Perhaps someone could offer some help...

The system:
Sun Ultra30 running Solaris 8, Seagate 36GB (ST-136403LC) in slot 0,
original equipment CDROM in slot 6; used as a file sharing and print
server.

The problem:
I would like to add a new drive (Seagate 50GB ST-150176LC) to slot 1;
when I install the device in the machine and perform a reconfiguration
boot, I get the following warnings:

WARNING: /pci@1f,4000/scsi@3 (glm0)
SCSI bus DATA IN phase parity error
Target 1 reducing sync. transfer rate
SCSI bus DATA IN phase parity error
Target 1 disabled wide SCSI mode
Target 1 reverting to async. mode

What I've tried:
Removing the drive, inspecting the pins on the 80-pin connectors on
both the drive and the Ultra30 backplane for bent or otherwise mangled
components
Formatting the drive, which succeeded 100%, but took 20 hours (format,
verify pass 0, and verify pass 1); the format utility told me that it
would take 185 minutes.
Looking through the archives at length for anything remotely
resembling this problem.
Research at sun.com and seagate.com, as well as Seagate Customer
Service, which was nearly useless, since they wouldn't give me any
information.

Additional relevant information:
First, the original configuration of the system included a 10GB
Seagate Barracuda (slot 0) and the 36GB drive (formerly in slot 1),
and there were no warnings generated. The root partition on the 10GB
was transferred to the 36GB to facilitate the addition of the 50GB
drive (Thank you to Logan Shaw and Michael Tosch for the information
on moving file systems using 'ufsrestore').
Second, why can't every SCSI device manufacturer use the same
terminology to describe the type of interface on their units? I'm
having trouble identifying exactly which permutation of SCSI all of
the devices involved actually use. Nearest I can tell, the Sun bus is
a SCSI-3, Fast20, UltraSCSI unit. I don't know if this is low-voltage
differential or single-ended; the hard drives both claim to be Ultra2,
SCSI Wide, LVD units. The CDROM is SCSI-2.

The questions:
Will this drive work with this system?
Why do the 10GB and 36GB work, but not the 50GB?
How can I make it work?
Since it evokes no warnings, is the 36GB drive running in optimum
configuration (wide, LVD), and if so, what configuration is it?
Does the Ultra30 support low-voltage differential devices?
What am I missing?

Please be copious and verbose with your responses. I am at wit's end
here, both with understanding the problem and solving it.

Thank you
John Muller

On Sat, 15 May 2004 17:05:02 -0700, John Muller wrote:

> The questions:
> Will this drive work with this system?
> Why do the 10GB and 36GB work, but not the 50GB?
> How can I make it work?
> Since it evokes no warnings, is the 36GB drive running in optimum
> configuration (wide, LVD), and if so, what configuration is it?
> Does the Ultra30 support low-voltage differential devices?

The Ultra 30 doesn't support LVD and will run both disks in single-ended
mode. AFAIK the seagate disks will support this.

> What am I missing?

I get lots of hits when I copied your error messages to a google search;
looks like most people think it's a hardware issue...

Klaasjan

John Muller wrote:
> Perhaps someone could offer some help...
>
> The system:
> Sun Ultra30 running Solaris 8, Seagate 36GB (ST-136403LC) in slot 0,
> original equipment CDROM in slot 6; used as a file sharing and print
> server.
>
> The problem:
> I would like to add a new drive (Seagate 50GB ST-150176LC) to slot 1;
> when I install the device in the machine and perform a reconfiguration
> boot, I get the following warnings:
>
> WARNING: /pci@1f,4000/scsi@3 (glm0)
> SCSI bus DATA IN phase parity error
> Target 1 reducing sync. transfer rate
> SCSI bus DATA IN phase parity error
> Target 1 disabled wide SCSI mode
> Target 1 reverting to async. mode
>
> What I've tried:
> Removing the drive, inspecting the pins on the 80-pin connectors on
> both the drive and the Ultra30 backplane for bent or otherwise mangled
> components
> Formatting the drive, which succeeded 100%, but took 20 hours (format,
> verify pass 0, and verify pass 1); the format utility told me that it
> would take 185 minutes.
> Looking through the archives at length for anything remotely
> resembling this problem.
> Research at sun.com and seagate.com, as well as Seagate Customer
> Service, which was nearly useless, since they wouldn't give me any
> information.

John, what you have is a disk drive which is not sync-ing at
the same mode and transfer rate as the disk in slot 0. This
is a scsi interface issue.(By the way, you don't have a slot
6, your cdrom is using scsi target id 6).

