Les Mikesell wrote:
On Fri, 2006-05-26 at 16:56, Barry L. Kline wrote:
Note that unless it has changed recently, SATA drives don't pass errors up to the software raid layer correctly. If a drive dies it doesn't get kicked out as it should.
Would you please elaborate as this seems to run counter to my experience?
I have had 3 drive fail events with software raid (md) tools and SATA drives in the last year. The drive was taken off-line and I as the sysadm received a warning message via email.
I downed the systems (after-hours) and replaced the failed drive. (These machines don't have hot- swappable h/w) No problems were encountered.
Two of these systems are running a WBEL-3 and a 2.4 kernel and haven't been updated in a while. The other was a CentOS-3, same story.
Regards Ray