Reclone of Recovery HD solved 10.8.3 boot failure

After installing the OSX 10.8.3 update this weekend, my 3 year old stock 15" MBP refused to complete the boot sequence. I finally was able to fix this problem by recloning the Recovery HD using Carbon Copy Cloner, and for the good of the community I wanted to pass on what I learned in case others encounter the same problem.

I had enabled both FileVault2 and a firmware password, so after the initial login screen the boot sequence would start but always hang just before launching the Finder with the message “Your computer restarted because of a problem. Press a key or wait a few seconds before starting up.” Of course, pressing a key or waiting would restart the login sequence and result in the same alert screen.

Booting into the Recovery HD, I removed the firmware password and ran DiskUtility. DU gave the alert “Journal needs to be replayed but volume is read-only” and finally the dreaded “DiskUtility can’t repair this disk” with the instructions to make a backup, erase the drive and restore.

I had made a bootable clone of the drive using CCC before installing the 10.8.3 update, so I booted the MBP from it. Running DiskUtility from the external drive ended with the message “Updating boot support partitions for volume as required” and the final alert “Couldn’t unmount disk” to complete the repair.

Using my bootable clone, I ran DiskWarrior on the MBP’s internal drive. But rebuilding the directories made no difference to the boot up problem.

Again from the bootable clone, I opened the Startup Disk system preference pane and tried to manually reconfirm the MPB’s internal HD as the boot volume. Upon confirming with the “Restart …” button, Startup Disk gave me the error message “Building boot caches on boot helper partition failed.”

Following other suggestions I encountered in support forums, I also tried rewriting the com.apple.Boot.plist system preference and messing with the kextmanager and com.apple.text.chaches, but nothing fixed the boot problem.

I finally gave up and decided to reformat the MPB’s drive and reinstall everything from my clone. But upon clicking on the “Erase” button in DiskUtility, it told me “Disk Encryption Failed - couldn’t unmount disk.” So now I really was hosed. I couldn’t repair the drive and I couldn’t reformat it.

Researching each of these error messages for some clue as to what might be going on, I had earlier come across a support thread on the CCC site where the customer was getting the same “Building boot caches on boot helper partition failed” error in the Startup Disk preference pane. Mike Bombich at one point responded “Alternatively, you could open CCC’s Disk Center, click on the Recovery HD tab, click on your ‘OLD HD’ volume, then click the button to reclone the Recovery HD. Perhaps with a freshly-restored Recovery HD, the OS won’t have any trouble updating the kernel extension cache on that volume.” I assumed that made sense and I was desperate, so running CCC from my bootable clone I followed Mike’s suggestion.

Voilà - boot problem fixed! DiskUtility now reports no problems or errors with the drive, and everything is back to running the way it should.

So did the update to 10.8.3 somehow hose the boot instructions affecting the Recovery HD partition, thereby stopping a successful boot up process? Did the presence of a firmware password and FileVault2 create this situation during the update? It appears that drives encrypted with FileVault2 require a healthy Recovery HD partition, and if this is the case it explains what happened to me.

What is clear to me is that my CCC clone backup and its Disk Center utility saved my MPB’s hard drive. I’m pretty sure that without CCC’s ability to reclone its Recovery HD partition I would have had to buy and install a replacement drive - and here in Brazil that would not have been cheap or easy.

Hope this helps someone else in a similar situation.