You can workaround this issue by creating or editing a file
called /kernel/drv/glm.conf and tuning the scsi properties
for target 1 on that scsi bus. This is covered in the manpage
for "driver.conf" -- section 4. "Formatting" the drive is not
going to assist you with a scsi interface issue.

If you have the option, I would recommend that you return the
drive to the vendor and request a replacement -- this 50Gb
drive is clearly not operating at Ultra2 Wide scsi capabilities.

One other thing, before dissing customer service and the websites
of Sun and Seagate, it would be nice if you'd let the froup know
what you tried to do / search for, and what questions you asked.

....
> Second, why can't every SCSI device manufacturer use the same
> terminology to describe the type of interface on their units? I'm
> having trouble identifying exactly which permutation of SCSI all of
> the devices involved actually use. Nearest I can tell, the Sun bus is
> a SCSI-3, Fast20, UltraSCSI unit. I don't know if this is low-voltage
> differential or single-ended; the hard drives both claim to be Ultra2,
> SCSI Wide, LVD units. The CDROM is SCSI-2.

Might I suggest that you do a bit of research into scsi via the
SCSI Trade Association (www.scsita.org) website -- that will
give you at least the basic pointers on the hardware you are
working with. With the SCSI specification being an actual ANSI
Standard, once you understand the specification's terminology
you should not have a problem with figuring out what your hardware
can and can not do.

You should find that LVD and single-ended targets can live
together on the same scsi bus. Just don't mix HVD with anything
else or you'll lose some magic smoke!

The on-board scsi buses on an ultra30 are UltraSCSI (SCSI-3),
Fast40 (40Mbyte/sec).



> The questions:
> Will this drive work with this system?

probably.

> Why do the 10GB and 36GB work, but not the 50GB?

quite possibly a dodgy drive.

> How can I make it work?

use a driver.conf file for your glm controller, OR
get the drive replaced

> Since it evokes no warnings, is the 36GB drive running in optimum
> configuration (wide, LVD), and if so, what configuration is it?

yes. If there are no messages then it is configured on the
scsi bus at its optimum capabilities. Question in return,
why does this matter to you?

> Does the Ultra30 support low-voltage differential devices?

yes.

> What am I missing?

a working drive!

> Please be copious and verbose with your responses. I am at wit's end
> here, both with understanding the problem and solving it.


docs.sun.com
sunsolve.sun.com
www.scsita.org
www.t10.org
http://www.scsita.org/aboutscsi/termsTermin.html
http://www.seagate.com/support/disc/...t150176lc.html

What's confusing to me is that two of the seagate disks operate with
no errors, and the 50GB evokes these errors. Is there a command to
determine what mode the devices are operating in? The prom-level
probe-scsi-all command does not return the information (unless there
is a verbose mode that I'm missing).

Is single-ended the opposite of LVD? And therefore, are all devices
running in single-ended mode? Will it help to enable the
'force-single-ended mode' jumper on the disks?

Thank you

John Muller


Klaasjan Brand <kjb@dds.nl> wrote in message news:<pan.2004.05.16.09.02.36.552972@dds.nl>...
> On Sat, 15 May 2004 17:05:02 -0700, John Muller wrote:
>
> The Ultra 30 doesn't support LVD and will run both disks in single-ended
> mode. AFAIK the seagate disks will support this.
>
> > What am I missing?
>
> I get lots of hits when I copied your error messages to a google search;
> looks like most people think it's a hardware issue...
>
> Klaasjan

HI,
John Muller wrote:
> What's confusing to me is that two of the seagate disks operate with
> no errors, and the 50GB evokes these errors. Is there a command to
> determine what mode the devices are operating in? The prom-level
> probe-scsi-all command does not return the information (unless there
> is a verbose mode that I'm missing).
>
> Is single-ended the opposite of LVD? And therefore, are all devices
> running in single-ended mode? Will it help to enable the
> 'force-single-ended mode' jumper on the disks?
>
Yes it is the "opposite" of LVD, Single ended measn that one physical
wire is used for one signal(say Data bit 0), LVD or as the real name
differential is when you use two physical lines for signal in order to
get a better signal to noise ratio.

I would say that all modern drives use LVD and then do a autoconfig to
single ended mode.

The other tstandard is High Voltage diff, HVD and that one you can not
use at all!

That 50GB drive I would replace with a new standard
Seaget/Maxtor/Hitachi drive 73GB for a $300 and then you would have no
problem at all.

/michael