SGI: Hardware

Hi,

I just picked up an O2 from the dumpster at work. It's not a massive spec but there were a few others in a bad state so managed to salvage some memory to boost it up to an R5000 300MHz with 256MB RAM. It came with no hard drives, but does have both caddies. I had a brand new, sealed, new old stock 72GB IBM Ultra 320 SCSI disk kicking around from an old system so fitted that into the caddy - and now the machine no longer powers up! Previously I could get to the 'BIOS' screen and go into the Maintenence screen and so forth. Now with the drive in, the power LED just remains orange, no boot up chime, and no display. Presumably the drive is not compatible but reading the FAQ it does say Ultra 320 disks should be ok. Is it the size of the disk that's the problem or should that be OK?

Is there something I'm missing?

Thanks in advance.

The 300 is top end for an r5k, so it's not eactly "low-spec" either The r5's are cooler and quieter than the R10's, they make for a nice O2. Just add more memory. O2's love memory, since they share it with graphics.


No, should be fine. I've run 300 gig U320 disks in an O2.

I would first try sticking it in the other slot, just in case there was some spooge or something in slot 0 that was creating a problem. In fact, since you say it came from the dumpster, did you clean out the old ham and cheese sandwich residue ?

Excellent - thanks for the quick response! It was one of three machines - the other two were R5K 200MHz's with the earlier type skins and this has the later type skins so I decided to take this one. And it's only metaphorically a dumpster - really it's a computer disposal storage unit, so no jam sandwiches or other things likely to be clogging up the slots

I'll have another tinker with it at lunch time and see if I can get it to come to life with the drive in, and I'll report back.

SGI systems run a power on diagnostics routine when powered up that might provide you with a little more detail concerning the problem.

As is, your only indication of the results of those tests are limited to the color/status of the LED display (and the boot chime).

You may be able to capture additional detail from the power on diagnostic routine by connecting a PC/Mac a terminal emulator program to the first serial port.

But for no longer than it'll take I'd probably give the O2 Jump Start Procedure a shot first - there have been more than a few mentions of a jumpstart reviving comatose O2s.

You might also find fu's O2 Diagnostics Aggregator useful.

Thanks for that information. Annoyingly I do have the correct serial cable for an Indy / Indigo2 but not a regular serial cable for an O2.

Worth mentioning that it works just fine with the drive removed - it's only when I insert the drive that I get this behaviour. Without it installed, the system comes on fine (but obviously doesn't boot!).

OK - total non problem. The O2 is indeed powering on, but it just takes a fair bit longer to do so with a drive installed in it. The drive was so quiet I wasn't hearing it spin up! But once it does, I get the jingle and the O2 welcome screen comes up - if I enter command monitor and run hinv, I can see it listed in the hardware.

Cheers for the advice & help folks

Some drives have a "delayed start" jumper, so they don't trip the circuit breaker if they're used in large disk arrays

If you check the drive's documentation, you should see if it has any such jumpers and if they're set on your drive.

I did have a damn good look all over the thing but couldn't see any jumpers. It's possible the O2 drive caddy has obscured one but it's hard to say. I don't plan to reboot it an awful lot once it's built so.. shouldn't really be much of a deal

The jumpers for some of the IBM drives I've had are at back of the PCB, towards the motor, not at the connector side. You may have to take the drive out of the caddy, but I'd recommend reading the documentation first because the jumpers may or may not be labeled properly.

I'm at my wits end trying to get a hard-drive that works in this machine. The problem is I have an abundance of old HP SCSI drives from old servers, here at work - the problem is they all seem to be modified versions of compatible drives (going by the information on Ian Mapleson's site) that have the jumpers (for forcing SE mode, parity, etc) all deleted from the PCB - I can get the O2 to boot into fx just fine, and repartition the disk but then as soon as I attempt to install IRIX I get a whole screen full of parity errors along with some ugly messages, before being spat out to the 'Safe to power off' screen.

Think I'm reluctantly going to have to fork out for a drive off Ian or someone. Not exactly turning out to be a free machine!

I've used HP branded disks in SGIs before. An O2 is not *that* picky...

It's possible that something else is going on, better check before you spend money for nothing.

* I believe the U320 spec makes 'narrow' SCSI support optional, but I've never seen an LVD disk that had issues negotiating SE mode. The O2 is ultra-wide (SE, 40MB/s).

* Maybe the disk doesn't insert securely into the backplane -- you could try to clean things or try with an external disk.

* Maybe these disks are from sort of array and have a non-standard sector size (other than 512bytes/sector)?. This usually affects FC disks, though. With 'fx' you can verify this, 'options' > 'show' > 'geometry', look for data bytes/sec.

* Maybe you should boot the O2 into 'fx' and run a disk wipe: 'exercise' > 'sequential' > 'write-only' > from 0 to whole disk range. See if that works. You have to re-create a disklabel before you can start the IRIX installation.

Good info, thanks. Sorry for these lame questions! I'm totally clueless when it comes to this SGI kit, and SCSI in general!

The disks would almost all have been in arrays of some kind even if 4-6 disk raidsets on a server.

Of course it's also possible that there's a fault with the O2 - I could be jumping through hoops for a problem that's actually more terminal. But I'll try the steps you've suggested and see how I get on! I thought it odd as I am running one of these types of drives in my Indigo2 at home with no problems though that's going through one of Ian Mapleson's adaptors..

I think I might have come across a Fujitsu that needed to be forced with a jumper. I know I came across drives that needed to be forced narrow, so it's worth a try.

Try doing a low level format of the drive.

OK - so I found a new old stock / sealed Ultra 320 36GB 10k SCSI disk at work this morning and threw that into the O2. Came up nice and quickly and I booted into fx, to create a partition on the disk and then launched the installer - same parity error! I'm running it now through the low level format process (being going for about 3 hours!) and will report back on whether it works after that.

I think you'll find out that a low level format works just fine. And then the install will still fail.

A low level format works like this: computer says to disk: plz format yourself and report back when you're done. And then it sits there for a couple of hours until the disk says it's done.

In your case, I have the feeling it's the SCSI bus that's causing the problem. You're better served with a disk exercise test which stresses the bus while the test is running. Read my previous post about how to use 'fx' to exercise the disk. Choose the sequential read and/or write tests to move lots of data fast, the butterfly test will waste lots of time doing the same thing because it does random access.

An individual disk may fail, but if you have a number of (different!) disks and none work, it's very unlikely that the problem is with the disk(s). And a new disk shouldn't need a low level format anyway.

Doh.. I think I worded my reply wrong - I'm doing the exercise on the disk that you advised me to do Not a low level format!

Successfully completed the write exercise test with no warnings or errors. Rebooted into the installer, and getting the same error:



Starting to wonder if there's something bad on this O2

Doing some more reading up I'm seeing some comments about checking the internal CDROM drive (which in my case is dead - and disconnected, pending a replacement coming from Ian Mapleson). I wonder if that's causing me this trouble as it's not terminating the bus properly?!

That's it! You can't get away with an unterminated bus. Well, you could with a 4D/35 maybe but not with ultra/wide SCSI ...