Re: veryPC, Seagate, raid - caught in the middle
Thanks Christian. It's really bugging me as I can't understand why 2 months ago one drive had problems and replacing it fixed those apparently. But then two weeks ago a different drive suddenly goes down. Maybe that series of Seagate drives are "evil "? But then.. I did stress test the broken drive (the one from 2 months ago) heavily using a USB controller and it hasn't shown any problems so far (using badblocks and Seatools).
Maybe it's a strange combination of factors as an i2c module I inserted after running sensors-detect was spamming my syslog with this message:
i801_smbus 0000:00:1f.3: Bus collision! SMBus may be locked until next hard reset. (sorry!)
Yesterday I took out what I believe to be the culprit module from /etc/modules, rebooted, rebuilt the md2, started stressing the drives and so far so good. But then the problems seem to only show after a week or two...
I don't know enough about i2c or any of this but could their be a connection between the i2c part of the machine and the SATA controller?!
On Friday 06 August 2010 12:11:03 Christian Svensson wrote:
> I've seen this before when one drive was about to crash.
> One array of my three software RAID-5 arrays would randomly fail - same
> member drive though.
> dmesg would log a failed write and that particular partition would be kicked
> out of the array.
> Replacing the drive solved the problem for me.
> Christian Svensson
> System administrator - Kalmar NDC