[Noguri]

Hi there!

This is my very first post so maybe I shall introduce myself first...
I may say I caught the "SGI virus" few months back when I decided to learn Unix and found no better way but to buy an O2 to help me on my quest (no one can deny it's way more classy to learn on an O2 rather than on a VM ). To have an SGI machine was an old dream of mine anyway and God only knows why it took me so long before making up my mind.
Well, thanks to Ian who (expertly!) prepared one machine for me, this old dream became true.

Since, time have past and I had the opportunity to acquire an Onyx2 Rackmount  (the leap with the O2 was huge!)
The price was fair, the specs were OK, the skins almost new and no sign of rust at all. So this was a pretty fair deal!
Of course, the "tricky" part is that the graphics module doesn't start at all. Therefore my post!

I am looking for advice so as to minimize the risk to mess up what is still working inside the rack and identify which board/part may be faulty.

At first, I was thinking of removing all boards of the rack so as to test the PSU/midplane/MSC only. I am pretty sure the graphics PSU is OK (despite I cannot be 100% sure but it delivers the correct voltage when tested apart).
Therefore my question: is there any minimum configuration (in terms of boards installed) inside the graphics module for it to able to start?
Another question: according to your personal experience, do you have any "best practice" or something else I could apply to identify the faulty parts?


Thanks in advance for your lights!

PS: I can post later the hw inventory of the "beast" if you want...

[CB_HK]

Hey there! Congrats on snagging an Onyx2! I’m currently restoring a deskside version myself.

There are a few things you can check initially to make sure equipment like the power supply is getting the proper voltage. I’d recommend getting a copy of the Onyx2 Rack Users Guide, here.

As for diagnostics, there’s a good walkthrough that lists how to check things one piece at a time. You can check that out here.

Best of luck! Please post up photos if you have the time and are interested in doing so.

[indy99]

Hi there!
Since, time have past and I had the opportunity to acquire an Onyx2 Rackmount  (the leap with the O2 was huge!)
The price was fair, the specs were OK, the skins almost new and no sign of rust at all. So this was a pretty fair deal!
Of course, the "tricky" part is that the graphics module doesn't start at all. Therefore my post!

I am looking for advice so as to minimize the risk to mess up what is still working inside the rack and identify which board/part may be faulty.

At first, I was thinking of removing all boards of the rack so as to test the PSU/midplane/MSC only. I am pretty sure the graphics PSU is OK (despite I cannot be 100% sure but it delivers the correct voltage when tested apart).
Therefore my question: is there any minimum configuration (in terms of boards installed) inside the graphics module for it to able to start?
Another question: according to your personal experience, do you have any "best practice" or something else I could apply to identify the faulty parts?

PS: I can post later the hw inventory of the "beast" if you want...

Hello Noguri, congrats on the newly acquired Onyx2.

I have some experience with Onyx2 Rack systems, but it's probably been ten years or so since the last one left the building. I do still have an Onyx350 IR3 system that I can use for reference. While the compute module is obviously different, the Graphics Module /GBrick chassis is the same other than the fascia, the the Ktown interface board, and the MSC (or L1) control relay module.

An inventory would probably prove helpful. To make sure we're operating on the same frequency, a few questions if I might.

Was the system known to run before it was transported to your location?

Never tried a power up a graphics module with less than a minimum set of graphics boards installed. 

When you connect the graphics module to power and flip on the power toggle does the cooling fan start? In the case of my Onyx350, the cooling fan starts up in standby mode as soon as the power toggle (located on the rear of the chassis) is placed in the 'on' position. This occurs even if the compute module isn't running to provide a "power up" signal via the L1 connection. If your Onyx2 was previously known to work and the fan doesn't come on, it might be worth checking to ensure the power supply didn't get dislodged during transport or not get fully re-inserted into the chassis if the system was disassembled.

Does the MSC in your graphics module show signs of life when the power toggle located on the back of the graphics module chassis is in the on position? If the MSC is receiving power you may be able to obtain some minimal diagnostics info by connection a serial terminal configured to 9600-8-N-1 to the 9-pin din port on the rear of the graphics module chassis. Use a null modem cable.

Good luck, hope I can provide some helpful information.

As for diagnostics, there’s a good walkthrough that lists how to check things one piece at a time. You can check that out here.

Thanks for the link. While I haven't visited that site before, that diagnostics walk through seems familiar.

[Noguri]

Hi,

Thank you guys for your replies, your very useful tips and documents and the support yuo gave me so far!!!

To answer indy99 questions at first:

Was the system known to run before it was transported to your location?
=> The graphics module was already known to be faulty

When you connect the graphics module to power and flip on the power toggle does the cooling fan start?
=> Yes! Fan is starting in standby mode, "AC OK" LED on the graphics power module is steady YELLOW. Things look fine if I refer to user's manual

Does the MSC in your graphics module show signs of life when the power toggle located on the back of the graphics module chassis is in the on position?
=> Yes The version is displayed.

I didn't perform the diagnostic test yet but here is the things I tried so far:
- Power on the graphics module (through MMSC) with all boards plugged => The graphics power supply RED LED immediately shows up ("FAULT"), a strange noise (like a click) is emitted and a strong smell of burn appears (The MSC displays "POWER UP" all the time).
NB: of course, when I smelt something I immediately cut the power off...
- Power on the graphics module (through MMSC) with NO boards => Same behaviour (RED LED, noise and strong smell + MSC displays "POWER UP").

Because the fan was working all this time, I could not figure out where the smell came from (neither the "click" sound anyway)...

I would say now there are only 4 parts involved in this mess: fan, PSU, MSC and midplane
(I would personnaly remove the MSC and probably the fan too from the list anyway...)

Well, that's where I am so far....
(note that the CPU module is working well, which is already great )

Wish you all a nice day/evening/night!

[Noguri]

Well,

After complete teardown of the graphics module, clean-up and tightening cables, reassemble all the parts again, there is no smoke neither faults anymore...
I cross my finger but maybe it was just some loose contacts or dust...

Next step is to start up all racks and check if everything is ok... (my current plug only supports 16A, I'm pretty sure it's not enough so I still have to work on this...)

[Irinikus]

Post some pics of your system, and the progress you're making!

[CB_HK]

Great to hear! I’m glad it was just a case of needing to clean and reseat everything.

Definitely post some pictures when you can!

Btw, are you running the system off 110v or 220v?

[Noguri]

Hi folks,

Well, I have some good news
Cleaning up the graphics module and tightening up the power connectors was finally the key!
I'm now an (proud!) owner of an Onyx2 IR2 rackmount fully working!


Here are some (very compressed and ugly) pics:










@CB_HK =>  the system is running under 230V.

Quick abstract for the hardware:
- The graphics module has 2xGE boards, 4xRM7-boards and 2xDG5-8 boards (so up to 16 monitors!)
- The processor module contains 4xIP27 R10k/250Mhz with 1152MB RAM (as previously shown on the hinv picture)
- There is an additional SCSI bay with 6x9Gb HDD

Wish you all a nice evening!

[Irinikus]

Wow, that's very nice!

[CB_HK]

That’s fantastic! A really nice looking system and very well equipped!