I broke the number one rule of non-cluster-aware filesystems Thursday night, and accidentally mounted the iSCSI volume that hosts my Aperture library on two systems at the same time, both of which were in read/write mode. The volume was only mounted for 15 seconds or so, but the damage was done: pages of error messages in Disk Utility, ending in “invalid thread count” and an error that the volume could not be repaired. Disk Warrior was also unable to do anything with the disk, reporting it damaged beyond repair.

I also broke the number one rule of using iSCSI on a Mac: my last backup of the volume, which holds my Aperture library (but not the raw images themselves) was on Monday, after I had finished importing all the pictures from Dov and Nikki’s wedding last week.

After a number of restarts of both the Mac and the Solaris box, and repeated runs with Disk Utility and fiddling with the globalSAN iSCSI initiator, I was able to mount the volume, even though I got a pile of error messages about “THIS VOLUME IS DAMAGED. BACK UP WHAT YOU CAN AND REFORMAT IT.” Rsync ran for a long time, and thankfully, pulled a working copy of my 103GB Aperture library back out of the burning wreckage. All 125,000 images files verified, and I was able to do an “Update metadata from master” for everything.

Yes, I’m making an extra set of regular backups now.