[mythtv-users] Another take on large-capacity disk enclosures

[Steve Heistand]

Robin Hill <myth at robinhill.me.uk>, said on Tue Jun 16, 2009 [11:12:34 PM]:
} On Tue Jun 16, 2009 at 11:11:07PM +0100, Robin Hill wrote:
} 
} > On Tue Jun 16, 2009 at 12:10:25PM -0700, Yeechang Lee wrote:
} > 
} > > Steve Heistand <steve at heistand.org> says:
} > > > any journaled filesystem is easy to fsck, its the rebuilding a drive
} > > > that takes a while these days. 20 drives would take a while to init
} > > > and or rebuild if you had to. not like days of course but unless you
} > > > crank up the cpu/bandwidth limits way up it would be 10+ hours.
} > > 
} > > Building, yes; said 6.6TB/16 takes 4-6 hours, as mdadm (still!)
} > > doesn't use more than one core.
} > > 
} > 
} > I'd have a serious look at your system if the mdadm build is CPU
} > limited.  Every system I've used over the last few years has been I/O
} > limited for any mdadm operation.
} > 
} Oh, and just to add - mdadm is actually doing very little here.  The
} array construction/rebuild is all done by the md code within the kernel.
} 

tweaking the settings in the files in : /proc/sys/dev/raid
can turn it into a nice md5sum limiting operation.
or make your system unusable depending on how high
you go..


-- 
"Why is it so hot inside this handbasket?"
--
Steve Heistand
steve at heistand.org