Re: SCSI ID vs. unit ID

With respect to that external plate and connector. I have added tape to that over the years to use for migrating customer to a new system with incompatible tape. Hook up the cable and assure the drive has power from an external source and it was recognized. Of course it had to be a drive that IBM i expected to see on that internal bus. So for example the 7208 can't be hooked up there as it needs to be on a dedicated SCSI card.

I believe the original function of the connection was for system testing at the point of manufacture and it wasn't intended for us to use in the field so there is little knowledge about it

Funny story when we built the box that had all those DVDs in it on one chain they were all recognized. IBM i actually created four DVD drives. BUT somehow, despite the multiple drives, they were all seen as one. So when you put a disk in and closed the drive, it attempted to mount it four times in /QOPT. This of course is a problem, and it simply kicked out the second drive, but it actually kicked open all four LOL! That's when I messed around and discovered that odd inverse addressing.

So your drive at 3, while not normal for a factory machine, is still valid and works fine.

With respect to correlation. As was mentioned by others many systems (all the later ones) set the SCSI address by the location in the canister while this guy, and the model 200, 400, and 436 (perhaps others?) required jumpers set on the drive per the chart on the inside of the cover.

We added external disk to a 170 (Frankie I) By cabling a disk canister from a model 620 to the open port on the RAID card. The system was happy with that, though it couldn't tell where those disks were located. Initially for heat we put a drive in slots 1,3,5 to allow more cooling. Those drives showed up in order to IBM i in Service tools despite being out of SCSI order. When we added a fan and then filled slots 2 and 4 those drives appeared at the end of the order in service tools.


- DrF

On 1/17/2022 10:05 AM, Patrik Schindler wrote:

Moin,

Am 17.01.2022 um 15:18 schrieb Larry DrFranken Bolhuis <midrange@xxxxxxxxxxxx>:

They used a sort of inverse SCSI numbering for optical back then.

Aha?

I had to decode it when I built Frankie II and Frankie III because we had as many as 4 DVD-RAM drives on the same SCSI string. (Get it wrong and IBM i think all the drives are the same one!) I believe if you set the jumpers for 12 the device became address 3. Something like that anyway.

This is weird and I honestly don't understand it. But since this seems to be some logical association, SCSI IDs are set as usual. Have had no issues with a Model 400 (my first AS/400) years ago, multiple model 150's, and now with a P02, and P03.

The tape as you see must be device 0 and has no jumpers installed (Unlike Optical).

It is.

The load source disk then is address 6 and I believe that is the only one that mattered.

Correct. The highest ID wins the bus if multiple bus requests are to be arbitrated.

All hard disks needed to be higher than that.

On PCs, the SCSI to PCI (or ISA, or whatever) chip usually has ID 7, so Narrow- and Wide SCSI components on the same bus can be addressed by it. I think, that's the same with AS/400 and later hardware, because there is often also a mix of Wide- (Disks) and Narrow SCSI components (Tape, CD-ROM) on the same SCSI bus.

That thing only accepted 4 drives though given my other experiments I do suspect that if you found a way to extend the SCSI cable and added a 5th and 6th unit they might function.

The 150 has a stock 68 Pin "external" HD connector hidden behind a metal coverage, just above the PSU. Problem is after the external connector, the cable runs 2 cm to its real end, to a permanently attached bus terminator. I have not yet found out how automatic SCSI bus termination really works, to verify if that thing can switch itself off magically. At the same time, I highly doubt that said bus terminator is automatic: The connector is normally hidden, and thus more likely another "debug" port, to attach a *short* cable to a logic analyzer or so.

While having more than one terminator on one bus isn't harmful from the viewpoint of electrical interference, the additional current being delivered to the bus' lines might put unusual strain to the chips in SCSI devices. I don't want to fry anything.

I've not yet come to terms if I want to cut the internal terminator from the cable, and use an external one, either with, or without additional external devices. ATM, I'm using just two internal 15k disks on my 150, which leaves more than enough space, and reasonable paging performance.

Having more disk arms is interesting for my P02, as I have four AS/400 compatible IBM 0663 drives spare, I intent to hook up externally to the P02. "More disk arms are more good." :-)

Just keep incrementing the SCSI address.

Yes, very important!

I think the unit number will more likely match the sequence they were added to the machine.

Okay, so no definitive answer if and how the unit number is derived from the SCSI ID.

Thanks anyway!

:wq! PoC