[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: raid 1 recovery



HDA
----
hda1 /        ext2
hda2 /swap
hda3 /home    fd filetyepe

HDC
---
hdc1 /        ext2
hdc2 /swap
hdc3 /home    fd filetype

/dev/md0:  hda3 + hdc3


/etc/raidtab:
raiddev /dev/md0
        raid-level 1
        nr-raid-disks 2
        nr-spare-disks 0
        chunk-size 4
        persistent-superblock 1
        device /dev/hda3
        raid-device 0
        device /dev/hdc3
        raid-device 1


the root partition, i do a manual rsync from hda to  hdc.

if i unplug ide cable from hda, the system can't find
/dev/hda to boot, and showing 04 04 04...


BTW, how to mkraid for root partition?
The partition is always mounted and i can do a mkraid to it.

David


On Sat, 12 Oct 2002 23:43:10 -0700 (PDT), Alvin Oga wrote
> hi ya
> 
> if raid1 mirroring is properly setup...
> 	- make sure your partition types are all "FD" type
> 
> - what does your /etc/raidtab look like ?
> 
> the system WILL boot properly all by itself into /dev/hda or /dev/hdc
> when one of the disk has failed 
> ( simulate by pulling the power to it or the ide cable )
> 	- you do NOT need tomsrtbt or any other standalone boot media
> 	to fix one failed disk of a properly configured raid1
> 	(mirror) system
> 
> 	- there is probably a misconfiguration someplace if your system
> 	doesn't boot with just one disk ( of the 2 in the mirrored system)
> 
> -- first problem is to make sure the system keeps running even if a
>    disks died ... ( runs in degraded mode ... maybe use hdparm to 
> shut the   drive down ) 	and that it is still bootable w/ any one of 
> the two disks
> 
> -- resync is of the new disks is also done automatically
> 	hotswapremove, hotswapadd combination of commands
> 
> while the system is resyncing, if you lose the 2nd disk, you're dead...
> unless you start to manually build up the raid before it failed
> with things like mdctl
> 
> c ya
> alvin
> 
> On Sun, 13 Oct 2002, david hong wrote:
> 
> > there is a good article at linux journal about raid 1 setup
> > http://www.linuxjournal.com/article.php?sid=5898&mode=thread&order=0
> > 
> > I tested the suggested method, it works quite well.
> > machine still boot up ok if any hdc partition fail.
> > 
> > i tested many combinations of the failures, the difficult part
> > is still the boot disk fails. I simulate it by unplug the cable from hda
> > HD, while hdc still attach to my 2nd controller.
> > 
> > the hdc root partition is a resync copy from the orginal date from hda,
> > filesystem ext2.
> > 
> > i boot up from tomsrtbt bootable floppy and amended the hdc  /etc/fstab 
file
> > to point to hdc instaed of hda.
> > then i reboot it using a mkbootdisk floppy, and do a LILO:linux 
root=/dev/hdc1
> > system boot up with all the superblock warnings....
> > 
> > what is the fastest way to recover if bootdisk (hda) fail?
> > 
> > (if did try move the hdc to primary controller, then i see
> > superblock warnings again....)
> > 
> > 
> > 
> > > with raid 0.90 it may be different, what does cat /proc/mdstat
> > > show when you only have 1 disk?  
> > it display the default content. no md* active.
> > 
> > David
> > 
> > 
> > 
> > 
> > 
> > 
> > 
> > 
> > On Fri, 11 Oct 2002 20:33:59 -0700 (PDT), nate wrote
> > > david hong said:
> > > 
> > > > if my hda (boot disk) fails, can i boot up from a
> > > > mini bootable floppy and mount the hdc partition
> > > > and recover the data?
> > > >
> > > > i tried that, but it keeps complaining filetype error
> > > > when i try to mount the hdc partition.
> > > 
> > > I haven't tried(or needed to) raid recovery with raid 0.90
> > > but I have with raid 0.36(raid1). This uses normal linux partition
> > > types instead of fd, one of my drives died and I switched
> > > lilo and friends over to the seocnd disk, disabled raid and continued
> > > to use the system for another few months till the 2nd disk died.
> > > 
> > > with raid 0.90 it may be different, what does cat /proc/mdstat
> > > show when you only have 1 disk?  also be sure you have raid support
> > > on that floppy, the new raid may "need" it to access the disk(perhaps
> > > it is in a slightly different format which the raid tools provide
> > > transparent access to).
> > > 
> > > nate
> > > 
> > > -- 
> > > To UNSUBSCRIBE, email to debian-user-request@lists.debian.org 
> > > with a subject of "unsubscribe". Trouble? Contact 
> > listmaster@lists.debian.org
> > 
> > 
> > 
> > 
> > -- 
> > To UNSUBSCRIBE, email to debian-user-request@lists.debian.org 
> > with a subject of "unsubscribe". Trouble? Contact 
listmaster@lists.debian.org
> >





Reply to: