couple of seemingly dumb questions

couple of seemingly dumb questions

[Maurice Hilarius]

On software RAID/raidtools how does one change the rebuild priority on a 
replacement disk or resync?
I don't mind giving higher CPU priority, since in many cases the boxes I am 
using have lots more CPU than is being used most of the time.

Secondly:
With a software RAID5 setup, with disks in removable hot-swap trays, I want 
to be able to pull a failed disk and replace it and have it then rebuild. 
Problem is, when I pull the disk I get lots of error messages, and then 
when I replace it there is nothing happening to redetect that the disk is 
changed, and allow me to rebuild. If I reboot then the machine does resync 
with the new drive, but surely there must be a way without a reboot?

Thanks all for suggestions and info!

With our best regards,

Maurice W. Hilarius       Telephone: 01-780-456-9771
Hard Data Ltd.               FAX:       01-780-456-9772
11060 - 166 Avenue        mailto:maurice@harddata.com
Edmonton, AB, Canada      http://www.harddata.com/
    T5X 1Y3

Ask me about the UP1500 Alpha - Full systems from $3,500!

Re: couple of seemingly dumb questions

[Danilo Godec]

On Thu, 18 Apr 2002, Maurice Hilarius wrote:

> On software RAID/raidtools how does one change the rebuild priority on a
> replacement disk or resync?
> I don't mind giving higher CPU priority, since in many cases the boxes I am
> using have lots more CPU than is being used most of the time.

With 2.2 kernels, you can simply use 'cat /proc/sys/dev/md/speed-limit',
to see the *minimum* rebuild rate. You can change it using
'echo 200000 > /proc/sys/dev/md/speed-limit'.

With 2.4 kernels, there are two files in '/proc/sys/dev/raid/', called
'speed_limit_max' and 'speed_limit_min'. I think the rest is pretty
obvious.

> Secondly:
> With a software RAID5 setup, with disks in removable hot-swap trays, I want
> to be able to pull a failed disk and replace it and have it then rebuild.
> Problem is, when I pull the disk I get lots of error messages, and then
> when I replace it there is nothing happening to redetect that the disk is
> changed, and allow me to rebuild. If I reboot then the machine does resync
> with the new drive, but surely there must be a way without a reboot?

I suppose you have SCSI disks (I think hot-swap IDE is just not that far
along).

First you have to make sure, that the failed disk is not in use anymore.
This means using 'raidhotremove' for every partition of that disk, that
may be in use (also check for non-RAID partition, if they exist). When
you're sure the disk is not used anymore, make the kernel 'release' the
disk from SCSI bus using:

echo "scsi remove-single-device 0 1 2 3" >/proc/scsi/scsi

Check 'cat /proc/scsi/scsi' to check if the failed disk is still listed.
If it is, it's probably still in use and can't yet be removed. Re-check
the first step... Else, you can pull the disk out and install a new one.

Before you can use the new disk, you have to make the kernel accept it to
SCSI. Use:

echo "scsi add-single-device 0 1 2 3" >/proc/scsi/scsi

Check 'cat /proc/scsi/scsi' if the new disk has appeared. If it has,
you're all set to partition the disk and use it at will.

   D.

PS: See /usr/src/linux/drivers/scsi/scsi.c for more about this
/proc/scsi/scsi commands.

Re: couple of seemingly dumb questions

[bo]

Danilo,

If I have a h/w support for IDE hot swap then what do you recommand for
the same as you suggested for scsi below?

Thansk,

Bo
----- Original Message -----
From: "Danilo Godec" <danci@agenda.si>
To: "Maurice Hilarius" <maurice@harddata.com>
Cc: "Linux-raid" <linux-raid@vger.kernel.org>
Sent: Thursday, April 18, 2002 11:20 PM
Subject: Re: couple of seemingly dumb questions


