Perc 5 driving me crazy

Steve Thompson smt at
Thu Apr 12 18:50:17 CDT 2012

CentOS 5.7, x86_64, OMSA 6.5.

I have a PE1900 with a Perc 5/E card, to one port of which is connected 
two MD1000 shelves (daisy chained). In one shelf is 15 x 1 TB SATA drives 
(sourced from Dell) and in the second shelf is 15 x 2 TB SATA drives 
(Seagates). Interposers are used in all 30 slots. Each shelf contains two 
7-member RAID5 arrays and one global hot spare. All temps, etc are fine.

The system will run like a champ for anything from a week to several 
months, and then will mark anywhere from four to eight drives as failed 
(there is nothing wrong with them). Many times there are several failed 
drives in one or more RAID5 arrays, so its contents are lost. Power cycle 
the system, and many drives are marked as foreign. Clear the foreign 
config, rebuild the RAID arrays, and then everything is fine until the 
next time.

I am seeking enlightenment as to what I can do to make this thing work 
consistently, because I'm about to take an axe to it. The alert log does 
show some SCSI sense events, and the Perc appears to be overly sensitive 
to these, when nothing is wrong with any of the disks. Solution: replace 
the Perc? Use software RAID? Seek another line of work?


More information about the Linux-PowerEdge mailing list