Re: Redhat 9.0 Rebuild raid 5 array

[Mark Farmer <farmerma cromwell co uk>]

Reuben D. Budiardja wrote:
> On Monday 11 October 2004 08:27, Mark Farmer wrote:
>
> > Brian McGrew wrote:
> >
> > > I've got the following:
> > >
> > > Personalities : [raid1] [raid5]
> > > read_ahead 1024 sectors
> > > md2 : active raid5 hde1[0] hdf1[1] hdg1[2](F) hdh1[3]
> > >      468864768 blocks level 5, 64k chunk, algorithm 0 [4/3] [UU_U]
>
>
> I've got similar problem (and it's interesting that it's not covered in 
> Software RAID Howto).
>
>
> > raidhotremove will allow you to remove the failed disk
> > raidhotadd will put the new disk into the array.
> >
> > If I remember correctly, the array will rebuild itself once the new disk is
> > added.
>
>
> But do I need to format the harddrive file with fdisk before doing raidhotadd 
> or anything like that ? or can I just plug in a new harddrive directly from 
> the factory to the machine? 
> If I do "fdisk /dev/hde1" and list the partiton (where hde1 is part of md0), I 
> see that the partiton type is "Linux RAID autodetect". So I don't know if 
> "raidhotadd' will take care of this or what (ie. creating and formating 
> partition).
>
> Thanks.
>
> RDB

Once you have replaced the disk, you'll need to use fdisk to recreate the partion table & set 
partion types. Raid partions MUST be set to "Linux Raid Autodetect"

This is freaky, seeing the original post prompted me to check one of our servers and hey 
presto..... FAILED DRIVE!  Don't know how long it's been down so thanks!

--
Mark Farmer
Linux System Administrator