> On Thu, 18 Apr 2002, Maurice Hilarius wrote:
>
> > On software RAID/raidtools how does one change the rebuild priority on a
> > replacement disk or resync?
> > I don't mind giving higher CPU priority, since in many cases the boxes I
am
> > using have lots more CPU than is being used most of the time.
>
> With 2.2 kernels, you can simply use 'cat /proc/sys/dev/md/speed-limit',
> to see the *minimum* rebuild rate. You can change it using
> 'echo 200000 > /proc/sys/dev/md/speed-limit'.
>
> With 2.4 kernels, there are two files in '/proc/sys/dev/raid/', called
> 'speed_limit_max' and 'speed_limit_min'. I think the rest is pretty
> obvious.
>
> > Secondly:
> > With a software RAID5 setup, with disks in removable hot-swap trays, I
want
> > to be able to pull a failed disk and replace it and have it then
rebuild.
> > Problem is, when I pull the disk I get lots of error messages, and then
> > when I replace it there is nothing happening to redetect that the disk
is
> > changed, and allow me to rebuild. If I reboot then the machine does
resync
> > with the new drive, but surely there must be a way without a reboot?
>
> I suppose you have SCSI disks (I think hot-swap IDE is just not that far
> along).
>
> First you have to make sure, that the failed disk is not in use anymore.
> This means using 'raidhotremove' for every partition of that disk, that
> may be in use (also check for non-RAID partition, if they exist). When
> you're sure the disk is not used anymore, make the kernel 'release' the
> disk from SCSI bus using:
>
> echo "scsi remove-single-device 0 1 2 3" >/proc/scsi/scsi
>
> Check 'cat /proc/scsi/scsi' to check if the failed disk is still listed.
> If it is, it's probably still in use and can't yet be removed. Re-check
> the first step... Else, you can pull the disk out and install a new one.
>
> Before you can use the new disk, you have to make the kernel accept it to
> SCSI. Use:
>
> echo "scsi add-single-device 0 1 2 3" >/proc/scsi/scsi
>
> Check 'cat /proc/scsi/scsi' if the new disk has appeared. If it has,
> you're all set to partition the disk and use it at will.
>
>    D.
>
> PS: See /usr/src/linux/drivers/scsi/scsi.c for more about this
> /proc/scsi/scsi commands.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: couple of seemingly dumb questions

[Maurice Hilarius]

With regards to your message at 09:49 AM 4/19/02, bo. Where you stated:
>Danilo,
>
>If I have a h/w support for IDE hot swap then what do you recommand for
>the same as you suggested for scsi below?
The issue seems to be one of "while the hardware supports replacing drives 
and bus rescans, the Linux IDE driver may not"

Or am I completely missing the point here???


With our best regards,

Maurice W. Hilarius       Telephone: 01-780-456-9771
Hard Data Ltd.               FAX:       01-780-456-9772
11060 - 166 Avenue        mailto:maurice@harddata.com
Edmonton, AB, Canada      http://www.harddata.com/
    T5X 1Y3

Ask me about the UP1500 Alpha - Full systems from $3,500!

Re: couple of seemingly dumb questions

[Danilo Godec]

On Fri, 19 Apr 2002, bo wrote:

> Danilo,
>
> If I have a h/w support for IDE hot swap then what do you recommand for
> the same as you suggested for scsi below?

I'm not sure, but I guess such H/W support for IDE hot swap would include
an IDE RAID card that would probably handle such things by itself and
kernel wouldn't use disks directly. Instead the RAID card will provide
some 'pseudo' devices which will appear as disks to the kernel.

Most probably this is handled by the special IDE RAID card drivers that
exists in late kernels (Promise Fasttrak).

But if you just want 'normal' hot-swap IDE disks on Linux, you're out of
luck (somebody, correct me if I'm wrong). You could also check out
http://www.linux-ide.org/, where I saw something about IDE hot-swap...

  D.

Re: couple of seemingly dumb questions

[Danilo Godec]

On Fri, 19 Apr 2002, Maurice Hilarius wrote:

> >If I have a h/w support for IDE hot swap then what do you recommand for
> >the same as you suggested for scsi below?
> The issue seems to be one of "while the hardware supports replacing drives
> and bus rescans, the Linux IDE driver may not"
>
> Or am I completely missing the point here???

Something like that. However, IDE RAID cards *may* handle the issue by
providing a 'pseudo' device acting like a disk to the kernel. I guess the
kernel does need special support (a driver) for such card(s).

In a couple of days I may know more about it as I just got a motherboard
with an on-board Promise Fasttrak... :)

  D.