Re: Soft Raid5 Problems
hi ya nik
if you think the hard disk is still working...
you can confirm it by:
a) check for any hda problems in /var/log/*
replace hda with the suspect drive that
is claimed to be bad by the raid5 sw
-you'd have some checking/gambling to do to see
if you have a bad disk or just an erroneous
nit-picking software that thinks it bad when its not
- try replacing the IDE cables... use good quality
SHORT cables... one disk per cable...preferrably
only 2 ide connector ( one for mb and one for disk )
- dont use the old style 40pin ata-33 cables...
( use the newer 80-pin ata-66/ata-100 cables )
b) hdparm -tT /dev/hda
- to see if you can at least read some bios info
off the disk controller on the disk itself
c) take the disk to another linux box and
install linux on it... do a disk stress test on it
- assumes that you have your raid5 data backed
up and dont care about the one presumed "dead disk"
d) if the disk is working in the other boxes...
plug it back in with raidhotadd and presumably
you did a raidhotremove to take it out...
( might need raidsetfaulty too before removing )
( but it already knew that since it thinks its a bad disk
( and not listed in /proc/mdstats
-- when you plug the disk back in... you do NOT format it...
- the software raid will do its magic and resync
it back into its "sw raid5"
-- you only mke2fs /dev/md0 once at the beginning
c ya
alvin
On Thu, 25 Apr 2002, Nik Engel wrote:
> Hi Alvin!
> > if you do have your raid intact... and just a dead disk
> > than you can take that disk out and add a new one
> >
> > -- if you were manually able to recreate /dev/md0
> > -- backup all your data now to some other disks
> > -- on your lan
> >
> > -- than continue fixing your raid box
> >
> > raidhotremove /dev/md0 /dev/hdc1
> > -- power down your system ...
> > -- swap out the bad disks ( /dev/hdc )
> > -- put the new replacement disk in
> >
> > -- power up as normal...
> > -- note that it will be running in degraded mode...
> > -- keep an eye on /proc/mdstat for "resync status"
> >
> > when its done with its housekeeping... i'd add
> > the replacement disks ...
> > raidhotadd /dev/md0 /dev/sdc1
> >
> > - watch it resync again...
> > and i'd power down and back up one ore time for good measure
> >
> Thank you very much, very detailed help, only one question remaining:
> as i am able to format "dead" disk under the running os -> is there a
> way try to sync manually ? or how can i check the dead disk, if it is
> really dead ?
> As i don't want to buy new disks as long as i am not quite sure if it is
> defekt ?
--
To UNSUBSCRIBE, email to debian-user-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: