IP30, Octane, strange behavior on the SCSI chain

SGI hardware problems, solutions, tips, hacks, etc.
Forum rules
Any posts concerning pirated software or offering to buy/sell/trade commercial software are subject to removal.
jpstewart
Donor
Donor
Posts: 376
Joined: Tue Sep 21, 2010 3:31 pm
Location: Southwestern Ontario, Canada

Re: IP30, Octane, strange behavior on the SCSI chain

Unread postby jpstewart » Mon Jun 08, 2015 7:51 am

ivelegacy wrote:I am experimenting troubles with an external HardDrives box with an Octane2 (not mine). The Box contains 2 Barracuda SCSI/SE 50, and it's connected to the Octane2 SCSI port on the back of the machine.

When you say "SE 50" I presume you mean they are Single Ended, 50 pin drives. If that's correct, how are they connected to the Octane's 68 pin SCSI port?

Octanes can be very picky about termination. If you're using a cheap 68->50 pin adaptor or cable, then that might be the problem. You'll need one that properly terminates the high byte of the SCSI bus. Or you'll need a good external SCSI enclosure such as the Sun 611 which takes a 50 pin device internally but has a 68 pin interface to the Octane and handles termination automatically for you.

Basically you need to ensure something (cable, adaptor, or enclosure) is terminating the high byte between the 68 pin Octane and the 50 pin drives, and then also ensure the 50 pin portion of the bus is terminated after the external drives.
:Indigo2IMP: :Octane: :Indigo: :O3x0:
Sun SPARCstation 20, Blade 2500
HP C8000

User avatar
jan-jaap
Donor
Donor
Posts: 4514
Joined: Thu Jun 17, 2004 11:35 am
Location: Wijchen, The Netherlands

Re: IP30, Octane, strange behavior on the SCSI chain

Unread postby jan-jaap » Tue Jun 09, 2015 3:59 am

Your problem is that you have a wide-scsi capable host, and wide-scsi capable devices, but they are connected with a narrow cable. Take the wide capability out on one end and it works. So, the Octane works with narrow devices (CDROM), and the wide disks + enclosure works with a narrow host e.g. an Indy.
To accentuate the special identity of the IRIS 4D/70, Silicon Graphics' designers selected a new color palette. The machine's coating blends dark grey, raspberry and beige colors into a pleasing harmony. (IRIS 4D/70 Superworkstation Technical Report)

User avatar
jan-jaap
Donor
Donor
Posts: 4514
Joined: Thu Jun 17, 2004 11:35 am
Location: Wijchen, The Netherlands

Re: IP30, Octane, strange behavior on the SCSI chain

Unread postby jan-jaap » Tue Jun 09, 2015 4:46 am

IIRC (I haven't dealt with this stuff for years), SCSI commands are all the way backwards compatible. So they use only the slowest frequency and the narrow signal path (the resulting unnecessary long induced latency was an ever bigger problem as SCSI speeds went up).

So an inquiry about capabilities, IDs etc would work, and the disk will tell the controller it can do wide SCSI. I don't think you should expect a terminator to filter out SCSI commands or responses.

As soon as actual data transmission occurs, half the data will be missing and all hell breaks loose.

SCSI disks often have a jumper to force narrow mode. Try that.
To accentuate the special identity of the IRIS 4D/70, Silicon Graphics' designers selected a new color palette. The machine's coating blends dark grey, raspberry and beige colors into a pleasing harmony. (IRIS 4D/70 Superworkstation Technical Report)

robespierre
Posts: 1234
Joined: Mon Sep 12, 2011 2:28 pm
Location: Boston

Re: IP30, Octane, strange behavior on the SCSI chain

Unread postby robespierre » Wed Jun 24, 2015 11:13 pm

If you put a narrow SCSI device in the middle of a Wide SCSI bus, you can use a regular 50-pin-to-68-pin SCSI adapter (an adapter without built-in high-byte termination) and still have all wide SCSI devices transferring data at wide speeds.

This is not a good idea because it presents different loading to different lines on the bus, which can cause skewing of signals. The confusion between "speed" and "width" is also not comforting.
:PI: :O2: :Indigo2IMP: :Indigo2IMP:

User avatar
recondas
Moderator
Moderator
Posts: 5399
Joined: Sun Jun 06, 2004 5:55 pm
Location: NC - USA

Re: IP30, Octane, strange behavior on the SCSI chain

Unread postby recondas » Thu Jun 25, 2015 5:26 pm

Probably won't directly resolve your issue, but I did notice in the photo you posted of your Octane PCI Cardcage that you've got a 32-bit PCI device (the USB board) in PCI slot one, and a 64-bit PCI device (the QLA1040B) in PCI slot three (to minimize any visual confusion I'll mention the PCI Cardcage is shown upside down in your photo, - I added PCI slot numbers and a this-side-up indicator for clarity):
ivelegacy wrote:
sgi-pci-cardcage.jpg
To eliminate the potential for conflicts, when both 64 and 32-bit cards are installed, install the 64-bit PCI devices in the numerically lowest slots (if more than one is 64-bit you might have to do some trial and error low-slot placement testing with the 64-bit devices to find the optimal arrangement). There's some additional background this thread.

Might not resolve your issue, but might get you a step or two ahead of what could be the next problem.
***********************************************************************
Welcome to ARMLand - 0/0x0d00
running...(sherwood-root 0607201829)
* InfiniteReality/Reality Software, IRIX 6.5 Release *
***********************************************************************

robespierre
Posts: 1234
Joined: Mon Sep 12, 2011 2:28 pm
Location: Boston

Re: IP30, Octane, strange behavior on the SCSI chain

Unread postby robespierre » Fri Jun 26, 2015 1:01 pm

Two more points: the SCSI spec requires drives up to U160 to support narrow transfers, but U320 removes this requirement. So a compliant U320 drive may simply not support narrow operation.
And the SCA-50 or 68-pin adapters sometimes are just broken by design:
viewtopic.php?f=3&t=16729267
:PI: :O2: :Indigo2IMP: :Indigo2IMP:


Return to “SGI: Hardware”

Who is online

Users browsing this forum: No registered users and 1 guest