1

Got a customer that has an issue with NTFS corruption that is popping up on SAN attached storage. The LUN is RAID-10 (4:1 spindle ratio). The machines are Windows Server 2003 (64-bit), and they are in the planning stages to migrate to Windows 2008 R2 (64-bit).

The manufacture of the SAN (who shall remain nameless) is at a loss for what to do (or why they are occurring), and Microsoft Support has said to go to backups. (Not an easy thing, the backup method is BLOCK-LEVEL, not FILE-LEVEL. Any corruption on the drive is copied into the backup).

I know it's a long shot, but would it be worthwhile to run SpinRite on the LUN?

  • What transport protocol are they using (i.e. FC/FCoE, iSCSI etc.)? and what additional software and/or drivers are in place? – Chopper3 May 06 '13 at 17:45
  • I know it's Fibre Channel. Not sure about additional software and/or drivers. – Jerry Sweeton May 06 '13 at 17:48
  • FC SANs, especially when configured using Server 2003, are notoriously picky about using a validated stack of components. This means ensuring that your OS, drivers, MPIO software and HBA firmware are all in a supported chain of levels - ensure you're using a supported set. Also be aware that 2003 and 2008 deal with FC a fair bit differently and require changes on most/many SAN arrays during migration plus you don't want one LUN being shared by BOTH 2003 and 2008 at the same time if you can avoid it. – Chopper3 May 06 '13 at 17:55
  • I Don't think spinrite would help. If spinrite found errors at what level would they be occurring? Fibre channel presents as a physical drive attached to the system but in reality is a networked drive with a bunch of different devices in the way. Are you trying to move the lun owner between hosts or is the corruption occurring during copy between luns? – Jim B May 06 '13 at 18:34

0 Answers0