More extended maintenance is planned for the server that hosts /home directories and web service this Friday March 18th starting at around 11PM through Saturday around 4AM Pacific Daylight Time (GMT-7).  This time frame is approximate at best.

     The mate to the drive that failed developed two bad sectors right after this drive was replaced so I guess there is some value to the recommendation that you not buy drives in a RAID array from the same place since they will likely be manufactured at the same date and thus prone to simultaneous failure.

     There also seems to be a firmware bug with these particular drives, they have two spare tracks for sector re-allocation but neither of them automatically re-allocated the failed sectors.  I don’t have a spare now so have one on order but probably won’t make it in time for Friday’s maintenance.

     This Friday, unless the spare arrives, the primary maintenance will be installing this new flash drive and copying existing data over to it.  I don’t know how long this copy will take which is primarily why the uncertainty of the time interval.

     If the replacement drive arrives by then I’m going to change out the drive as well even though it only has two flawed sectors.  The firmware is supposed to handle automatic re-assignment internally, but two drives of the same model failed to do so, so I assume this is a firmware bug.  The new drives have a 4x larger cache anyway so worth replacing from a performance standpoint anyway.

     If the drive does not arrive in time for replacement Friday, then when it does arrive I’m going to attempt to manually force re-assignment, but I don’t want to do this until I have a spare just in case I brick the drive.  At any rate, I will replace at the next time it is convenient to do a maintenance on a Friday night.