All,
Well, unmounting our RAID (which is using the disk) is not
acceptable, since there will be a bunch of NFS and Samba
shares hanging off it at all times. So (groan) I guess I do
need the latest and greatest. Can someone point me to the
right recipe for going straight from RedHat 6.0 (2.2.5-15) to
the best stuff?
More...
At 01:42 AM 9/14/99 +0100, Paul Jakma wrote:
>On Mon, 13 Sep 1999, Lawrence Dickson wrote:
>
> >1. unmount everything that's directly on the disk that isn't raid.
> >2. swapoff if you have swap on that disk
>
> Didn't understand this. Our system is on hda and we can't
> unmount it.
>
>indeed. you can't do it on the disk with the root-device on it.
>However you should be able to do it if you have root-raid setup. Ie
>root on a RAID-1 device and boot the kernel from a non-significant
>device like a floppy or an old spare ide disk.
>
> >3:
> >
> >echo "scsi remove-single-device c b t l" > /proc/scsi/scsi
>
> Tried it and it didn't work. But I noticed /proc/mdstat had
> an (F) after sdo1. So I tried from another terminal
>
> raidhotremove /dev/md0 /dev/sdo1
>
>forgot to mention this above.. absolutely nothing must be using the
>disk, no mounted partitions, no swap active, not being used by
>md/lvm.. etc.
So I take it the 18 disk RAID has to be unmounted before I
can do this. That's a killer.
>
> which hung THAT terminal and gave me a D in ps ax.
>
> THEN I tried
>
> echo "scsi remove-single-device 2 0 2 0" >/proc/scsi/scsi
>
> and that worked; it was gone from /proc/scsi/scsi.
>
>[snip]
>
> >
> >We never.. never.. never.. ever.. reboot linux!!
>
>a bit tongue in cheek this.. :)
>
> I guess we will have to - which is really a bummer. We were
> hoping to make our RAID product more friendly to customers.
>
>The latest raid patches are pretty much configure, take notes and
>forget. It'll take care of itself really.
>
> Larry Dickson
>
> PS. We would really appreciate, if possible, some route that
> does NOT require massive kernel recompiles and leading-edge
> alpha code.
>
>If you're reffering to hot-swapping SCSI disks, then it does work.
>I've used the above method quite a few times and it works for me.
>Although in every case it was for disks with bad-blocks (ie the
>drives interface logic was fine). But if the electronics are gone
>funny and the hardware won't co-operate there's no way around a
>reboot.
>
>Caveat: I've only tried on servers with hot-swap bays and drive
>caddies. Hot-(un)plugging plain SCSI cable connectors probably won't
>work, there's even a (smallish) chance of damaging the device being
>swapped and other devices on the same bus (including the controller).
We have proper hot-swap hardware but that does not help when
the scsi driver forces us into uninterruptible sleep.
Thanks,
Larry
>
>As for Alpha code, the latest raid patches are the ones to use for
>production machines. I don't understand why they've not been folded
>back into the main kernel source. But make sure to use the latest
>RAID patches, as they are the ones with the best hot-swap support.
>
>regards,
>--
>Paul Jakma
>[EMAIL PROTECTED] http://hibernia.clubi.ie
>PGP5 key: http://www.clubi.ie/jakma/publickey.txt
>-------------------------------------------
>Fortune:
>linux: because a PC is a terrible thing to waste
>([EMAIL PROTECTED] put this on Tshirts in '93)
>
>
