Re: Raid trouble
On Sat, Jan 27, 2007 at 09:41:59AM -0500, Roberto C. Sanchez wrote:
> On Sat, Jan 27, 2007 at 02:26:01AM -0800, Land Haj wrote:
> > Hi!
> >
> > I decided to try etch out and use raid-1. All went fine until I mounted one of the raid disks from another debian installation on the same computer, to copy a library. My set up looks like this:
> >
> > sda ---> / in a sid installation
> > sdb ---> /home in the sid installation
> > sdc ---> first disk in the raid cluster where the etch installation is
> > sdd ---> second disk in the raid cluster where the etch installation is
> >
> > When on the sid installation, I did this:
> >
> > mount /dev/sdc3 /mnt/raid
> > cp -a /mnt/raid/me/photolibrary /home/anotherme/
> > umount /mnt/raid
> >
> > When I logged back into my etch+raid installation, I had massive fs errors and had to do fsck manually. Now my question is: did I cause this simply by mounting one of the raid partitions, or did I encounter a bug in etch?
> >
> > If I casued it on my own, what is the proper way to copy information from a raid partition when on another installation?
> >
> You can't just mount a raid partition and have it work. You have to
> start the raid first (You can force it to start in degraded mode of you
> only have one disk of the pair) and then you can mount the /dev/md?
> wherever you like. What you did was really Not Good (TM) for you RAID
> partition.
>
But shouldn't Etch's raid system have noticed an inconsistancy in the
data at the raid level and rebuilt the array?
Doug.
Reply to: