This is part of The Pile, a partial archive of some open source mailing lists and newsgroups.
Subject: Re: How to ADD new HDD ?
From: Jan Carlson <janc@iname.com>
Date: Thu, 01 Jul 1999 16:32:06 -0400
Manoj Saxena wrote:
>
> We have a Redhat 5.2 server installed with 9Gb SCSI hard disk. Due
> to huge volume of data we want to add another 9Gb SCSI hard disk.
> This is a brand new unformatted hard disk.
>
> So please let me know how can I add it with no risk to the
> present running server. Though I can shut it down at times.
Give the new disk a higher SCSI ID number than
the existing ones, or the /dev/sdxx entry for existing disks
WILL CHANGE. That will force you to fix EXISTING entries
in /etc/fstab and /etc/lilo.conf, which is no big deal if you
are aware it will be necessary.
If the new disk is the LOWEST SCSI ID, your machine will not boot,
unless you do tricks in the scsi card bios to make the old
disk the 'boot' drive again, or you reinstall lilo in the
drive that becomes the first, or 'boot' drive.
/dev/sda is the lowest SCSI ID NUMBER
/dev/sdb is the next lowest, etc.
Make sure:
. there is a SCSI device at the end of each scsi cable
. every SCSI device at the end of a cable is terminated
. every SCSI device NOT at the end of a cable is NOT terminated
. If there is only one scsi cable, the scsi card must have
termination enabled or automatic.
. If there is more than one scsi cable, see instructions for
the scsi card regarding termination on the card.
===
Subject: Re: Mounting problems
From: Steve Borho <sborho@ststech.com>
Date: Mon, 14 Jun 1999 14:36:20 -0500
On Mon, Jun 14, 1999 at 01:55:36AM -0700, Mike A. Lewis, CNE wrote:
> RH6.0
>
> I'm trying to add a new SCSI hard drive that has an extended partition and
> two logical partitions. I've formatted the hard drive with no problems, yet
> when I go to mount it with :
>
> mount -t ext2 /dev/sdb5 /mnt/part1
>
> the system hangs, does not give a prompt or anything. I can go to another
> console and do a 'ps ax' and get the following:
>
> 481 tty1 D 0:00 mount -t ext2 /dev/sdb5 /mnt/part1
if you run dmesg, are you seeing any SCSI errors?
===
Subject: Re: adaptec 7895 not recognized?
From: Jan Carlson <janc@iname.com>
Date: Thu, 17 Jun 1999 21:02:26 -0400
Don Brown wrote:
>
> >Don Brown wrote:
> >>
> >> It seems that neither 5.2 or 6.0 Redhat support the adaptec 7895 or
> >> 7896 scsi adapters.
>
> >Those are definitely supported.
>
> Yes, but how would I go about installing RH6.0 from the boot disks. I can't get linux
> loaded on the drive in order to put the module in place?
> My apologies if this is a stupid question. It is something that I am unfamiliar with.
Those cards are listed at RH as "Tier 1 Supported" which means they
are known to work well for installation, etc. so it seems they should
be detected automatically.
Here are some educated guesses at what might help:
1. try the 'other' BIOS setting for "PLUG AND PLAY".
2. remove any other cards that are not vital during installation,
like sound, ethernet, modem, etc.
I tried starting an install with the new updated
boot.img, supp.img, and modules.img floppies, typing "linux supp"
as directed in the README. It simply autodetected my NCR scsi card,
so it's hard to say, but indeed I did not see any opportunity to
choose 'aic7xxx' which is the name of the driver.
Maybe then you get that choice if you type 'expert supp' instead of 'linux supp'.
In any case, since your scsi card is on the Tier 1 supported list,
RH seems to promise to help you if you bought an official boxed set.
The kernel source for that aic7xxx driver has these comments:
aic7xxx.c: * Form: aic7xxx=extended
aic7xxx.c: * aic7xxx=no_reset
aic7xxx.c: * aic7xxx=ultra
aic7xxx.c: * aic7xxx=irq_trigger:[0,1] # 0 edge, 1 level
aic7xxx.c: * aic7xxx=verbose
aic7xxx.c: * append="aic7xxx=verbose,tag_info:{{32,32,32,32},{32,32,32,32}}"
aic7xxx.c: * ie. aic7xxx=unpause:0x0A,extended
which tell me it might help to type
linux aic7xxx=extended
at the initial boot floppy prompt, as well as the other options listed.
Maybe one of them will work, but boy, this is stabbing in the dark.
I hope some of this helps though.
===
Subject: Changing from IDE to SCSI on working RH6.0 system
From: Darryl Harvey <dharvey@connect.com.au>
Date: Tue, 14 Sep 1999 12:32:58 +1000
I have a working RH6.0 system running off 1 x IDE drive and 1 x SCSI drive.
It currently boots from the IDE Drive (/dev/hda) and the partitions are as
follows:
/dev/hda1 /boot
/dev/hda5 /usr
/dev/hda6 /tmp
/dev/hda7 /var
/dev/hda8 SWAP PARTITION
/dev/hda9 /
/dev/sdb1 /home <-- existing SCSI partition to stay.
My Aim:
I want to change this so that the whole machine runs off 2 x SCSI drives,
and I want to change the IDE to run WinDoze.
The machine will boot directly off the 1st SCSI drive (SCSI ID: 0) and LILO
will allow me to boot windoze from the IDE drive (/dev/hda I hope)
/dev/sda = SCSI ID 0
/dev/sdb = SCSI ID 1
to boot from /dev/sda
/dev/sda1 /boot
/dev/sda5 /usr
/dev/sda6 /tmp
/dev/sda7 /var
/dev/sda8 SWAP PARTITION
/dev/sda9 /
/dev/sdb1 /home <-- existing SCSI partition
/dev/hda will be Windows only (at this stage)
How do I convert the system to the new drive ? I am getting stuck and
have nearly lost my booting capabilities several times.
Time to call on help.
I have the system booting off /dev/sda with a new install of RH6.0, but I
want to copy the existing setup and configs of the old system over the new
install, so all is "as it was" but on the new drives.
So as of now, I am booting a new system with /dev/sda as the boot device,
/dev/sdb is mounted as /home and does not need to be touched. but all the
setups and filesystems from /dev/hda need to be put on /dev/sda in the
right spot, but I am having trouble getting them on the system.
/dev/hda is available to me (I have not reformatted yet) and mountable.
What should I do in what sequence to make this conversion work?
===
Subject: RE: Changing from IDE to SCSI on working RH6.0 system
From: Gregory Hosler <gregory.hosler@eno.ericsson.se>
Date: Tue, 14 Sep 1999 13:42:49 +0800 (SGT)
One method (certainly not the only one) is to boot the old system,
mount all the new partitions starting under (for example) /mnt, and
then tar copy thru a pipe (excluding /mnt and /proc, of course).
This method obviates the need to install the new system (which you say
you have already done).
OTOH, to do as you state, i.e. copy only the setup and config files, you
will need to know which files those are, and manually copy/merge then
over/into your new files. Someties it is easier to just copy the whole
system (assuming that your were previously satisfied w/ the former
installation), then to hit/mis the config files.
===
Subject: RE: Changing from IDE to SCSI on working RH6.0 system
From: Jerry Winegarden <jbw@oit.duke.edu>
Date: Tue, 14 Sep 1999 09:20:42 -0400 (EDT)
Greg Hosler asked about adding another scsi drive and moving file
systems around. Others suggested using a "tar pipe" to move the file
systems.
The tar pipe command to move a directory (in this case a whole file
system from its mount point) is in the man page of my old DEC Ultrix
system (perhaps the only useful example of how to USE a command
I've ever seen in a man page!;-), but is not there in the GNU tar man page.
It's simple:
(Make sure the desired file system is already mounted on "todir")
cd fromdir; tar cvpf - . | (cd todir; tar xvpf -)
(the above can be done without the v verbose option above, but I like
to see what's going on to make sure I didn't type the wrong fromdir, etc.)
The first tar defines it's output file as - which means standard output
which means the pipe in this case. The . is the directory to tar up,
which of course means the current directory.
The paraentheses around the two statements on the right side of the pipe,
(separated by the comma) lets you change directory to the destination
and then unpack, as part of the flow through the pipe.
The file spec of - in the second tar command tells it to take its
input (it's an "x" (unpack) command) from the file - which means
standard input, which here means the pipe. Way cool!
===
Subject: Hot-swap SCSI drives
From: Eric Sisler <esisler@westminster.lib.co.us>
Date: Tue, 21 Sep 1999 12:51:31 -0600
Greetings all,
I have an HP NetServer with hot-swap SCSI drives and I was wondering if
there's any way to get the kernel to remove a SCSI device from its table
without rebooting. From the proc man page:
"An echo 'scsi singledevice 1 0 5 0' > /proc/scsi/scsi will cause host
scsi1 to scan on SCSI channel 0 for a device on ID 5 LUN 0. If there is
already a device known on this address or the address is invalid an error
will be returned."
To illustrate my point, let's say I want to replace a 4Gb drive with a new
9Gb drive, but I want to use the same hot-swap tray. The physical drive
sizes are obviously different and the partition tables will be as well. Is
there a way to remove the partition table & SCSI device from the kernel,
remove the old drive, install the new drive and issue the above command so
the kernel will "find" the new drive? Or does this require a reboot?
BTW - Unloading / reloading the SCSI driver unfortunately isn't an option -
the boot disk is also SCSI and uses the same driver. :-(
===
Subject: external SCSI id selector Q
From: Charles Galpin <cgalpin@lighthouse-software.com>
Date: Tue, 21 Sep 1999 18:32:18 -0400 (EDT)
Hi
I have a SCSI hard drive I want to put into an external enclosure. It
comes with a cable to connect an exernal id selector to the jumpers on the
drive. Other than trial and error, does anyone know of a generic ordering
or color coding for this?
The cable has a flat/wide 5 pin conector on one end that fits on the id
selector switch (which way I'm not sure), and then for connectors on the
other end. These each have two pin holes, but only one has two wires
going to it ( red and brown ), the other colors are yellow, blue and grey.
On the flat end, they are in this order - brown, red,yellow, blue, grey.
My drive is an IBM Ultrastar DDRS-39130, which has 4 pairs of jumpers to
control the id.
The enclosure is a DT3F-XXX made by JMR Electronics (www.jmr.com i think),
but their website doesn't list this model.
===
Subject: Re: How can you test the performance of scsi drives?
From: Steve Borho <steve@borho.myip.org>
Date: Tue, 5 Oct 1999 15:57:20 -0500
On Tue, Oct 05, 1999 at 04:10:59PM -0500, johnsj20 wrote:
> I like using hdparm for ide drives. Is there anything that I can use for
> my scsi hd drives?
errr... hdparm?
hdparm can query SCSI drive statistics, but it can't tweak IDE settings
for obvious reasons.
===
Subject: Re: How can you test the performance of scsi drives?
From: Randy Carpenter <rcarpen@celt.network1.net>
Date: Tue, 5 Oct 1999 21:33:02 -0400 (EDT)
> On Tue, 5 Oct 1999, johnsj20 wrote:
>
> > I like using hdparm for ide drives. Is there anything that I can use for
> > my scsi hd drives?
What's wrong with hdparm? -t and -T work fine for SCSI drives.
===