A halted rebuild, and a SQL database that refused to attach
Batch and QA records at a Croydon food manufacturing facility resided on a six-bay server that had quietly operated one member short since the summer. When an amber light came up on a second carrier, a replacement disk was installed, and the rebuild began. Unfortunately, the rebuild halted mid-stream. The virtual disk then went offline, and upon the next restart the job database identified itself as SUSPECT. The server was unplugged later that day, and each individual drive arrived with its bay number still written on its carrier. The two weak drives were imaged prior to imaging the remaining drives; one of them required four passes. All remaining drives read clean from start to finish. What the controller had recorded about the set was read from the images created during recovery, and a volume was built over these images, with the .mdf and .ldf files copied off it. Each database was attached and tested here before anything went home.