Informix/NCR/partition table problem

We have recently had a serious problem with one of our NCR 3555 systems. Please read the following and let us know if anything similar has happened to you, or if you've heard of similar. We're especially interested in causes and/or solutions. This system is configured as such: 16 - 90 MHz Quad Processors with LARC 1GB RAM 2 6299 disk arrays housing 230 GB configured with RAID 5 Dual Active Controllers with RDAC 04.00.04.12 MP-RAS 3.00.00 Raid Manager 03.05.00.03 In a nutshell, this is the problem. The Informix log shows an error 5 on a chunk within the root DBspace. Streams showed an aborted read error on a block of the disk within that chunk. No fault lights appeared at any time on the drive or the disk array, nor did any other warnings or errors appear on the Unix system in the OSM or in the Raid Manager. At this point, the database was taken offline due to the fact that the rootdbspace was marked offline. The database administrator dialed in to check things out, and contacted Informix Regency Support. Informix dialed in, and attempted to mark the chunk back online with a support utility. This did not succeed. It reported an error in the Informix log indicating that the disk was bad. AT&T GIS support was contacted to try to determine why, if the disk was bad, no fault light or error message was reported; or alternately, why Informix thought the disk was bad if it was not. A support tech at NCR/AT&T GIS dialed in to look at the disk; they found nothing wrong, could not make a recommendation for correction, but could not read the disk. One of our in-house engineers suggested executing a prtvtoc on the disk, to look at the slice layout. The command resulted in a response of "no active unix partition". Fdisk showed no partitions of any kind. At NCR's request, we re-created the partition and slice information using edvtoc and recovery files created in advance for such an emergency as a disk failure. AT&T's statement was that it would not adversely affect the data on the disk. Informix support tried again to restart the database, but still could not. It appeared to them that the rootdbspace was corrupt. A dd of the first 100MB of the drive returned all null bytes (0's). At this point, there was nothing left to do but begin a full database restore... Thanks, -- /---------------------------------------------------------------\\ | Bobby Wilkins Harrah's Entertainment, Inc.| | Systems Engineer 850 Ridge Lake, Suite 300 | | email: hwilkins@harrahs.com Memphis, TN 38120-9460 | | voice: 901-820-6062 fax: 901-820-6029 | \\---------------------------------------------------------------/