guess that its more likely to be for adding capacity to Apple’s content delivery services: iTunes and App Store servers, which have in the past been outsourced to companies like Akamai.

Copy folders like in Windows, on the Mac

John — Sun, 24 May 2009 09:51:44 +0000

Apple’s OS X has one “feature” which periodically causes me a lot of grief. Not often, because once I have fallen over it, it takes several years until I forget and make the mistake again. I use Windows PCs at work, and there, if I copy a directory’s contents into another directory, it effectively merges the source directory files with those in the target directory.

OS X however treats each folder as a single object and replaces the complete contents of the target folder with the contents of the source directory. This usually catches me out when I decide to update my local (test environment) WordPress installation with the latest and greatest version of WordPress. The standard way to do this is to copy the new kit containing all of the WordPress files and sub-directories into the folder where the test site is installed. In the target folder containing my test installation there are, however, a few extra files that aren’t part of the WordPress installation kit – configuration files, the site favicon.ico file and one or two other files and folders. An OS X copy command deletes these without even stopping and asking if that is what I intend to do. Time to be glad that I run Time Machine in the background and can usually restore the missing stuff in a couple of minutes. But very annoying.

The solution is first of all to remember about this nasty difference between Windows and OS X. Then, I ideally need a way to perform a Windows type of copy, which OS X considers to be a folder-merge. The best solution I have found is to download the mergefolders script. This little script can be stored anywhere and dragged to the dock where it sits, waiting to be run. It asks the names of source and target folders and then performs the necessary merge into the the target folder, just like Windows.

If you want to read up a bit about this difference between the way the two operating systems perform a copy, then look here and if a summary of all the differences between OS X and Windows Vista is what you are looking for, you can find that here.

Junco-operative

John — Mon, 23 Mar 2009 11:13:31 +0000

Sometimes I ask myself what the developers of the Apple Mac spelling checker are smoking…

Fortunately, most people don’t accept that particular suggestion – not everyone was paying attention however…

Getting the most out of iTunes

John — Mon, 09 Mar 2009 09:24:36 +0000

Here’s a good article on Gizmodo, which lists a bunch of useful additions to iTunes, some for the Mac, some for Windows and even a couple to allow you use Linux to play from iTunes libraries. Take a look.