Talk:ND-110 Satellite 9883.21238

Boot problems

What is the settings of the CPU card? It might have come from another machine not set up for booting from the SCSI disk. The Satellites have been mixed around a lot and I think I had to add a CPU-card to that machine. Looking at pictures from my working one I see

• Thumb wheel on CPU card 9
• Thumb wheels on SCSI FLOPPY 0 0 7
• Thumb wheels on memory card 0 0 0

I'm not at the machine right now, but the CPU thumb wheel is currently at 9 (the ALD switch). 30.003.7 says 'No loading done' for this position, but the document doesn't seem to be updated for SCSI disks so it may be slightly out of date. I don't know what the physical device number for SCSI is supposed to be. As for the other boards, as far as I rembember, the SCSI board is indeed 0 0 7, and the memory card 0 0 0 (will double check when I get home).

Then pull out any ethernet, PIOC or HDLC-card as there might be address conflicts. I had problem with my ethernet card and couldn't get the machine to boot at first. I will check the configurations of the other cards when I get back out to the workshop. --Gandalf (talk) 12:01, 26 September 2016 (UTC)

Ok, thanks! I will try without those other boards (no PIOC, but I've got two terminal boards - I can presumably run with only the SCSI, Memory and CPU boards).

--TArntsen (talk) 12:45, 26 September 2016 (UTC)

Got it booting! The memory card was at 0 0 1, not 0 0 0 - what I remembered was the LEDs which are 0 0 0. The system is still a bit finicky at poweron, it needed three tries before the console&panel responded. Now the problem is UE.. User Environment started at the console, and IIRC UE uses a different set of passwords - I never used that abomination UE, so I'm not sure. But certainly it doesn't recognise user system and the system password. So for the moment I'm stuck. I can continue when I'm back from Japan. Obviously I would like a clean shutdown though.. which needs a login! :-) --TArntsen (talk) 14:37, 26 September 2016 (UTC)

Hard drive - figure out SCSI parameters

I see you have hardware now - congrats! One thing about the SCSI hard drive (in addition to make an image of it) - it would be nice if you could figure out what SCSI parameters or changes to firmware (if any) ND have changed on the drive in order to make it work with ND machines. Some things you can probably figure out with the drive connected to a Linux or other unix box. Examples:

• does the drive report 1024 byte blocks instead of the more common 512 byte blocks?
• Has the name or firmware version changed (compared to a non-ND drive of the same model)?
• Does any of the SCSI / drive parameter pages contain unusual values?

Apologies if you have thought about this already. --Torfinn (talk) 08:43, 31 August 2016 (UTC)

I'll try to check for all of that if and when I can get the drive connected to something.. right now I don't have a SCSI interface on any of my Linux boxes (single ended or otherwise). My SCSI SGI O2+ got a hardware failure recently, so that one is also not an option. I should have access to an Octane with differential SCSI, but will need some interfacing work. TArntsen (talk) 09:50, 31 August 2016 (UTC)

If you need or want a PCI SCSI card, I have several of the common Adaptec cards, they work fine with FreeBSD and should work in Linux as well. I could send you one, if you want to. --Torfinn (talk) 10:11, 31 August 2016 (UTC)

Thanks, I may have to take you up on that (the last PC card (Adaptec 1542 ISA) I had broke some time ago). I'll come back to you if I can't get the SGI solution to work.TArntsen (talk)

If I knew you needed a SCSI card I would have thrown in a PCI one. The system I have was easily imaged by just plugging it to a linux computer and using dd, I don't believe there is any exotic settings for the disk format but I might be wrong. This is purely speculation but I believe that the problem with getting a generic disk working with SINTRAN is probably because the disk formatter is reading the model number and without a known model don't know how big system it is going to create. A dump of the formatting program contains strings with the name of the disks it supports. --Gandalf (talk) 12:58, 31 August 2016 (UTC)

I found an Adaptec AHA-2940UW card, so I'll try that. TArntsen (talk) 13:58, 31 August 2016 (UTC)

Quick update (more later): The drive uses 1024-byte sectors. And it has an ND-specific name. But I believe the 1024-byte sectors are enough to explain the weird image created when you tried to install SINTRAN on a 512byte-sectored disk. Because that's what it looked like. The type etc. probably only matters when formatting. So, try finding a 1024-byte sectored disk, plug it in, install SINTRAN M again, and see how it goes. --TArntsen (talk) 20:38, 3 September 2016 (UTC)

sg3_utils with sg_format in linux can reformat a drive to change block size, see this forum post.--Gandalf (talk) 08:01, 4 September 2016 (UTC)

I tried sg_format on a drive, and ended up with a bricked drive. Evidently, not all drives can be reformatted to 1024-byte blocks. More info on my sg3_utils subpage. --Torfinn (talk) 21:57, 4 September 2016 (UTC)

I tried setblocksize on the "bricked" drive, and it manged to fix it. I did a SINTRAN install on the drive afterwards. Unfortunately, the machine still fails to boot. either it is still something wrong with the drive settings (parity enable - on or off?), or maybe the SCSI controller on the 3201 card is broken? I will image the drive later, so the image can be analyzed. --Torfinn (talk) 22:03, 6 September 2016 (UTC)

Power cord

Hint: it is possible (but not necessarily suggested) to modify a power cord with a C13 plug to fit into the C16 inlet on the Satellite. --Torfinn (talk) 20:23, 8 September 2016 (UTC)

Done with a knife, a drill, and a woodworking tool. TArntsen (talk) 15:48, 9 September 2016 (UTC)

I don't know why they used that plug and why some manufacturers still uses it. I recently installed a new HP PoE switch with that plug. I had to borrow the cable from my satellite when configuring the switch. :) --Gandalf (talk) 11:22, 10 September 2016 (UTC)