[iansmith]

For my next project I want to power up my Iris but first I want to try and take a backup image of the hard drive.

This powered up into the GUI 20 years ago but I don't have a monitor for it anymore, so am trying to back up the HD as a first step.

I've managed to image Amiga, Sun and Iris Indigo drives with my setup so I know it's working.

When I plug this drive in it seems to spin up and it reports "CDC 94171-0184" to the SCSI controller.  I can hear the solenoid locking the read/write heads in place click when it turns on, then the drive spins up but I never hear the heads move after that.

Linux reports...

[sda] Read Capacity (1) failed: Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[sad] Sense Key : Not Ready [current]
[sda] Add. Sense: Logical unit not ready, cause not reportable

Is this an issue with my setup, or does this look like the drive has failed?

Thanks!

[Raion]

Linux doesn't support the old XFS IRIX uses, I wouldn't be surprised if it choked on the ancient SCSI that uses or the disklabel format it has.

I suggest perhaps using a more period-accurate machine to dd the drive. Modern Linux is a tossup at best with these ancient devices.

[robespierre]

It sounds like an issue with the drive controller or HDA. First thing to check is any visibly damaged components on the controller board.

[weblacky]

You likely need to load sg3_Utils and perform an sg_start and then an sg_readcap (likely use old 10 vs 16 encoding). Then use sg_dd to grab an surface scan image.

Raion is right (I hate to have to agree, but it's a poor truth), Linux has been terrible at preserving past functionality. Basically if it's NOT used in a major consumer PC anymore...it's gone! Just because linux worked and ran great on your PIII in 1999, doesn't mean it still does. That's a sad statement I'm loth to make it, but that's why I gave up on Linux outside of server use. It's a constantly changing target with no respect for its own history or POSIX standards. So "old Linux software" now has the same damn problem as "old mac software" or old UNIX software. Binaries aren't compatible, kernel's not compatible, and for some reason they keep ripping up subsystems and replacing crap (breaking old code).

When Linux came out it kept it's history for a LONG time, but it seems somewhere around like 2010-ish...someone got their hand in it and started "cleaning up" old hardware support and assumptions. So it's like a new OS (generationally). There is still source compliance at the userland...not much more. I wish they had take PC history and tech archaeology more seriously, but the real reason is, most of the Linux funding for the past 10+ years has been large corporations and that only pay to add in what they want...not maintain legacy support. No one wants to do it and no one is paying for it.

So just like Windows, maybe use a much older Linux (turn of the century)?

[iansmith]

Thanks for the replies.

I'm using an old Pentium 4 PC as that is the only computer I have with a PCI slot for my old BT-948 SCSI adapter, so it's Ubuntu 16. So it's not a modern machine but nothing from back then either. It reads other drives from that era but this one is giving me trouble.

It's not a problem with the filesystem as I can't even read raw data with dd. I can't imagine anything will read it, but I'd like top copy the data so I can contiue trying to work with the hardware and not worry about the drive dying on me.

No visible hardware damage, no burn marks, no leaking or swollen caps.

I tried sg_start, it sits there for a while before returning. sg_readcaps returns the same information as I found in dmesg, drive not ready.

Any other ideas? My last resort would be to pop the lid on the drive and see if anything is stuck. I was around for the Segate stickation issues, but it's obviously after all other options have been tried, which includes just putting it back in the Iris and powering it up to see if it runs on that hardware, could be something specific there. Need a monitor though. I'm still terrified of plugging that monster in and having it explode.

And yeah, I've been around to watch Linux have it's ups and downs, I was using some of these machines back in the 90s.

[robespierre]

If it is anything like the 5.25" SCSI disks I have used (Seagate Hawks that looked basically identical), there should be some head activity even when no host is connected. The drive runs thermal calibration when it starts and that requires seeks across the surface.
I'm sticking with my first answer.

[jan-jaap]

To those bashing Linux here: no 4D/70 ever used the XFS filesystem. Whatever is on there must be disklabel + several EFS partitions. Linux supports that.

But I'd simply 'dd' it. And add 'conv=noerror' so it doesn't trip over the first bad sector. Everything else can be done later.

[weblacky]

The problems being discussed are not file system related. Drive not ready. For whatever reason, could be disk failure or could be Linux SCSI support.

[Raion]

For Jan: I've never used SCSI under GNU/Linux so I was spitballing as to the answer. My point was if the Linux kernel doesn't support X then there's a good chance Y is left unsupported. It could be the adapter doesn't negotiate the SCSI version properly due to a weak driver, it could've been that his OS version doesn't package the disklabel or EFS support. 

His replies clearly clarified the issue is either with the drive itself or compatibility issues with the controller.

Therefore, I think the next step logically is to throw caution to the wind and power the 4D/70 up, since it's a good chance his drive is screwed.

[jan-jaap]

I just saw 'XFS' mentioned twice in the thread and triggered on it, that's all ;-)

Regarding the system, let me tell you about my own 4D/70 which spent years in storage when I put my hands on it:

* The PSU, a blue 2000W "shoebox" made by L&H Research made some terrible noises. It was basically screaming.
* The twin-tower enclosure has dual centrifugal blowers in the base. They are not monitored in any way. In mine one had burned out.
* The disk was basically in the same shape as OP describes: when powered up it made some bleep-bleep (head seeking) noises, spindle motor runs, but not much else happened.
* In the end the VME backplane turned out to be jumpered incorrectly causing VME bus errors during graphics initialization.

I still can't explain the backplane problem, but everything else is to be expected if a system sits abandoned for a decade or more. So I wouldn't just cross my fingers and power it on. You risk damaging things impossible to replace these days.

* First test the PSU, outside the chassis if possible. Be careful, it puts out at least 200A@5V, screwdriver-melting stuff.
* If it works correctly, fire it up inside the chassis with all cards removed. Check fan operation (they run off the -12V rail)
* The IP4 should work with nothing else (graphics, VME ethernet) in the chassis. If you complete POST you will find the NVRAM battery long expired. IP4 doesn't understand 'resetenv', you have to enter correct values for each env value manually using the serial console.
* Now you can put back the rest of the cards and hope for the best.
* You will likely have to look for a new hard disk. Don't try latest & greatest SCSI, U320 doesn't require backwards compatibility with narrow / sync SCSI. I tried a Seagate 15K.4 in a PowerSeries once -- no joy.
* If you have a QIC tape drive it's probably QIC-24, not QIC-150. (Very) old IRIX tapes are QIC-24, otherwise there's always network installation.

I have experience with older SGIs, but it took me quite a bit of time to get my 4D/70 going again...